WO2013054712A1

WO2013054712A1 - Sleep assessment system and sleep assessment apparatus

Info

Publication number: WO2013054712A1
Application number: PCT/JP2012/075638
Authority: WO
Inventors: 由佳本田; 千秋山谷; 佐々木　敏昭; 松本　崇
Original assignee: 株式会社タニタ
Priority date: 2011-10-14
Filing date: 2012-10-03
Publication date: 2013-04-18

Abstract

Provided are a sleep assessment system and sleep assessment apparatus for accurately assessing quality of sleep. A sleep assessment system and sleep assessment apparatus for multiplying: sleep determination data corresponding to a predetermined item calculated from the subject's biological signals; and a principal component factor for each of a plurality of types of predetermined items in a sleep assessment score obtained by running a principal component analysis on the predetermined items, the predetermined items including at least an item relating to the depth of sleep, an item relating to the rhythm of sleep, and an item relating to interrupting arousal, extracted on the basis of PSG measurement data; and calculating a sleep assessment score, calculating a sleep disorder conditional probability obtained by running a logistic regression analysis on the sleep assessment score, and computing a sleep index of the subject on the basis of the sleep disorder conditional probability.

Description

Sleep evaluation system and sleep evaluation apparatus

The present invention relates to a sleep evaluation system and sleep evaluation apparatus for evaluating the quality of sleep.

Generally, the quality of sleep is determined by a polysomnography (PSG) obtained by acquiring each data of electroencephalogram, eye movement and mental electromyogram and judging based on these data. However, when acquiring the data, it is necessary to have a hospitalized examination at a medical institution, affixing electrodes to the human body, etc., which is not only burdensome for the test subject, but also requires specialized knowledge such as operation of the measuring device. Since it cannot be measured at home, a sleep evaluation apparatus that can easily evaluate the quality of sleep in a general home has been developed (see, for example, Patent Document 1).

JP2008-301951

In such a sleep evaluation apparatus, unspecified regarding sleep determination data measured by the apparatus (for example, the length of time of sleep, the amount of awakening in the middle, the amount of deep sleep, the amount of body movement, etc.) Sample data of a large number of subjects are collected and analyzed, a regression equation for determining the quality of sleep is created, and this regression equation is incorporated into the sleep evaluation device to be a product.
However, the regression equation created in this way can incorporate the same regression equation for the same type of sleep evaluation device. For example, if it is incorporated in another newly improved sleep evaluation device as it is, There is a risk that an error may occur due to a difference in apparatus. Therefore, when developing a new sleep evaluation device, it is necessary to collect and analyze sample data relating to sleep determination data measured by the device again and recreate a new regression equation for the new sleep evaluation device. There was development effort and cost.
Further, when creating a regression equation for calculating a sleep index such as a sleep score, it is desired to further improve the accuracy of sleep quality evaluation by reflecting another predetermined item. In particular, in the evaluation of sleep quality, it is important to consider items related to sleep depth, items related to sleep rhythm, and items related to mid-wakefulness. In the evaluation, since the items related to the sleep rhythm (cycle) are not evaluated, and the deep sleep rate is not considered, there is room for improving the evaluation ability of the items related to the sleep depth.
Furthermore, a sleep evaluation system and a sleep evaluation apparatus capable of making a difference in the evaluation results of sleep quality between a sleep disordered person and a healthy person such as a patient with sleep apnea syndrome (SAS) are desired. It is.

Moreover, in the conventional sleep evaluation apparatus, what calculates | requires the sleep score which is an objective numerical value as a parameter | index which evaluates a test subject's sleep quality, and displays on a display part is known. The test subject can know the quality of the sleep state of the day by seeing such a sleep score.
However, in the conventional sleep evaluation device, the subject confirms the sleep score, and if the value is high, it can be known that it is “good sleep”, and if it is low, it is “bad sleep”. It is impossible to grasp at a glance whether the score was good (or bad).
Also, since the sleep depth, sleep cycle, sleep time, etc. may vary from day to day, for example, even if the sleep score of yesterday and yesterday's sleep score are substantially the same value as a result, the day before yesterday The quality of sleep and yesterday's sleep are not always the same. Therefore, the subject cannot find the quality difference only by confirming the sleep score.
In order to know the factor that the sleep score as described above was good (or bad) and the quality of sleep, it was necessary for the subject to read and judge from the sleep stage graph and other information. For this reason, for a subject who does not have accurate and detailed knowledge regarding sleep determination, it is likely to be a sensory evaluation, and accurate evaluation and countermeasures may not be performed properly.
In particular, subjects in general-use sleep evaluation devices often do not have sufficient knowledge to perform accurate evaluations themselves, so a result display format that makes it easy for such people to understand the judgment results of sleep quality Is required.

The present invention has an object to provide a sleep evaluation system and a sleep evaluation apparatus that use a regression equation for evaluation of a general-purpose sleep quality created based on PSG measurement data. Moreover, an object of this invention is to provide the sleep evaluation apparatus and sleep evaluation system which are easy to confirm the judgment result of the quality of sleep.

In order to solve the above-described problems, a sleep evaluation system according to the present invention includes a measurement device including biological information detection means that detects biological information of a subject and outputs the biological information as a biological signal, and sleep of the subject based on the biological signal. An information processing terminal for calculating an index, wherein the information processing terminal relates to at least an item relating to sleep depth extracted based on measurement data of PSG and a sleep rhythm Calculated from a principal component coefficient for each predetermined item of a sleep evaluation score obtained by performing a principal component analysis on a plurality of types of predetermined items including items and items relating to mid-wakefulness, and the biological signal of the subject Obtained by calculating a sleep evaluation score by multiplying the sleep determination data corresponding to the predetermined item, and performing logistic regression analysis on the sleep evaluation score Calculating a sleep disorder judgment probabilities, characterized by calculating the sleep index of said subject based on said sleep disorder discrimination probability.
For the determination of sleep disorders, apnea hypopnea index (AHI), Pittsburgh sleep quality index (PSQI), Visual Analog Scale (VSA), OSA sleep survey (OSA sleep survey) Shirawa-Azumi sleep inventory (Kwansei-gakuin Sleepiness Scale; KSS), St. Mary's Hospital Sleep Questionnaire (St. Mary's Hospital Sleep Question Scale) Epworth sleepiness scale (ESS) ), Stanford Sleepiness Scale (SSS) and the like are suitable, but it is possible to use a wide range of indices related to sleep, and the present invention is not limited to these various indices.

In the sleep evaluation system according to the present invention, the item related to the sleep depth includes at least one of a deep sleep rate or a deep sleep appearance amount, and the item related to the sleep rhythm includes a sleep cycle or a differential sleep cycle score. And at least one of sleep efficiency or number of times of awakening in the middle and long period is included.

In the sleep evaluation system according to the present invention, the sleep evaluation score includes any one or more of a sleep depth score, a sleep cycle score, a sleep time score, and a midway awakening score.

Moreover, in the sleep evaluation system according to the present invention, the plurality of types of predetermined items are deep sleep rate, differential sleep cycle score, total bedtime, sleep cycle, deep sleep appearance amount, differential total bedtime score, medium length It includes the number of time awakenings, the number of short-time awakenings, and sleep efficiency.

Moreover, the sleep evaluation apparatus according to the present invention includes biological information detection means that detects biological information of a subject and outputs it as a biological signal, and a determination unit that calculates the sleep index of the subject based on the biological signal. In the sleep evaluation device, the determination unit includes at least an item related to sleep depth, an item related to sleep rhythm, and an item related to mid-wakefulness extracted based on measurement data of PSG A principal component coefficient for each predetermined item of a sleep evaluation score obtained by performing principal component analysis on a plurality of types of predetermined items, and sleep determination data corresponding to the predetermined items calculated from the biological signal of the subject. Multiplying to calculate a sleep evaluation score, calculating a sleep disorder discrimination probability obtained by performing a logistic regression analysis on the sleep evaluation score, and based on the sleep disorder discrimination probability Characterized by calculating the sleep index of the subject Te.

In the sleep evaluation device according to the present invention, the item related to the sleep depth includes at least one deep sleep rate or deep sleep appearance amount, and the item related to the sleep rhythm includes a sleep cycle or a differential sleep cycle score. Including at least one of the items, and the item related to mid-wake awakening includes at least one sleep efficiency or number of mid-long-time awakenings.

In the sleep evaluation device according to the present invention, the sleep evaluation score includes any one or more of a sleep depth score, a sleep cycle score, a sleep time score, and a midway awakening score.

Further, in the sleep evaluation device according to the present invention, the plurality of types of predetermined items are deep sleep rate, differential sleep cycle score, total bedtime, sleep cycle, deep sleep appearance amount, differential total bedtime score, medium length It includes the number of time awakenings, the number of short-time awakenings, and sleep efficiency.

Moreover, in order to solve the said subject, the sleep evaluation apparatus which concerns on this invention detects the biological information of a test subject, outputs the biometric signal as a biomedical signal, and the test subject's sleep state based on the said biometric signal. A plurality of types including at least an item related to sleep depth, an item related to sleep rhythm, and an item related to mid-wake awakening A sleep evaluation score is calculated for each of the predetermined items, and it is determined based on the sleep evaluation score whether the sleep state of the subject corresponds to a predetermined sleep type.

Moreover, in the sleep evaluation apparatus according to the present invention, the sleep type is a type stored in advance according to a feature of sleep content, and the determination unit uses the calculated sleep evaluation score to calculate the sleep type. A determination value is calculated for each, and based on the determination value, which of the predetermined sleep types is determined is determined.

In the sleep evaluation device according to the present invention, the sleep evaluation score used for determining which of the predetermined sleep types includes at least a sleep depth score, a sleep cycle score, and a midway awakening score, or In addition to these, a sleep time score is further included.

Moreover, the sleep evaluation apparatus according to the present invention has a display unit capable of collectively displaying the sleep evaluation scores calculated for each of the plurality of types of predetermined items.

Moreover, in the sleep evaluation device according to the present invention, the sleep evaluation score displayed in a lump on the display unit includes a sleep evaluation score used to determine which of the predetermined sleep types corresponds, or these In addition to a body movement frequency score and / or a sleep habit score.

Moreover, the sleep evaluation apparatus according to the present invention is characterized by having a display unit capable of displaying the sleep evaluation score calculated for each of the plurality of types of predetermined items on a radar chart.

In addition, the sleep evaluation system according to the present invention includes a measuring device including biological information detection means for detecting biological information of a subject and outputting the biological information as a biological signal, and information processing for determining the sleep state of the subject based on the biological signal. A plurality of types of predetermined items including at least an item related to sleep depth, an item related to sleep rhythm, and an item related to mid-wakefulness. A sleep evaluation score is calculated for each, and it is determined based on the sleep evaluation score whether the sleep state of the subject corresponds to a predetermined sleep type.

According to the present invention, since the regression equation for evaluating the quality of sleep is created based on the measurement data of PSG, it can be widely applied to sleep evaluation devices in general, and a new sleep evaluation device is developed. Even in this case, the development process of recreating the original regression equation can be omitted. Further, according to the present invention, the regression equation for evaluating the quality of sleep is created based on the measurement data of PSG. Therefore, the information processing terminal incorporating this regression equation and the sleep determination data acquisition device are connected. Thus, the sleep determination data of PSG can be directly substituted into this regression equation and used for the evaluation of sleep quality, which is also useful for the evaluation of sleep quality in medical institutions.

In addition, according to the present invention, when creating a regression equation for evaluating the quality of sleep, an item indicating how much the sleep cycle differs from a predetermined reference time, and the total bedtime is a predetermined reference time. This also reflects the item of how much difference there is, so that the time used as a reference for the sleep quality evaluation is set, so that the explanatory power can be further improved.

In addition, according to the present invention, a sleep depth score, a sleep cycle score, a sleep time score, a mid-wake awakening score can be calculated as a sleep evaluation score based on a plurality of types of predetermined items, and thereby a sleep cycle score is also evaluated. The accuracy of the sleep quality evaluation is improved as compared with the prior art. Furthermore, since the deep sleep rate is also reflected in the sleep quality evaluation, the evaluation ability of items related to sleep depth is improved.

In addition, according to the present invention, since the probability of being a sleep disorder person such as a SAS patient is calculated, the sleep evaluation system can make a difference in the sleep quality evaluation result between the sleep disorder person and the healthy person And a sleep evaluation apparatus can be provided.

In addition, according to the present invention, it is determined which of the predetermined sleep types, so that the subject himself can understand what his / her sleep was very easily.

In particular, the sleep evaluation score includes a sleep time score and / or body movement frequency score in addition to a sleep depth score, a sleep cycle score, and a midway awakening score, so that the sleep depth, sleep cycle, sleep time, midway awakening, body From various viewpoints such as motion frequency, it becomes easier to find improvements such as what can be improved to improve sleep evaluation.

Moreover, since the sleep evaluation score can be displayed in a lump, particularly on a radar chart, the score affecting the sleep evaluation can be visualized, and the sleep contents can be easily analyzed and self-evaluated.

In addition, the sleep type is a type stored in advance according to the characteristics of the sleep content, and the determination value for each sleep type is calculated using the calculated sleep evaluation score. It becomes possible. As a result, provision of advice for each sleep type can also be automated.

It is an external appearance perspective view of the sleep evaluation apparatus of an Example at the time of use. It is an electrical block diagram of the sleep evaluation apparatus of an Example. It is a flowchart which shows main operation | movement. It is a flowchart which shows the flow of sleep stage determination. It is a flowchart which shows entering / leaving judgment. It is a figure which shows the respiration waveform of a bed leaving state. It is a flowchart which shows body movement determination. It is a figure which shows the respiration waveform of an inbody movement state. It is a figure which shows the respiratory waveform of a rough body movement state. It is a figure which shows the respiration waveform of a thin body movement state. It is a figure which shows the relationship between body movement determination and arousal determination. It is a flowchart which shows arousal determination. It is a flowchart which shows sleep onset determination. It is a flowchart which shows deep sleep determination. It is a flowchart which shows REM and shallow sleep determination. It is a flowchart which shows midway awakening determination. It is a flowchart which shows midway awakening condition determination. It is a flowchart which shows wakeup determination. It is a flowchart which shows a sleep point calculation. It is a flowchart which shows deep sleep appearance amount calculation. It is a flowchart which shows the number of times of awakening in middle and long time. It is a flowchart which shows the frequency | count of awakening for a short time. It is a flowchart which shows sleep efficiency calculation. It is a flowchart which shows the standardization of data. It is a flowchart which shows a principal component score calculation. It is a flowchart which shows discrimination | determination probability calculation. It is a flowchart which shows a sleep point calculation. It is a time chart for demonstrating a predetermined item. It is a flowchart which shows a determination result display. It is explanatory drawing which shows the screen of a determination result. It is explanatory drawing which shows the screen of a determination result. It is a graph which shows the comparison with the prior art regarding sleep score and this invention, (a) is a graph which shows the correlation with the conventional sleep score and deep sleep rate, (b) is the sleep score of this invention, and deep sleep. It is a graph which shows the correlation with a rate. It is a graph which shows the comparison with the prior art and this invention regarding a sleep score. It is an external appearance perspective view of the sleep evaluation apparatus of an Example at the time of use. It is an electrical block diagram of the sleep evaluation apparatus of an Example. It is a flowchart which shows main operation | movement. It is a flowchart which shows the flow of sleep stage determination. It is a flowchart which shows entering / leaving judgment. It is a figure which shows the respiration waveform of a bed leaving state. It is a flowchart which shows body movement determination. It is a figure which shows the respiration waveform of an inbody movement state. It is a figure which shows the respiratory waveform of a rough body movement state. It is a figure which shows the respiration waveform of a thin body movement state. It is a figure which shows the relationship between body movement determination and arousal determination. It is a flowchart which shows arousal determination. It is a flowchart which shows sleep onset determination. It is a flowchart which shows deep sleep determination. It is a flowchart which shows REM and shallow sleep determination. It is a flowchart which shows midway awakening determination. It is a flowchart which shows midway awakening condition determination. It is a flowchart which shows wakeup determination. It is a flowchart which shows a sleep point calculation. It is a flowchart which shows deep sleep appearance amount calculation. It is a flowchart which shows the number of times of awakening in middle and long time. It is a flowchart which shows the frequency | count of awakening for a short time. It is a flowchart which shows sleep efficiency calculation. It is a flowchart which shows the standardization of data. It is a flowchart which shows a principal component score calculation. It is a flowchart which shows a sleep disorder discrimination | determination probability calculation. It is a flowchart which shows a sleep point calculation. It is a time chart for demonstrating a predetermined item. It is a radar chart which shows the example of the 1st sleep type. It is a radar chart which shows the example of the 2nd sleep type. It is a radar chart which shows the example of the 3rd sleep type. It is a radar chart which shows the example of the 4th sleep type. It is a radar chart which shows the example of the 5th sleep type. It is a flowchart which shows a body movement frequency score calculation. It is a flowchart which shows sleep type determination. It is a flowchart which shows a determination result display. It is explanatory drawing which shows the screen of a determination result. It is explanatory drawing which shows the screen of a determination result. It is explanatory drawing which shows the screen of a determination result. It is a graph which shows the comparison with the prior art regarding sleep score and this invention, (a) is a graph which shows the correlation with the conventional sleep score and deep sleep rate, (b) is the sleep score of this invention, and deep sleep. It is a graph which shows the correlation with a rate. It is a graph which shows the comparison with the prior art and this invention regarding a sleep score. Is a diagram showing an example of a sleep diary SD _1. It is a diagram showing an example of a sleep time table ST ₁ and wake-up time table WT _1. It is a figure which shows the example of a sleep habit pattern table. Is a diagram showing an example of a sleep diary SD _2. It is a diagram showing an example of a bedtime table ST ₂ and wake-up time table WT _2. Is a diagram showing an example of a sleep diary SD _3. It is a diagram showing an example of a bedtime table ST ₃ and waking-up time table WT _3. Is a diagram showing an example of a sleep diary SD _4. It is a diagram showing an example of a bedtime table ST ₄ and waking-up time table WT _4. It is explanatory drawing which shows the screen of a determination result.

[First Embodiment] A mode for carrying out a sleep evaluation apparatus according to a first embodiment of the present invention will be described below with reference to the drawings.

First, the configuration of the sleep evaluation apparatus of this embodiment will be described with reference to FIGS. 1 and 2. FIG. 1 is an external view when the sleep evaluation apparatus 1 is used, and FIG. 2 is a block diagram of the sleep evaluation apparatus 1. As shown in FIG. 1, the sleep evaluation device 1 is connected to the sensor unit 2 (biological information detection means) that detects biological information of a subject lying on the bedding and outputs it as a biological signal, and is in the sleep stage. And a control box 3 that performs determination and evaluation of sleep quality. The control box 3 includes a display unit 4 that performs guidance display such as a sleep stage determination result and a sleep evaluation index, and an operation unit 5 that performs operations such as power on / off or measurement start / end.

The sensor unit 2 detects, for example, a pressure fluctuation of a mattress enclosing an incompressible fluid by using a microphone (for example, a condenser microphone). As shown in FIG. It is intended to detect changes in the biological signal and posture of the subject in the supine position.

As shown in FIG. 2, in the control box 3, the sensor unit 2, the display unit 4, and the operation unit 5 are connected to the CPU 6. The CPU 6 also includes a biological data detection unit 7 that detects each of a respiratory signal, a body motion signal, and a heartbeat signal from the biological signal detected by the sensor unit 2, a determination unit 8 that performs various determinations and calculations for sleep evaluation, A storage unit 9 that stores various conditional expressions, determination results, and calculation results for sleep stage determination and sleep evaluation, an evaluation unit 20 that evaluates sleep quality, and a power supply 10 that supplies power to the sleep evaluation device 1 And connected to. In this case, the CPU 6 includes a control unit that controls the sleep evaluation apparatus 1 and a time measuring unit that measures time. More specifically, the determination unit 8 includes an entering / leaving determination unit 11, a body movement determination unit 12, a wake determination unit 13, a sleep determination unit 14, a deep sleep determination unit 15, a REM / light sleep determination unit 16, a midway An awakening determination unit 17 and a wakeup determination unit 18 (not shown) are included. Each of these determination units will be described later using a flowchart.

Furthermore, the determination unit 8 includes a sleep determination data calculation unit 30 (not shown), a sleep evaluation score calculation unit 40 (not shown), and a discrimination probability calculation unit 50 (not shown).

The sleep determination data calculation unit 30 calculates sleep determination data (a plurality of variable data) as a basis for calculating a sleep evaluation score (described later). Sleep determination data includes deep sleep rate (%), differential sleep cycle score, total bedtime (minutes), sleep cycle (minutes), deep sleep appearance amount (minutes), differential total bedtime score, mid-long time awakening It is preferable to use nine types of data such as the number of times (times), the number of times of short-term awakening (times), and sleep efficiency (%). Therefore, as the sleep determination data calculation unit 30, the deep sleep rate calculation unit, the difference sleep cycle score calculation unit, the total bedtime calculation unit, the sleep cycle calculation unit, the deep sleep appearance amount calculation unit, the difference total bedtime score calculation unit , A medium / long-time wake-up number calculating section, a short-time wake-up number calculating section, and a sleep efficiency calculating section (all not shown). In the present embodiment, a sleep evaluation system and a sleep evaluation device using these nine types of sleep determination data will be described, but sleep determination data (variable data) other than these may be further added.

Here, the nine types of sleep determination data, which are a plurality of variable data, will be further described. Deep sleep rate (%) means the ratio of deep sleep in sleep time, and “(deep sleep time / sleep time) × 100”, that is, “(deep sleep time / (time from falling asleep to final awakening)” ) × 100 ”. The differential sleep cycle score is a score representing how much the sleep cycle (minute) is different from the reference time (for example, 90 minutes). “− | Sleep cycle−predetermined reference time |” (|| represents an absolute value). Therefore, when the reference time is set to 90 minutes, if the sleep cycle of the subject is 90 minutes, the value of the differential sleep cycle score is 0 and shows the maximum value, and if the sleep cycle is 120 minutes or 60 minutes. The value of the differential sleep cycle score is −30. Total bedtime (minutes) means the time from bed to bed. The sleep cycle (minutes) means an average value of the cycle when one cycle is from the end of the REM sleep to the end of the next REM sleep. However, the first period is from the end of sleep to the end of REM sleep that appears first. Deep sleep appearance amount (minute) means the sum total of deep sleep time. The difference total bedtime score is a score indicating how much the total bedtime (minutes) is different from the reference time (for example, 6.5 hours (390 minutes)). “− | Total bedtime−reference time |” (|| represents an absolute value). Therefore, when the reference time is set to 390 minutes and the total bedtime of the subject is 390 minutes, the value of the difference total bedtime score is 0 and shows the maximum value, and the total bedtime is 420 minutes. Alternatively, if it is 360 minutes, the value of the difference total bedtime score is −30. The number of awakening times (times) means the number of times of awakening over a reference time (for example, 2 minutes 30 seconds) that appears during sleep. The number of short-time awakenings (times) means the number of times of awakening within a reference time (for example, 2 minutes) that appears during sleep. The sleep efficiency (%) means the ratio of the actual sleep time to the total bedtime, which is “(total sleep time / total bedtime) × 100”, that is, “((total bedtime−sleeping time) (Total sum of hours awakened) / total bedtime) × 100 ”.

In the present invention, the nine types of sleep determination data (predetermined items) are extracted as having good correlation between the PSG measurement data and the measurement data of the existing sleep evaluation device, and in particular (as an example, 90 Since it has a differential sleep cycle score (based on minutes) and a differential total bedtime score (based on 6.5 hours (390 minutes) as an example), it also has a sleep time and sleep cycle not found in the prior art It is possible to calculate the sleep quality evaluation in consideration, that is, the sleep score, and to improve the correlation of the sleep score with respect to the result of the sleep determination of PSG.

The sleep evaluation score calculation unit 40 calculates a sleep evaluation score that is a basis for evaluation of sleep quality. As an example, the sleep evaluation score is composed of a four-component score including a sleep depth score as the first component score, a sleep cycle score as the second component score, a sleep time score as the third component score, and a mid-wake score as the fourth component score. It is preferable to do this. Therefore, the sleep evaluation score calculation unit 40 includes a sleep depth score calculation unit, a sleep cycle score calculation unit, a sleep time score calculation unit, and a midway awakening score calculation unit (all not shown). When evaluating the quality of sleep more accurately, the depth of sleep, the sleep cycle, the sleep time, and the degree of arousal during mid-level are important indicators. Therefore, in the present embodiment, the four types of component scores are used. Although the sleep evaluation device will be described, sleep evaluation scores other than these may be further added.

The discrimination probability calculation unit 50 calculates a discrimination probability (described later) indicating the probability of being a sleep disorder person such as a SAS patient.

The evaluation unit 20 calculates a sleep score (sleep index) based on the sleep evaluation score calculated by the sleep evaluation score calculation unit 40 based on the sleep determination data and the determination probability calculated by the determination probability calculation unit 50. The result of sleep quality evaluation including this sleep score is displayed on the display unit 4. The processes executed by the sleep determination data calculation unit 30, the sleep evaluation score calculation unit 40, the discrimination probability calculation unit 50, and the evaluation unit 20 will be described later using each flowchart. The biometric data detection unit 7, the determination unit 8, the evaluation unit 20, the sleep determination data calculation unit 30, the sleep evaluation score calculation unit 40, and the discrimination probability calculation unit 50 are executed by the CPU 6 executing a predetermined program. Those functions may be realized.

Next, the main operation of the sleep evaluation apparatus 1 will be described with reference to the flowcharts of FIGS. FIG. 3 is a flowchart showing a main operation, and FIG. 4 is a flowchart showing a flow of sleep stage determination using the respective determination units 11 to 18.

First, as shown in FIG. 3, when the sleep evaluation apparatus 1 is turned on by turning on the operation unit 5, the guidance for instructing to take a sleeping posture and perform the measurement start operation of the operation unit 5 in step S 1. Is displayed on the display unit 4 to determine whether or not a measurement start operation has been performed. If measurement start operation is not performed, it will progress to NO and will continue displaying the said guidance in step S1. When the measurement start operation is performed, the process proceeds to YES, and in step S2, a biological signal is detected by the sensor unit 2 and stored in the storage unit 9 as biological signal data together with the time measured by the time measuring unit built in the CPU 6.

In step S3, it is determined whether or not the measurement end operation has been performed. If the measurement end operation has not been performed, the process proceeds to NO. The detection and storage of the biological signal in step S2 is continued, and if the measurement end operation has been performed, the process proceeds to YES. In step S4, each part is controlled to process the biological signal detected by the control part in CPU6. That is, the biological signal data stored in the storage unit 9 is read out, the biological data detection unit 7 detects the respiratory signal, the body motion signal, and the heartbeat signal, and the respective waveforms obtained from these respiratory signal, body motion signal, and heartbeat signal. Are stored in the storage unit 9 as respiratory data, body movement data, and heartbeat data. At this time, it is assumed that respiratory data, body movement data, and heart rate data are stored for each unit section having a predetermined time, for example, 30 seconds as one unit (hereinafter, this unit section is referred to as “epoch”). Note that the calculation of the amplitude and period of the waveform of the respiratory signal, the body motion signal, and the heartbeat signal is already known and will be omitted. The length of the epoch is not limited to 30 seconds, and can be set to an arbitrary value within a range that does not impair the accuracy of determination.

When respiration data, body motion data, and heart rate data are detected and stored for all the biological signal data stored in the storage unit 9, the respiration data, body motion data, and heart rate data are stored in step S5. Using each of the determination units 11 to 18 in the determination unit 8, a sleep stage determination (described later) is performed.

In step S6, based on the result of the sleep stage determination, the sleep determination data (a plurality of variable data) is calculated, the sleep evaluation score is calculated, and the discrimination probability is calculated to calculate the sleep score. In step S 7, the result of the sleep quality evaluation including the sleep score is displayed on the display unit 4. In step S8, it is determined whether or not a power-off operation has been performed on the operation unit 5. If the power-off operation has not been performed, the process proceeds to NO, and the display in step S7 is continued. If the power-off operation has been performed, the process proceeds to YES. Then, the sleep evaluation apparatus 1 is turned off and the process ends.

Next, the flow of sleep stage determination using the determination units 11 to 18 in the determination unit 8 will be described using the flowchart of FIG. The determination unit 8 is controlled by the CPU 6 and sequentially performs the following determination processing based on the respiratory data, body movement data, and heart rate data stored for each epoch in the storage unit 9 in step S4 of FIG. .

In step S11 (entrance / leaving determination step), the entrance / leaving determination unit 11 determines whether to enter or leave the floor between the start of measurement and the end of measurement based on changes in respiratory data, body movement data, and heart rate data. Make a decision. In step S12 (body motion determination step), the body motion determination unit 12 performs rough body motion that is a large motion such as rolling over based on the amplitude or period of a waveform obtained from respiratory data, body motion data, and heart rate data. Among the states of small body movements such as snoring and non-body movements obtained in a stable breathing / heartbeat / body movement state, it is determined which state each epoch is in. In step S13 (awake determination step), the awake determination unit 13 determines whether or not the state is a clear awake state based on the determined body movement state. In step S14 (sleep onset determination step), the sleep onset determination unit 14 determines in which epoch the sleep state has shifted from the awakened state immediately after entering the bed (hereinafter referred to as the sleep onset period or sleep onset latency). . In step S15 (deep sleep determination step), the deep sleep determination unit 15 determines whether or not the patient is in a deep sleep state from the fluctuations in the respiratory data and the heart rate data and the determined body movement state. In step S16 (REM / shallow sleep determination step), the REM / shallow sleep determination unit 16 determines whether the deep sleep determination unit 15 determines that the sleep state is a REM sleep state or a shallow sleep state. Determine whether. In step S 17 (midway awakening determination step), the midway awakening determination unit 17 determines the presence or absence of an awakening state during the sleep state based on the duration of body movement. In step S18 (wakeup determination step), the wakeup determination unit 18 determines in which epoch the transition from the sleep state to the wakeup state (hereinafter referred to as a wakeup section) is made.

When all the above determinations are completed, the process returns to the flowchart showing the main operation in FIG. 3, and after the sleep score calculation process in step 6 is executed, the result of the sleep quality evaluation including the sleep score in step S 7. Is displayed.

The processing of each of the determination units 11 to 18 will be described step by step with reference to the flowcharts of FIGS. However, it is assumed that the constants indicated by alphabets and the like are set based on the correlation between sleep stage determination based on polysomnographic examination data and actual measurement data obtained by the sleep evaluation device 1.

The processing of the entrance / leaving determination unit 11 will be described using the flowchart of FIG.
In step S21, the total number of epochs set for the respiratory data, body motion data, and heart rate data stored in the storage unit 9 is defined as an nmax section, and for each epoch from the n = 1 section to the n = nmax section. Therefore, n = 0 is initialized. Subsequently, in step S22, as n = n + 1, the respiration data of the corresponding epoch in the advance storage unit 9 is read by one epoch.

In step S23, A is the minimum value of the respiratory amplitude that is recognized when a person is in the normal supine position, and the amplitude of the respiratory waveform in the epoch n is greater than or equal to the magnitude t (sec). It is determined whether or not the process continues (see FIG. 6). Here, A and t are constants, and t <unit time. If this is the case, it is determined that respiration is detected, and the process proceeds to YES. In step S24, the subject is determined to be in the floor, and the epoch n is determined to be the floor section. In step S25, the corresponding epoch is detected. n is stored in the storage unit 9 in association with n. If the condition in step S23 is not satisfied, it is determined that no respiration is detected, and the process proceeds to NO. In step S27, the subject is determined to be out of bed, and epoch n is determined as a bed leaving section. In S25, it is stored in the same manner as described above. In step S26, it is determined whether or not the entry / exit determination has been made at all epochs nmax, that is, n = nmax. If all epochs have not been determined, the process proceeds to NO and again at step S22, n = n + 1 is set. When entering / leaving determination is repeated and all epochs are determined, the process proceeds to YES, and the process returns to the flowchart of FIG. 4 to proceed to the next determination. In addition, although the example using respiration data was demonstrated, it cannot be overemphasized that body movement data and heart rate data may be used. Based on the processing result of the entrance / exit determination unit 11, the total bedtime calculation unit and the differential total bedtime score calculation unit are configured to calculate the total bedtime (minutes) and the differential total employment as the sleep determination data. The floor time score can be calculated. Further, based on the processing result of the entrance / leaving determination unit 11, the sleep efficiency calculation unit can calculate “total bedtime” necessary for calculating sleep efficiency (%).

The process of the body movement determination part 12 is demonstrated using the flowchart of FIG.
The process of the body movement determination unit first determines the magnitude of body movement from the amplitude of the waveform of the respiratory signal regardless of the epoch n, and then, depending on the presence or absence of the determined body movement in the epoch n, The body movement state of each epoch n is determined.

Therefore, in step S31, the total respiratory rate from the start of measurement to the end of measurement is set to imax, and i = 0 is initially set. Subsequently, in step S32, i = i + 1 is set to advance by one respiration rate, and a respiration waveform corresponding to each respiration rate i from i = 1 to i = imax in the storage unit 9 is read.

In step S32, the i = i + 2 and i = i + 3 respiration waveforms are further read, and the presence / absence of body movement is determined based on variations in the amplitudes of the three consecutive respiration waveforms. That is, in step S33, it is determined whether or not the standard deviation of the amplitudes of the three respiratory waveforms ≧ B1 (where B1 is a constant indicating a threshold value indicating whether the respiratory waveform is stable). If the standard deviation is smaller than B1, it is determined that the respiration waveform is stable because the variation in respiration is small, and the process proceeds to NO. In step S38, of the three consecutive respiration waveforms, i = i + 1th respiration is It is determined to indicate an inanimate state.

If the standard deviation is greater than or equal to B1 in step S33, it is determined that there is a body movement due to large variations in respiration, and the process proceeds to YES. In step S34, the amplitude of the i = i + 1th respiration waveform is greater than or equal to ≧ It is determined whether or not B2 (where B2 is the maximum value of the respiration amplitude recognized when the person is in the normal supine position, and B2> A). If the magnitude of amplitude is greater than or equal to B2, the process proceeds to YES, and in step S35, it is determined that i = i + 1-th breathing is in a rough body motion state (see FIG. 9). Further, if the magnitude of the amplitude is smaller than B2, the process proceeds to NO, and in step S36, the magnitude of the body motion is determined based on the period of the respiratory waveform. That is, it is determined whether or not i = i + 1th respiration cycle ≧ B3. If the breathing cycle ≧ B3, the process proceeds to YES, and it is determined that the body motion and state are present. If respiratory cycle <B3, the process proceeds to NO. In step S37, it is determined that both the amplitude and cycle of the respiratory waveform are small, but the fluctuation is large, and it is determined that the body is in a thin body motion state (see FIG. 10).

As described above, when the state of each body motion of coarse body motion, thin body motion and inbody motion is determined, in step S39, it is stored in the storage unit 9 in association with the corresponding respiration rate i. It is determined whether or not the body movement determination is made at the respiration rate imax, that is, whether i = imax. If all the epochs are not determined, the process proceeds to NO, and the body movement determination is repeated again from step S32 as i = i + 1. When the determination of the total respiratory rate is made, the process proceeds to YES, and this time the body movement determination for each epoch n after step S41 is performed (see FIG. 11).

That is, similarly to step S21 and step S22 of FIG. 5, epoch n = 0 is initially set in step S41 of FIG. 7, and respiration data of the corresponding epoch is read as n = n + 1 in step S42. In the following step S43, it is determined whether or not the read epoch n contains the respiratory waveform determined to be in the rough body movement state. Judged as a section. If not, the process proceeds to NO. In step S45, it is determined whether or not the epoch n has the respiratory waveform determined to be the thin body movement state. If there is, the process proceeds to YES. Epoch n is determined to be a thin body motion section. If not, the process proceeds to NO, and in step S47, this epoch n is determined to be a non-movement section.

As described above, when the rough body motion section, the fine body motion section, and the non-body motion section are determined, the determination result is stored in the storage unit 9 in association with the corresponding epoch n in step S48. It is determined whether or not the above determination has been made at the epoch nmax. If all epochs have not been determined, the process proceeds to NO, and body movement determination for each epoch is repeated again from step S42 with n = n + 1, and all epochs are determined. The process proceeds to YES and returns to the flowchart of FIG. 4 to proceed to the next determination.

Processing of the awakening determination unit 13 will be described using the flowchart of FIG.
In order to make a determination for each epoch as described above, in step S51, epoch n = 0 is initially set, and in step S52, the respiration data of the corresponding epoch n is read as n = n + 1. Hereinafter, as shown in FIG. 12, in the subsequent step S53, for example, it is determined whether or not there are a total of 5 epochs in the storage unit 9 of ± 2 sections before and after the read epoch n. If it does not exist, the process proceeds to NO, returns to step S52 again, and proceeds as n = n + 1. If the five sections exist, the process proceeds to YES, and the five sections are read from the storage unit 9. In subsequent step S55, the body motion value Z of each epoch in the five sections is obtained. Based on the determination of the body motion section of the body motion determination unit 12 described in detail with reference to FIG. 7, the body motion value Z is Z = 2 for a rough body motion section, Z = 1 for a thin body motion section, In the case of an inbody movement section, the value is defined as Z = 0. At the same time, based on the body motion value Z of each epoch, the sum of the body motion values Z of the five sections (here, 0 ≦ Z sum ≦ 10, hereinafter, the sum of Z may be referred to as ΣZ). )

In step S56, it is determined whether or not the sum of the body motion values Z of the five sections = 10. If the sum of Z = 10, the process proceeds to YES. In step S57, since all the five sections are coarse body movement sections, the epoch n (the epoch at the center of the five sections) read in step S52 is awakened. It is determined as an awakening section in a state. If the sum of the body motion values Z is less than 10, the process proceeds to NO, and in step S58, it is determined whether or not 5 ≦ the sum of the body motion values Z ≦ 9 as in step S56. If the total sum of Z is within this range, the process proceeds to YES, and in step S59, the epoch n is changed to an unstable interval in which the respiratory state is relatively unstable and the possibility of a REM sleep or a shallow sleep state is high. Is determined. If the sum of Z is not within the above range, that is, if the sum of Z ≦ 4, the process proceeds to NO, and the epoch n can be in a deep sleep state or a shallow sleep state where the respiratory state is relatively stable. Judged as a stable section with high characteristics.

As described above, when the determination of the awakening period, the unstable period, and the stable period is made, the epoch nmax is stored in the storage section 9 in association with the corresponding epoch n in step S61. If it does not exist, the process proceeds to NO, returns to step S52 again, repeats the awakening determination for each epoch with n = n + 1, and proceeds to YES if it exists, as shown in the flowchart of FIG. Return to the next determination.

The process of the sleep detection unit 14 will be described with reference to the flowchart of FIG.
In order to determine an epoch (hereinafter referred to as a sleep interval) that shifts from the initial awake state immediately after entering the bed to the sleep state, the awake determination unit 13 described in detail in FIG. In addition, the sleep interval is defined by determining the initial awake interval in more detail based on the person's sleep tendency.

In order to make a determination for each epoch as described above, in step S71, epoch n = 0 is initially set, and in step S72, the respiration data of the corresponding epoch n is read as n = n + 1. In a succeeding step S73, it is determined whether or not the read epoch n is the unstable section detailed in FIG. However, this unstable section is an unstable section that appears for the first time after the continuation of the initial awakening section. Therefore, if it is not an unstable section, the process proceeds to NO, this epoch is replaced and stored as an awakening section, and the processing from step S72 is repeated until the unstable section is read again. At this time, even in the case where the epoch is in a stable section, in a normal person's breathing, the stable state suddenly appears without going through the unstable state from the initial awakening state immediately after entering the bed. Very Hard to think. Therefore, it can be easily estimated that this stable interval is data with low reliability, and it can be said that it is appropriate to replace it as an awakening interval.

Further, if the epoch n is an unstable section, the process proceeds to YES, and it is determined whether or not there is an epoch determined to be a wake-up section between the epoch n and the predetermined number of sections C1. Here, assuming that the epoch n is a sleep interval, it is difficult to wake up immediately after falling asleep in human sleep. Therefore, the predetermined number of intervals C1 is assumed that the person does not normally wake up immediately after falling asleep. A constant with a range set. Accordingly, if there is a wake-up interval between the epoch n and the predetermined number of intervals C1, the process proceeds to NO, the epoch n is replaced with the wake-up interval and stored, and the processing from step S72 is repeated again. If the awakening section does not exist, the process proceeds to YES. In step S75, the epoch n is set as the sleep (temporary) section, and the sleep section is determined more strictly by the processing after step S77.

The processing after step S77 is awakening up to which epoch among the epochs determined to be unstable after the sleep (temporary) interval, based on three respiratory fluctuation trends near the person's sleep, found by actual measurement. The epoch immediately after that is determined as a sleep interval by determining whether it should be replaced as an interval.

First, in step S77, out of each epoch determined to be an entrance section by the entrance / leaving determination unit 11 described in detail with reference to FIG. Up to the epoch immediately before the (temporary) section is set as a reference range, and in this reference range, the variance σ ² with respect to the respiratory rate for each epoch is obtained. Further, including the reference range, the number of constant intervals α, β and γ from the sleep (provisional) interval (where α, β and γ are constants set as α <β <γ, and the three The time interval suitable for discriminating the respiratory fluctuation tendency is calculated and set from actual measurement.) The epochs that set each range are the α range, β range, and γ range up to the incremented range. Are defined as an α section, a β section, and a γ section, respectively, and in the same manner as the reference range, the variance of the respiration rate of each of these ranges is obtained and is set as σα ² , σβ ^2, and σγ ² . Based on these variances σ ² , σα ² , σβ ^2, and σγ ² , the three respiratory fluctuation tendencies are determined as Condition D, Condition E, and Condition F, respectively.

The first respiratory fluctuation tendency is that a subject's breathing variation is rapidly reduced to a sleep state. Therefore, in step S78, the determination is made based on the condition D defined by the expression “σα ² > σβ ² (Expression 1)” and “σβ ² ≦ C2 (Expression 2)”. That is, as shown in the above equation 1, the variation in the respiratory rate rapidly decreases as the range increases, and as shown in the above equation 2, the variance accompanying the increase in the population becomes smaller than a certain number C2. It is. Here, C2 is a constant that can be determined to be significantly close to the variation in respiratory rate that appears after falling asleep. When this condition D is satisfied, it can be determined that the β section is already in the sleeping state.

Accordingly, if the condition D is satisfied, the process proceeds to YES. In step S79, it is determined that at least the α section is the awakening section, and the epoch immediately after the α section is determined as the sleep period. If the condition D is not met, the process proceeds to NO, and in step S80, the second respiratory fluctuation tendency is determined.

The second tendency of respiratory fluctuation is that a subject's breathing variation is gradually reduced to a sleep state. Therefore, it is determined by the condition E defined by the expression “σ ² × C3 ≧ σα ² ≧ σβ ² (Expression 3)”. Here, C3 is a constant satisfying C3 <1, and is reduced by some percent with respect to variations in the reference range. However, there is a relationship of σ ² × C3> C2 with C2 in the formula 2. Therefore, as shown in Expression 3, the variation of the α range is small and the variation of the β range is further reduced with respect to the variation of the reference range reduced by the C3.

Accordingly, if the condition E is satisfied, the process proceeds to YES. In step S79, since the variation is very gradual but tends to decrease, it is determined that the β period is the awakening period. The epoch immediately after is determined as the sleep interval. On the other hand, if the condition E is not satisfied, the process proceeds to NO, and the third respiratory fluctuation tendency is determined in step S81.

The third respiratory fluctuation tendency is that the fluctuation of the subject's breathing once decreases after the fluctuation once increases compared with the breathing fluctuation of the reference range. Therefore, it is determined by the condition F defined by the expressions “σ ² <σβ ² (Expression 4)” and “σγ ² <σβ ² (Expression 5)”. Here, since this tendency is a phenomenon seen in a relatively long span as compared with the conditions D and E, the conditions using the β range and the γ range are used.

Accordingly, if the condition F is satisfied, the process proceeds to YES, and in step S79, at least the β section in which the variation has increased is determined to be an awakening section, and the epoch immediately after the β section is determined. Determine as sleep interval.

If the condition F is not met, proceed to NO. This is a case where none of the conditions D, E, and F is satisfied. In step S82, the number of sections α, β, and γ is increased by the number of sections δ, respectively, and an α range, β After resetting each of the range and the γ range, the process returns to step S78 again, and the conditions D, E and F are repeated until the sleep interval is determined.

When the sleep interval is determined in step S79, the awake interval and the sleep interval are stored in association with the corresponding epoch n in step S83, and the process returns to the flowchart of FIG. 4 and proceeds to the next determination. . Based on the processing results of the awakening determination unit 13 and the sleep onset determination unit 14, the deep sleep rate calculation unit calculates “sleep time” necessary for calculating the deep sleep rate (%), that is, “from sleep onset to final awakening”. Can be calculated. In addition, based on the processing results of the awakening determination unit 13 and the sleep onset determination unit 14, the sleep efficiency calculation unit calculates the “total amount of time awakened during sleep” necessary for calculating sleep efficiency (%). Is possible.

The process of the deep sleep determination part 15 is demonstrated using the flowchart of FIG.
Here, in the deep sleep state, breathing has a gentle constant rhythm, and body movement hardly occurs, so the following determination is performed.

In order to make a determination for each epoch as described above, in step S91, epoch n = 0 is initially set, and in step S92, the respiration data of the corresponding epoch n is read as n = n + 1. In a succeeding step S93, it is determined whether or not the read epoch n is the stable section detailed in FIG. If it is not the stable section, the process returns to step S92 again and repeats until n = n + 1 and the stable section is met. Further, if it is a stable section, the process proceeds to YES, and in step S94, a condition G in which a large number of determination conditions are combined is determined.

The condition G is “respiration rate in epoch n ≦ H1”, “standard deviation of the period of the respiratory waveform in epoch n ≦ H2”, and “difference in respiratory rate between epoch n and ± 1 interval of epoch n ≦ H3” ”And“ Epoch n is a bodyless motion section ”, the epoch n is determined as a deep sleep section (where H1, H2, and H3 are constants obtained by actual measurement. ).

Accordingly, if the read epoch n satisfies the condition G in step S94, the process proceeds to YES. In step S95, the epoch n is determined to be a deep sleep section, and the determination result is stored in the storage unit 9 in step S96. . If the condition G is not satisfied, the process proceeds to NO. In step S97, it is determined that it is an unstable section. In step S96, the stable section is replaced as an unstable section and stored in the storage unit 9. In step S98, it is determined whether or not the above determination has been made for all epochs nmax. If all epochs have not been determined, the process proceeds to NO, and the steps from step S92 are repeated again. The process returns to the flowchart of FIG. 4 and proceeds to the next determination. Based on the processing result of the deep sleep determination unit 15, the deep sleep appearance amount calculation unit can calculate the deep sleep appearance amount (minute) as the sleep determination data. Further, based on the processing result of the deep sleep determination unit 15, the deep sleep rate calculation unit can calculate “deep sleep time” necessary for calculating the deep sleep rate (%).

The processing of the REM / light sleep determination unit 16 will be described using the flowchart of FIG.
Here, in the REM sleep state, since the increase and fluctuation of the respiration rate occur continuously and the body movement increases, the following determination is performed.

In order to make a determination for each epoch as described above, in step S101, epoch n = 0 is initially set, and in step S102, the respiration data of the corresponding epoch n is read as n = n + 1.

In step S103, it is determined whether or not the read epoch n is n ≠ nmax and is the unstable section detailed in FIG. If epoch n is n ≠ nmax and an unstable section, the process proceeds to YES, and in step S104, the number of continuations of the unstable section is counted as j = j + 1. The condition I is determined as follows: “Average value of respiration rate in epoch ≦ respiration rate of epoch n”. That is, as described above, since an increase in respiratory rate is observed in REM sleep, it is determined whether the respiratory rate of the epoch n is higher than the average respiratory rate during sleep. .

If this condition I is not satisfied, the process proceeds to NO, and in step S106, each epoch from the number of continuations j = 1 to j = j is determined as a shallow sleep section. If the condition I is satisfied, the process proceeds to YES, and an unstable section is detected again with n = n + 1 in step S102.

In step S103, if epoch n is n = nmax or not an unstable section, the process proceeds to NO. In step S110, it is determined whether the number of continuations in the unstable section j = 0, and j = 0. If there is, the process proceeds to YES, returns to step S102 again, and repeats until n = n + 1 corresponding to the unstable section. If j is not 0, the process proceeds to NO. In step S111, whether or not the continuation number j is equal to or greater than the predetermined number jx for the unstable section satisfying the condition I from the continuation number j = 1 to j = j. Or j ≧ jx (where jx is the number of continuations suggesting the possibility of a REM sleep state). If not, the process proceeds to NO, and each epoch from j = 1 to j = j shown in step S106 is determined as a shallow sleep section. If exceeded, the process proceeds to YES, and it is determined in step S112 that each epoch is a REM sleep (provisional) section, assuming that there is a high possibility that the continuation count j = 1 to j is in the REM sleep state.

Here, when there is an apnea state due to sleep apnea syndrome or the like, forced breathing occurs, and therefore, the “respiration rate of epoch n” of the condition I in step S105 increases, and this abnormality The condition I is determined based on the value, and the section to be determined as the shallow sleep section is determined as the REM (provisional) section. Therefore, in step S113, the stable section during sleep (that is, the deep sleep state or the shallow sleep state) is determined based on the determination of the condition K that “the number of stable sections in the entire bed section / (the number of all bed sections−the awake section) ≧ k”. ) Is determined whether or not at least a predetermined ratio k is present, thereby determining whether or not at least generally normal sleep is maintained. If the condition K is satisfied, sleep is normal and the determination of the condition I is determined to be appropriate, and the process proceeds to YES. In step S115, the epochs of the number of continuations j = 1 to j are defined as REM sleep intervals. decide. If the condition K is not satisfied, it is determined that there is an abnormal sleep state, the process proceeds to NO, and the determination based on the condition L is performed in step S114.

Condition L is the determination of “(maximum respiratory rate in each segment−minimum respiratory rate in each segment) / REM sleep (provisional) segment number ≧ Lx” in the REM sleep (provisional) interval from the number of continuous times j = 1 to j. Therefore, even if there is a variation in the respiration rate, it is determined whether or not the respiration rate is within a normal range, and Lx is a minimum value that defines that respiration is abnormal. That is, it is assumed that an apnea state appears in any of the REM sleep (provisional) sections from the number of continuous times j = 1 to j. Therefore, if the condition L is satisfied, that is, if there is an abnormality in breathing, the process proceeds to YES, and in step S106, each epoch from j = 1 to j as the REM sleep (provisional) section is set as the shallow sleep section. Determine as. Further, if the condition L is not satisfied, that is, if the breathing is normal, the process proceeds to NO, and in step S115, the epochs of the number of continuations j = 1 to j as the REM sleep (provisional) section are set as the REM sleep. Determined as a section.

When the REM sleep interval and the light sleep interval are determined, they are stored in the storage unit 9 in association with each epoch n in step S107. In step S108, the continuation number j is once reset to 0, and in step S109, all epoch nmax In step S102, it is determined whether the above determination has been made. If all epochs have not been determined, the process proceeds to NO. The steps from step S102 are repeated, and if all epochs have been determined, the process proceeds to YES. Return to the next determination. Based on the processing result of the REM / shallow sleep determination unit 16, the sleep cycle calculation unit and the differential sleep cycle score calculation unit may calculate a sleep cycle (minutes) and a differential sleep cycle score as the sleep determination data. It becomes possible.

The process of the midway awakening determination unit 17 will be described using the flowchart of FIG.
Even in the sleep state, if the body movement continues for a certain time or more, it can be understood that the user has awakened in the middle, and the following determination is made.

In the same manner as described above, in order to perform the determination for each epoch, in step S121, epoch n = 0 is initially set, and in step S122, the respiration data of the corresponding epoch n is read as n = n + 1. In step S123, the read epoch n is n ≠ nmax, and the coarse body motion, the fine body motion, and the non-body motion determined by the body motion determination unit 12 described in detail in FIG. It is determined whether it is either a section or a thin body movement section (hereinafter referred to as a body movement section).

If epoch n is n ≠ nmax and is a body motion section, the process proceeds to YES, and in step S124, the number of continuations is counted as m = m + 1. In step S122, detection of the body motion section is repeated with n = n + 1. If epoch n is n = nmax or is a body movement section, the process proceeds to NO, and in step S125, the number of continuations m is m ≧ mx (where mx is the possibility of mid-wakefulness) In the case of m ≧ mx, the process proceeds to YES, and in step S126, each epoch from m = 1 to m = m is in the awake state. Even if it is determined that each epoch is stored as a deep sleep interval, a shallow sleep interval, and a REM sleep interval, each epoch is replaced with an awakening interval and stored in the storage unit 9, and in step S127, The number of continuations m is once reset to zero.

If the continuation count m does not exceed mx, the process proceeds to NO, and the continuation count m = 0 is set as it is in step S127. In step S128, it is determined whether or not the above determination has been made for all epochs nmax. If all epochs have not been determined, the process proceeds to NO. The steps from step S102 are repeated again, and if all epochs have been determined, YES is determined. In the middle awakening condition determination described in detail in FIG. 17, the middle awakening is determined in detail according to each condition defined based on the tendency of the person during the middle awakening, found by the inventors. After this determination is made, the process returns to the flowchart of FIG. 4 and proceeds to the next determination. Based on the processing result of the midway awakening determination unit 17, the medium / long-time awakening frequency calculation unit and the short-time awakening frequency calculation unit calculate the medium / long-time awakening frequency (times) and the short-time awakening frequency as the sleep determination data ( Times) can be calculated.

Here, mid-wake condition determination will be described using the flowchart of FIG. In the mid-wake condition determination, epoch n = 0 is initially set in step S131, and respiration data of the corresponding epoch n is read as n = n + 1 in step S132.

In step S133, first, the state of body movement for each epoch n is obtained. That is, in the description of the body motion determination unit 12 described above, the coarse body motion, the thin body motion, and the non-body motion state stored in association with one breath i in step S39 of the flowchart of FIG. In the same way, U = 2 is set for the moving state, U = 1 is set for the thin body moving state, and U = 0 is set for the non-body moving state. The state of motion is obtained as the sum of U (hereinafter referred to as ΣU).

Next, it is determined whether or not ΣU ≧ 2 at the epoch n. If ΣU ≧ 2, the process proceeds to YES, and in step S134, the number of continuations is counted as m = m + 1. Further, if ΣU ≧ 2, the process proceeds to NO, and the number of continuations counted up to the previous time is m ≧ mp in step S135 without counting the number of continuations (where mp includes the possibility of mid-wakefulness) It is a constant indicating the number of continuation of body motion sections, and it is determined whether or not mp <mx. If m ≧ mp is not satisfied, the process proceeds to NO because there is no possibility of awakening, and the number of continuations is returned to m = 0 in step S140. If m ≧ mp, it can be said that there is a possibility that each epoch n from the number of continuations m = 1 to m = m is in the awake state, so the process proceeds to YES and the next condition determination is performed.

That is, in step S136, “(number of epochs n of ΣU ≧ 10) ≧ m1%” among the epochs from the continuation number m = 1 to m = m (where m1 is a ratio to the total continuation number m). It is determined whether or not it is a constant. If this condition is met, the process proceeds to YES, and in step S139, each epoch n from the number of continuations m = 1 to m = m is replaced as an awakening section and stored in the storage unit 9. If the condition is not met, the process proceeds to NO and the next condition determination is performed.

That is, in step S137, “(number of epochs n of ΣU ≧ 10) ≧ m2%” (where m2 is a constant indicating the ratio to the total number of continuations m, and m2 <m1). It is determined whether or not to do. If this condition is not met, it is determined that there is no possibility of awakening during m = 1 to m = m, and the process proceeds to NO, and the number of continuations is returned to m = 0 in step S140. If the above condition is met, it is determined that there is a possibility of awakening midway, the process proceeds to YES, and further conditions are added.

That is, whether or not “(average respiratory rate of all epochs from m = 1 to m = m) ≧ (average respiratory rate of all epochs from n = 1 to n = nmax) × mq” in step S138. Determine whether. Here, mq is a constant of mq> 1, and since the respiratory rate in the awake state is generally higher than the respiratory rate in the sleep state, from n = 1 including the sleep state. If the average respiratory rate of epochs from m = 1 to m = m is larger than mq times the average respiratory rate of epochs up to n = nmax, it can be said that it is clearly determined that the patient is in an awake state.

If the above conditions are not satisfied, it is determined that there is no possibility of awakening during m = 1 to m = m, and the process proceeds to NO, and the number of continuations is returned to m = 0 in step S140. If the condition is satisfied, the process proceeds to YES, and in step S139, the epoch n from the number of continuations m = 1 to m = m is replaced as a wake-up section and stored in the storage unit 9; In S140, the continuation count is returned to m = 0. In step S141, it is determined whether or not the above determination has been made for all epochs nmax. If all epochs have not been determined, the process proceeds to NO. The steps from step S132 are repeated again, and if all epochs have been determined, YES is determined. Proceed and return to the flowchart of FIG.

The process of the wakeup determination unit 18 will be described using the flowchart of FIG.
In step S151, epoch n = nmax is set, and in step S152, n = n−1 is set back in time, and the corresponding epoch n is read. In step S153, whether or not the read epoch n is determined to be a sleep state, that is, corresponds to any one of a deep sleep section, a shallow sleep section, and a REM sleep section (hereinafter referred to as a sleep section). Determine whether or not. If it does not correspond to the sleep section, the process proceeds to NO, and the detection of the sleep section is repeated with n = n−1 again in step S152. If the epoch n is a sleep section, the process proceeds to YES, and this epoch n is defined as a wake-up (temporary) section in step S154. In subsequent step S155, it is determined whether or not there is a wake-up section in each epoch that goes back from the wake-up (temporary) section to a certain number of sections R. Here, in the normal sleep of a person, it can be considered that awakening does not occur a certain time before waking up. Therefore, the certain number of intervals R defines the certain time. If the awakening interval exists, the process proceeds to YES, and in step S158, each epoch from the detected awakening interval to the wake-up (temporary) interval is defined as an awakening interval. In step S154, the detected awakening is detected. The epoch immediately before the section is newly redefined as a wake-up (temporary) section, and the fixed section number R is set again in step S155. In step S155, if there is no awakening section up to a certain number of sections R, the process proceeds to NO. In step S156, the wake-up (temporary) section is determined as the wake-up section. In step S157, The information is stored in the storage unit 9 in association with the corresponding epoch n, and the process returns to the flowchart showing the main operation in FIG. Based on the processing result of the wake-up determination unit 18, the deep sleep rate calculation unit calculates “sleep time” necessary for calculating the deep sleep rate (%), that is, “time from falling asleep to final awakening”. It becomes possible to do.

Then, CPU6 progresses to step S6 of FIG. 3, and performs a sleep point calculation process. In the sleep score calculation process, a sleep score (sleep index) that comprehensively indicates the level of sleep quality is calculated by calculation. In one sleep from when the subject takes a sleeping posture to wake up, the above-described bed / bed determination, body movement determination, arousal determination, sleep determination, deep sleep determination, REM / light sleep determination, mid-wake determination, Sleep determination data (e.g. deep sleep time, number of mid-wakefulness) indicating sleep state can be obtained by wakeup determination. Although these sleep determination data can evaluate the quality of sleep to some extent by itself, the evaluation by itself only evaluates a part of the sleep state. Therefore, in the present embodiment, a plurality of predetermined items (variables) reflecting “depth”, “cycle”, “time”, and “halfway awakening” of sleep are extracted from PSG measurement data, and existing sleep is extracted. As variables correlated with the evaluation device, deep sleep rate (%), differential sleep cycle score, total bedtime (minutes), sleep cycle (minutes), deep sleep appearance amount (minutes), differential total bedtime score, The number of mid- and long-term awakenings (times), short-time awakenings (times), and sleep efficiency (%) were selected. Furthermore, in order to derive a sleep score that aggregates these predetermined items (variables), a principal component analysis is performed to develop a sleep evaluation score, and this sleep evaluation score and sleep apnea syndrome risk are reflected. A regression formula for sleep scores was developed. In addition, the object of the measurement data of PSG analyzed when creating the regression equation in the present embodiment was 49 healthy persons and 112 SAS patients.
As described above, in this embodiment, a sleep evaluation score is selected by a principal component analysis method in order to derive a sleep score that is an evaluation index for comprehensively evaluating the quality of sleep.

First, for a certain population, a plurality of predetermined items (variables) are measured by PSG. In this example, as a plurality of predetermined items, deep sleep rate (%), differential sleep cycle score, total bedtime (minutes), sleep cycle (minutes), deep sleep appearance amount (minutes), differential total bedtime score The number of wakefulness (medium / long-time), the number of short-time wakefulness (times), and sleep efficiency (%).

Second, a correlation coefficient between the plurality of predetermined items is calculated to obtain a correlation matrix. The correlation matrix based on the nine items is given by the determinant shown in the following formula (1). However, r11 to r99 are correlation coefficients.

Third, first to ninth principal components Z1 to Z9 and eigenvectors a11 to a99 are calculated based on the correlation matrix. These are given by the following equations (2) to (7).
Z1 = a11X1 + a12X2 + a13X3 + a14X4 + a15X5 + a16X6 + a17X7 + a18X8 + a19X9 (2)
Z2 = a21X1 + a22X2 + a23X3 + a24X4 + a25X5 + a26X6 + a27X7 + a28X8 + a29X9 (3)
Z3 = a31X1 + a32X2 + a33X3 + a34X4 + a35X5 + a36X6 + a37X7 + a38X8 + a39X9 (4)
Z4 = a41X1 + a42X2 + a43X3 + a44X4 + a45X5 + a46X6 + a47X7 + a48X8 + a49X9 (5)
Z5 = a51X1 + a52X2 + a53X3 + a54X4 + a55X5 + a56X6 + a57X7 + a58X8 + a59X9 (6)
Z6 = a61X1 + a62X2 + a63X3 + a64X4 + a65X5 + a66X6 + a67X7 + a68X8 + a69X9 (7)
Z7 = a71X1 + a72X2 + a73X3 + a74X4 + a75X5 + a76X6 + a77X7 + a78X8 + a79X9 (8)
Z8 = a81X1 + a82X2 + a83X3 + a84X4 + a85X5 + a86X6 + a87X7 + a88X8 + a89X9 (9)
Z9 = a91X1 + a92X2 + a93X3 + a94X4 + a95X5 + a96X6 + a97X7 + a98X8 + a99X9 (10)
However, X1 to X9 are the nine items described above. Eigenvectors a11 to a99 are determined so that the first to ninth principal components Z1 to Z9 are orthogonal to each other. Orthogonal means that they are independent of each other and there is no correlation between them.

Fourth, it is determined whether the eigenvectors a11 to a99 appropriately reflect the meanings of the first to ninth principal components Z1 to Z9. The reason why the eigenvectors a11 to a99 do not appropriately reflect the meanings of the first to ninth principal components Z1 to Z9 is that the item is selected incorrectly. For this reason, the set of items is rejected and another set of items is adopted.

Fifth, eigenvalues λ1 to λ9 of the first to ninth principal components Z1 to Z9 are obtained from the following determinant.

The eigenvalues λ1 to λ9 are related to the dispersion of the first to ninth principal components Z1 to Z9. The larger the eigenvalue, the larger the dispersion, and the smaller the eigenvalue, the smaller the dispersion. Then, the greater the variance, the higher the importance of the corresponding principal component. That is, the larger the eigenvalue, the more appropriately the corresponding principal component represents the whole.

Sixth, the contribution ratios of the first to ninth main components Z1 to Z9 are calculated. The contribution rate is the ratio of the eigenvalues of each principal component to the total of all eigenvalues. When the eigenvalue is calculated based on the correlation matrix, the contribution rate is obtained by dividing each eigenvalue λ1 to λ9 by “9” which is the number of principal components.

Seventh, the first principal components Z1 to Z9 are arranged in descending order of contribution, and when the cumulative contribution exceeds 0.8, the previous principals are adopted as sleep evaluation scores. For example, when analysis of principal components is given in the following table and K3 <0.8 <K4, up to the fourth principal component is adopted as the sleep evaluation score.

Eighth, the factor loading is calculated by multiplying the eigenvector by the square root of the eigenvalue, and those below a predetermined reference value (for example, 0.5) are deleted. Needless to say, it may be used without being deleted.
As described above, through the first to eighth steps, four sleep evaluation scores are selected, and expressions (20) to (23) described later are derived.

Here, the four sleep evaluation scores are obtained by multiply-adding each of the nine items X1 to X9 and the first coefficients a11 to a99 shown in the equations (2) to (10). Since the first coefficients a11 to a99 are eigenvectors, the four sleep evaluation scores are in a linearly independent relationship with each other. That is, four sleep evaluation scores having smaller correlation coefficients than any two of the nine items are generated. Therefore, the sleep evaluation score is obtained by aggregating a plurality (nine in the present embodiment) of predetermined items (variables) from the viewpoint of sleep, and each expresses a characteristic of sleep. Therefore, by evaluating sleep using a sleep evaluation score, it is possible to obtain a more accurate evaluation index as compared to a single predetermined item or a combination of these appropriately. In the present embodiment, the predetermined items include deep sleep rate (%), differential sleep cycle score, total bedtime (minutes), sleep cycle (minutes), deep sleep appearance amount (minutes), differential total bedtime score, Select the number of mid- and long-term awakenings (times), short-time awakenings (times), and sleep efficiency (%). Based on these, four sleep evaluation scores are selected by the principal component analysis method. A “sleep depth score”, a “sleep cycle score” as the second principal component, a “sleep time score” as the third principal component, and a “halfway awakening score” as the fourth principal component could be obtained. “Sleep depth score” is an item indicating deep sleep, “sleep cycle score” is an item indicating sleep cycle, “sleep time score” is an item indicating sleep time, and “midway awakening score” is an item indicating midway awakening.

FIG. 19 is a flowchart showing the flow of each calculation in the sleep score calculation process. FIGS. 20 to 26 show the calculation processes in steps S165, S167, S168, S169, S170, S171, and S172. It is a flowchart which shows the detailed flow of. FIG. 28 is a time chart for explaining predetermined items calculated in the calculation processing. In the following description, it is assumed that the CPU 6 executes these processes according to a predetermined program.

This sleep score calculation process is executed after the subject has completely awakened (that is, woken up). The state in which the subject is completely awakened may be determined to be a complete awakening state when a state in which the subject's breathing is not detected continues for a predetermined period, or a measurement start / end button ( (Not shown) may be provided, and when the end button is pressed down, it may be determined that the state is completely awake. And the memory | storage part 9 memorize | stores the state (namely, result of the sleep stage determination process in step S5) in each epoch after measurement is started (time t0 of FIG. 28) until it ends (time te). And is used for sleep score calculation processing.

As shown in FIG. 19, in the sleep point calculation process, first, a deep sleep rate calculation process (step S161) is executed. Deep sleep rate (%) means the ratio of deep sleep in sleep time, and “(deep sleep time / sleep time) × 100”, that is, “(deep sleep time / (time from falling asleep to final awakening)” ) × 100 ”. “Sleep time” necessary for calculating the deep sleep rate (%), that is, “time from falling asleep to final awakening” is determined by the sleep determination unit 14 as a sleep interval and the associated epoch is read to It is calculated by incrementing to an epoch that is determined to be an awakening section by the determination unit 13 and associated therewith. In addition, as shown in FIG. 28, “deep sleep time” is the total number of epochs of deep sleep time from the start to the end of the measurement (time t0 to te). What is necessary is just to calculate similarly to quantity DT (min).

Subsequently, the CPU 6 executes a differential sleep cycle score calculation process in step S162 of FIG. The differential sleep cycle score is a score representing how much the sleep cycle (minute) is different from the reference time (for example, 90 minutes). “− | Sleep cycle−predetermined reference time |” (|| represents an absolute value). Since the sleep cycle is an average value when one cycle is from the end of the REM sleep to the end of the next REM sleep, the REM / shallow sleep determination unit 16 determines the REM sleep interval and associates it with the epoch The average value may be calculated based on this. The reference time may be 90 minutes, for example, but is not particularly limited.

Subsequently, the CPU 6 executes a total bedtime calculation process in step S163 of FIG. Total bedtime (minutes) means the time from bed to bed. What is necessary is just to calculate as a sum total of the epochs which are determined to be in the floor-entry state by the floor-in / bed-out determination unit 11 and are associated with each other.

Subsequently, the CPU 6 executes a sleep cycle calculation process in step S164 of FIG. As for the sleep cycle, the average value may be calculated based on the epoch that is determined to be the REM sleep section by the REM / shallow sleep determination unit 16 and is associated with the difference sleep cycle score calculation process.

Subsequently, the CPU 6 proceeds to the deep sleep appearance amount calculation process in step S165 of FIG. 19 and executes the deep sleep appearance amount calculation process shown in FIG. The deep sleep appearance amount DT (minutes) means the sum of deep sleep times, and more specifically, as shown in FIG. 28, from the start to the end of the measurement (time t0 to te ) Is the total number of epochs of deep sleep appearance amount. Therefore, as shown in FIG. 28, when the deep sleep appearance amounts DT1 and DT2 are measured during the period from time t0 to te, the deep sleep appearance amount DT = DT1 + DT2.

As shown in FIG. 20, in the deep sleep appearance amount calculation process, first, the process proceeds to the epoch next to the first epoch (step S221). Next, in step S222, it is determined whether or not the epoch is the final epoch Ee. If this determination condition is negative, then in step S223, it is determined whether or not the epoch is in a deep sleep state. If this determination condition is affirmed, the deep sleep appearance amount DT is incremented in step S224 (however, the initial value DT of DT = 0), and the process returns to step S221. On the other hand, if the determination condition in step S223 is negative, the deep sleep appearance amount DT is not incremented, and the process returns to step S221. The process from step S221 to S224 or the process from step S221 to S223 is repeated until the determination condition in step S222 is affirmed. If the determination condition in step S222 is affirmed, the process returns to the flowchart of FIG.

Subsequently, the CPU 6 executes a difference total bedtime score calculation process in step S166 of FIG. The difference total bedtime score is a score indicating how much the total bedtime (minutes) is different from the reference time (for example, 6.5 hours (390 minutes)). Floor time−reference time | ”(|| represents an absolute value). The total bedtime (minutes) may be calculated as the total number of epochs that are determined to be in the bedded state by the bed / leaving determination unit 11 and associated with each other, as in the total bedtime calculation process.

Subsequently, the CPU 6 proceeds to the medium / long-time awakening number calculation process in step S167 of FIG. 19 and executes the medium / long-time awakening number calculation process shown in FIG. The number of awakening times (times) means the number of times of awakening over a reference time (for example, 2 minutes 30 seconds) that appears during sleep. As shown in FIG. 21, in the mid-long-time awakening count calculation process, first, the process proceeds to the epoch next to the first epoch (step S191). Next, in step S192, it is determined whether or not the epoch is the final epoch Ee. If this determination condition is negative, then in step S193, it is determined whether or not the awakening continues for T minutes or more. As described above, since 1 epoch = 30 seconds in this embodiment, when T = 2.5, if 5 or more awakening epochs continue, it is considered that the state is awakening. Therefore, the CPU 6 affirms the determination in step S193 only when a predetermined number (for example, five or more) of epochs in the awake state are continued. Subsequently, the routine proceeds to step S194, and increments the number of wake-up times (WN; where initial value WN = 0). Subsequently, in step S195, the epoch is advanced by the number of continued awakenings, and the process returns to step S191. On the other hand, if the determination condition in step S193 is negative, the process returns to step S191. The processing in steps S191 to S195 or steps S191 to S193 is repeated until it is determined in step S192 that the epoch to be determined is the final epoch Ee. If the determination condition in step S192 is affirmed, the medium / long-term awakening count calculation process ends, and the routine returns to the flowchart of FIG.

Subsequently, the CPU 6 proceeds to the short-time awakening count calculation process in step S168 of FIG. 19 and executes the short-time awakening count calculation process shown in FIG. The number of short-time awakenings (times) means the number of times of awakening within a reference time (for example, 2 minutes) that appears during sleep. As shown in FIG. 22, in the short-time awakening count calculation process, first, the process proceeds to the epoch next to the first epoch (step S231). Next, in step S232, it is determined whether or not the epoch is the final epoch Ee. If this determination condition is negative, it is determined in step S233 whether or not the awakening continues for less than T minutes. As described above, since 1 epoch = 30 seconds in the present embodiment, when T = 2.5, if there are less than 5 continuous awakening epochs, it is regarded as a short-time awakening. Therefore, the CPU 6 affirms the determination in step S233 only when the number of continuous awakening epochs is less than a predetermined number (for example, 5). Subsequently, the routine proceeds to step S234, and increments the short-time awakening count (MN; provided that the initial value MN = 0). Subsequently, in step S235, the epoch is advanced by the number of continued awakenings, and the process returns to step S231. On the other hand, if the determination in step S233 is negative, the process returns to step S231. The processing in steps S231 to S235 or steps S231 to S233 is repeated until it is determined in step S232 that the epoch to be determined is the final epoch Ee. If the determination in step S232 is affirmed, the short-time awakening count calculation process ends, and the routine returns to the flowchart of FIG.

Subsequently, the CPU 6 proceeds to the sleep efficiency calculation process in step S169 of FIG. 19 and executes the sleep efficiency calculation process shown in FIG. The sleep efficiency (%) means the ratio of the actual sleep time to the total bedtime, which is “(total sleep time / total bedtime) × 100”, that is, “((total bedtime−sleeping time) (Total sum of hours awakened) / total bedtime) × 100 ”. That is, the sleep efficiency SE is the total number of epochs from the start of measurement (time t0 in FIG. 28) to the end (time te), which is IA, and the epoch determined to be awake in determination step S183 described later. This is a value obtained by (IA-I) / IA where I is the number of wakefulness (number of awakening epochs). Therefore, in the sleep efficiency calculation process, since the first epoch is in the awake state, the initial value of the number of awakening epochs (I) is set to 1, and it is determined in step S183 that the epoch to be determined is awake. It is incremented every time (step S184).

As shown in FIG. 23, in step S181, the CPU 6 first proceeds to the next epoch. Subsequently, in step S182, it is determined whether or not the epoch is an epoch immediately before the transition to the complete awake state (final epoch Ee in FIG. 28). If this determination is negative, it is determined whether or not the epoch is awake (step S183). If the determination result is affirmative, the process proceeds to step S184, the value of the awakening epoch number (I) is incremented, the process returns to step S181, and the process proceeds to the next epoch. On the other hand, if the determination condition in step S183 is negative, the process returns to step S181 without incrementing the value of the number of awakening epochs (I). The CPU 6 repeats the processes of steps S181 to S183 or steps S181 to S184 unless the determination result of step S182 is positive.

On the other hand, if the determination in step S182 is affirmed, the routine proceeds to step S185, and sleep efficiency SE is calculated. That is, the final value of the awakening epoch number I and the numerical value of the total epoch number IA are substituted into the above formula SE = (IA−I) / IA to obtain the sleep efficiency SE, and the sleep efficiency calculation process ends. Returning to the flowchart of FIG.

Subsequently, the CPU 6 executes each data standardization process in step S170 of FIG. As shown in FIG. 24, in each data standardization process, the standardization process of the value of the sleep determination data acquired in steps S161 to S169 described above is executed. First, in step S241, the standard value Za (st) of the deep sleep rate Za is obtained by Za (st) = (Za−average Za) / standard deviation Za. Here, each average Za and standard deviation Za are each average value and standard deviation value (fixed values) of deep sleep rate Za in the population based on the measurement data of PSG. The population is, for example, a group of X people in their 20s when the subject's age is in their 20s. The test subject inputs his / her parameters (for example, age and sex) in advance using the operation unit 5 to select data regarding an appropriate population and use it for the standardization process. Data regarding this population is stored in the storage unit 9 in advance. The CPU 6 reads the average value and the standard deviation from the storage unit 9 and executes the calculation of step S241. The same applies to the processing in steps S242 to S249.

Similarly, in each step S242 to S249, each sleep determination data is standardized. The standard value of each sleep determination data is obtained by the following formulas (12) to (19) (steps S242 to S249).
Difference sleep cycle score Zb (st) = (Zb−average Zb) / standard deviation Zb (12)
Total bedtime Zc (st) = (Zc−average Zc) / standard deviation Zc (13)
Sleep cycle Zd (st) = (Zd−average Zd) / standard deviation Zd (14)
Deep sleep appearance amount Ze (st) = (Ze−average Ze) / standard deviation Ze (Equation 15)
Difference total bedtime score Zf (st) = (Zf−average Zf) / standard deviation Zf (16)
Number of middle and long awakenings Zg (st) = (Zg−average Zg) / standard deviation Zg (17)
Number of short-time awakenings Zh (st) = (Zh−average Zh) / standard deviation Zh (18)
Sleep efficiency Zi (st) = (Zi−average Zi) / standard deviation Zi (19)
By this standardization process, the deep sleep rate Za, the differential sleep cycle score Zb, the total bedtime Zc, the sleep cycle Zd, the deep sleep appearance amount Ze, the differential total bedtime score Zf, and the number of mid- and long-term awakenings Zg of different scales. It becomes possible to handle the number of short-time awakenings Zh and sleep efficiency Zi by the same process. When the process of step S249 ends, the routine returns to the flowchart of FIG.

Subsequently, the CPU 6 executes each principal component score calculation process in step S171 of FIG. As shown in FIG. 25, in each principal component score calculation process, the standard value of each sleep determination data acquired in step S170 described above is used for the calculation. In the principal component score calculation process, a plurality of types of predetermined items including at least items related to sleep depth, items related to sleep rhythm, and items related to mid-wake awakening extracted based on PSG measurement data Sleep evaluation by multiplying the principal component coefficient for each predetermined item of the sleep evaluation score obtained by performing the principal component analysis on the sleep determination data corresponding to the predetermined item calculated from the biological signal of the subject Calculate the score. More specifically, the nine sleep determination data are deep sleep rate Za, differential sleep cycle score Zb, total bedtime Zc, sleep cycle Zd, deep sleep appearance amount Ze, differential total bedtime score Zf, medium From the long-time awakening count Zg, the short-time awakening count Zh, and the sleep efficiency Zi, four sleep evaluation scores, that is, a “sleep depth score”, a “sleep cycle score”, a “sleep time score”, and a “halfway awake score” calculate. The CPU 6 calculates the “sleep depth score”, “sleep cycle score”, “sleep time score”, and “halfway awakening score” according to the following equations (20) to (23) (steps S251 to S254).

First principal component score (sleep depth score)
= Coefficient C1a * Standard value Za (st) + Coefficient C1b * Standard value Zb (st) + Coefficient C1c * Standard value Zc (st) + Coefficient C1d * Standard value Zd (st) + Coefficient C1e * Standard value Ze (st) + Coefficient C1f * Standard value Zf (st) + Coefficient C1g * Standard value Zg (st) + Coefficient C1h * Standard value Zh (st) + Coefficient C1i * Standard value Zi (st) Equation (20)

Second principal component score (sleep cycle score)
= Coefficient C2a * Standard value Za (st) + Coefficient C2b * Standard value Zb (st) + Coefficient C2c * Standard value Zc (st) + Coefficient C2d * Standard value Zd (st) + Coefficient C2e * Standard value Ze (st) + Coefficient C2f * Standard value Zf (st) + Coefficient C2g * Standard value Zg (st) + Coefficient C2h * Standard value Zh (st) + Coefficient C2i * Standard value Zi (st) Equation (21)

Third principal component score (sleep time score)
= Coefficient C3a * Standard value Za (st) + Coefficient C3b * Standard value Zb (st) + Coefficient C3c * Standard value Zc (st) + Coefficient C3d * Standard value Zd (st) + Coefficient C3e * Standard value Ze (st) + Coefficient C3f * Standard value Zf (st) + Coefficient C3g * Standard value Zg (st) + Coefficient C3h * Standard value Zh (st) + Coefficient C3i * Standard value Zi (st) Equation (22)

Fourth principal component score (midway awakening score)
= Coefficient C4a * Standard value Za (st) + Coefficient C4b * Standard value Zb (st) + Coefficient C4c * Standard value Zc (st) + Coefficient C4d * Standard value Zd (st) + Coefficient C4e * Standard value Ze (st) + Coefficient C4f * Standard value Zf (st) + Coefficient C4g * Standard value Zg (st) + Coefficient C4h * Standard value Zh (st) + Coefficient C4i * Standard value Zi (st) Equation (23)

Here, each coefficient in the equations (20) to (23) is a constant obtained from a principal component analysis method based on nine predetermined items, and is stored in the storage unit 9. CPU6 reads each coefficient from the memory | storage part 9, and performs a calculation process. The principal component score coefficient matrix of the nine predetermined items and the principal component score is as shown in Table 2.

When the process of step S254 ends, the routine returns to the flowchart of FIG.

Subsequently, the CPU 6 executes the sleep disorder determination probability calculation process in step S172 of FIG. Here, the sleep disorder discrimination probability refers to the probability of being a sleep disorder person such as a SAS patient. As shown in FIG. 26, in the sleep disorder determination probability calculation process, a sleep disorder determination probability obtained by performing logistic regression analysis on the sleep evaluation score is calculated.
Logistic regression analysis (multiple regression analysis) of multiple sleep evaluation scores (explanatory variables) and sleep disorder dummy variables (objective variables) allows the probability of falling into sleep disorder from multiple sleep evaluation scores (sleep disorder) Discrimination probability).
In the logistic regression analysis in this embodiment, the objective variable is represented by a dummy variable of 0/1 (the presence or absence of SAS: SAS group 1 / non-SAS group 0). The explanatory variables are a first principal component score (sleep depth score), a second principal component score (sleep cycle score), a third principal component score (sleep time score), which are sleep evaluation scores calculated from sleep determination data, At least any three of the fourth principal component scores (halfway awakening scores) are used. This makes it possible to create a regression equation of sleep disorder discrimination probability. Here, an example using four sleep evaluation scores will be described with reference to FIG.
The CPU 6 calculates the first principal component score (sleep depth score), the second principal component score (sleep cycle score), the third principal component score (sleep time score), and the fourth principal component score (halfway awakening) calculated in step S171. Using the score, the variable P is calculated by the following equation (24) (step S261). At this time, a process of inverting the signs of the third principal component score (sleep time score) and the fourth principal component score (halfway awakening score) may be appropriately performed.
Variable P
= Fixed value F1 * first principal component score + fixed value F2 * second principal component score + fixed value F3 * third principal component score + fixed value F4 * fourth principal component score (24)
Here, the fixed values F1 to F4 are fixed values obtained from a principal component analysis method of a certain population. These fixed values are stored in the storage unit 9 and read out by the CPU 6 and used for calculation.

Next, the CPU 6 calculates the sleep disorder determination probability by the following equation (25) using the variable P calculated in step S261 (step S262).
Sleep disorder discrimination probability = 1 / (1+ (exp- (P))) (Equation 25)
When the process of step S262 ends, the routine returns to the flowchart of FIG.

Then, CPU6 performs the sleep point number calculation process of step S173 of FIG. As shown in FIG. 27, in the sleep score calculation process, the CPU 6 calculates the sleep score (sleep index) by the following equation (26) using the sleep disorder determination probability calculated in step S262 (step S265). ).
Number of sleep points = 100-(Sleep disorder discrimination probability * 100) Equation (26)
CPU6 memorize | stores the sleep score (Score) obtained by step S265 in the memory | storage part 9 (refer step S174 and FIG. 19).
Thus, from equation (26), when the sleep disorder determination probability is high, the sleep score is calculated low, and when the sleep disorder determination probability is low, the sleep score is calculated high. This makes it possible to predict whether or not SAS will develop before SAS develops.

Thus, the regression equations for obtaining the number of sleep points in the present invention are summarized in the following three equations described above.
Variable P = fixed value F1 * first principal component score + fixed value F2 * second principal component score + fixed value F3 * third principal component score + fixed value F4 * fourth principal component score (24)
Discrimination probability = 1 / (1+ (exp- (P))) (Equation 25)
Number of sleep points = 100− (discrimination probability * 100) Equation (26)

Next, the evaluation result display process (step 7, see FIG. 3) will be described. FIG. 29 is a flowchart showing the contents of the evaluation result display process executed by the CPU 6, FIG. 30 is an example of a sleep evaluation screen, and FIG. 31 is an example of a sleep score transition screen.

First, the CPU 6 executes a sleep score comparison process (step S271). In this sleep score comparison process, the sleep score Score obtained by the sleep score calculation process is compared with the first reference value W and the second reference value Y, and the sleep score Score is divided into three stages. Specifically, the sleep point average value + dispersion value of the sleep abnormal group included in a certain population is the first reference value W, and the sleep point average value + dispersion value of the healthy sleep group is the second reference value Y. When the sleep score Score is less than or equal to the first reference value W, the first division, and when the sleep score Score exceeds the first reference value W and is less than or equal to the second reference value Y, the second category and the sleep score Score When the reference value Y is exceeded, the sleep score Score is classified into the third category. Here, the first reference value W is the sleep point average value + dispersion value of the abnormal sleep group included in a certain population, and the second reference value Y is the sleep point average value + dispersion value of the healthy sleep group. is there. These fixed values W and Y are stored in the storage unit 9 and read in the sleep score comparison process.

Next, the CPU 6 displays “bad sleep” on the display unit 4 when the sleep score Score is classified into the first category, and “normal” when the sleep score Score is classified as the second category. If the sleep score is classified into the third category, “Good sleep” is displayed (step S272). For example, in the case of good sleep, the sleep evaluation screen shown in FIG. 30 is displayed on the display unit 4. In this case, it is preferable that the CPU 6 displays the transition of one sleep stage together using the processing results of the step S5 and the step S6 in addition to the processing result of the sleep score comparison processing. By displaying the category in addition to the sleep score, the user can know the quality of sleep quality. Furthermore, since the time course of the sleep stage such as awakening, shallowness, and deepness is displayed, it is possible to make use of it for self-condition management.

Next, the CPU 6 determines whether or not an operation for displaying the next screen has been performed (step S273). The variance value is calculated (step S274), and a sleep point transition screen is displayed on the display unit 4 (step S275). For example, as shown in FIG. 31, a vertical axis is a bar graph with the number of sleep points Score and the horizontal axis. In this case, if the sleep score Score is equal to or less than the average value + dispersion value calculated in step S274, the bar graph is colored and displayed. Thereby, the user can know the day when the quality of sleep was bad, and can use it for physical condition management.

Next, the CPU 6 determines whether or not an operation for displaying the next screen has been performed (step S276). If the operation has been performed, the process is terminated.

According to the present invention, since the deep sleep rate (%) is selected as one of the predetermined items, the deep sleep rate is reflected in the regression equation for evaluating the quality of sleep, and the evaluation ability related to the sleep depth Is improved. FIG. 32 is a graph showing a comparison between the related art regarding the sleep score and the present invention, (a) is a graph showing the correlation between the conventional sleep score and the deep sleep rate, and (b) is a sleep of the present invention. It is a graph which shows the correlation with a score and a deep sleep rate. The sleep evaluation device (Sleep Scan SL-501) manufactured by the applicant was used as the sleep score according to the prior art. Further, the deep sleep rate (%) on the vertical axis uses the measurement results obtained by the sleep evaluation device of the applicant's product for both (a) and (b). It is clear that the correlation between the sleep score of the present invention and the deep sleep rate (FIG. 32 (b)) is better than the correlation between the conventional sleep score and the deep sleep rate (FIG. 32 (a)), Among items related to sleep quality, items related to sleep depth, items related to sleep rhythm, and items related to awakening during sleep, the ability to evaluate items related to sleep depth can be improved.

According to the present invention, the probability of being a sleep disorder person such as a SAS patient (sleep disorder discrimination probability) is calculated, and the sleep score is calculated by reflecting this, so the sleep disorder person and the healthy person can sleep. Differences in quality assessment results can be made. FIG. 33 is a graph showing a comparison between the related art relating to the sleep score and the present invention. As a prior art, a sleep evaluation device (Sleep Scan SL-501) manufactured by the applicant is used. According to the present invention, it can be seen that the difference between the sleep score of a SAS patient and the sleep score of a healthy person appears more markedly than in the case of the prior art.

[Second Embodiment]
Hereinafter, the form for implementing the sleep evaluation system which is 2nd Embodiment by this invention is demonstrated. As shown in the external view of FIG. 1, the sleep evaluation device 1 according to the first embodiment is established as one device including the sensor unit 2 and the control box 3, and the control box 3 includes the device according to the present invention. Since a series of processing programs including a regression equation for obtaining the sleep score is already incorporated, the sleep determination data can be acquired and the sleep score can be calculated only by the sleep evaluation device 1.
On the other hand, the sleep evaluation system according to the second embodiment is for executing a series of processing programs including a measuring device for obtaining a biological signal of a subject and a regression equation for obtaining a sleep score (sleep index) in the present invention. And an information processing terminal. The determination part (equivalent to the determination part 8 of 1st Embodiment) which performs a sleep stage determination etc. based on the said biosignal should just be comprised in either a measuring device or an information processing terminal. The output of the data measured by the measuring device to the information processing terminal is not particularly limited, for example, using a wired or wireless connection means.
According to the present invention, since the regression equation for evaluating the quality of sleep is created based on the measurement data of PSG, a series of processing programs including such a regression equation is stored in the information processing terminal (for example, a personal computer). In addition, it is possible to calculate sleep determination data, which is a plurality of variable data, based on the biological signal detected from the subject by the measuring device, and to evaluate the sleep quality of the subject, that is, to calculate the sleep score. . In addition, since the regression equation for evaluating the quality of sleep according to the present invention is created based on the measurement data of PSG, the acquisition device can measure biological information capable of calculating a predetermined item (sleep determination data). If it is, it will not specifically limit. Therefore, for example, using a PSG measurement device as a measurement device, the sleep determination data can be directly substituted for use in the evaluation of sleep quality, which is also useful in the evaluation of sleep quality in medical institutions. Since the specific process flow is the same as that of the sleep measurement apparatus 1 of the first embodiment, detailed description thereof is omitted.

In the embodiment described above, four sleep evaluation components are selected from nine sleep determination data, but the present invention is not limited to this, and n (n is a natural number of 2 or more) indicating a sleep state. From the sleep determination data, m sleep evaluation scores (m is a natural number satisfying n> m) that are independent of each other may be calculated, and the sleep score may be calculated based on the m sleep evaluation scores. . In this case, sleep determination data includes sleep onset latency, sleep efficiency, number of mid- and long-term awakenings, deep sleep latency, deep sleep time, number of short-time awakenings, deep sleep rate, differential sleep cycle score, differential total bedtime Score, total bedtime, bed rest latency, sleep time, total sleep time, mid-wake time, REM sleep latency, shallow sleep time, REM sleep time, number of transitions to sleep stage, number of shallow sleep appearances, number of REM sleep appearances, Number of deep sleep appearances, REM sleep duration, REM sleep interval time, REM sleep cycle, sleep cycle, ratio of shallow sleep in the first and second half, ratio of REM sleep in the first and second half, ratio of deep sleep in the first and second half , Items related to sleep depth (for example, at least one of deep sleep rate or deep sleep appearance amount), items related to sleep rhythm (for example, at least one of sleep cycle or differential sleep cycle score), and midway awakening Items related to At least one) and the may be arbitrarily selected in sleep efficiency or medium long awakening times. If items relating to sleep depth, sleep rhythm, and awakening that are important indicators in the evaluation of sleep quality are selected, it is possible to appropriately derive an indicator that indicates the overall level of sleep quality. The sleep evaluation score may include any one or more of a sleep depth score, a sleep cycle score, a sleep time score, and a midway awakening score.

In the first embodiment, the sleep evaluation apparatus 1 is exemplified by detection of a respiratory signal using a mattress and a condenser microphone sensor. However, a piezoelectric sensor is assumed to be disposed under the mattress and directly detect a human body pressure fluctuation. A piezoelectric element such as a cable, a capacitive sensor, a film sensor, a strain gauge, or the like may be used, and a known device may be used as long as it can detect a respiratory signal, a body motion signal, and a heartbeat signal.

Further, in step S138 of the midway awakening condition determination described with reference to the flowchart of FIG. 17, “(average respiratory rate of all epochs from m = 1 to m = m) ≧ (n = 1 to n = nmax Under the condition of “epoch average respiratory rate) × mq”, the awakening determination based on the respiratory rate is performed, and the sleep stage is corrected using the heartbeat signal detecting means for detecting the heartbeat-related index and the heartbeat-related index. By further providing a correcting means, for example, “(average heart rate of all epochs from m = 1 to m = m) ≧ (average heart rate of all epochs from n = 1 to n = nmax) × mv” (Here, mv is a constant such that mv> 1), and a condition that satisfies this condition may be determined as an arousal state, and a more accurate arousal determination is possible.

Furthermore, the determination result may be corrected by taking a known correlation using the transition of the determination result of the sleep evaluation device 1 and the transition of the index related to the heartbeat detected by the heartbeat signal detecting means.

In the embodiment described above, nine predetermined items and four sleep evaluation scores have been described as examples. However, the present invention is not limited to this, and n (two or more natural numbers) predetermined items. The sleep score may be calculated using m (n ≧ m, where m is a natural number) sleep evaluation scores.

[Third Embodiment]
Hereinafter, with reference to drawings, the form for implementing the sleep evaluation apparatus which is 3rd Embodiment by this invention is demonstrated.

First, the configuration of the sleep evaluation apparatus according to the present embodiment will be described with reference to FIGS. 34 and 35. FIG. 34 is an external view when the sleep evaluation apparatus 101 is used, and FIG. 35 is a block diagram of the sleep evaluation apparatus 101. As shown in FIG. 34, the sleep evaluation apparatus 101 detects the biological information of the subject lying on the bedding and outputs it as a biological signal, and is connected to the sensor unit 102 and is in the sleep stage. And a control box 103 that performs determination and evaluation of sleep quality. The control box 103 includes a display unit 104 that performs guidance display such as a sleep stage determination result and a sleep evaluation index, and an operation unit 105 that performs operations such as power on / off or measurement start / end.

The sensor unit 102 detects, for example, a pressure fluctuation of a mattress enclosing an incompressible fluid using a microphone (for example, a condenser microphone). As shown in FIG. 34, the mattress is placed under a bedding. It is intended to detect changes in the biological signal and posture of the subject in the supine position.

As shown in FIG. 35, in the control box 103, the sensor unit 102, the display unit 104, and the operation unit 105 are connected to the CPU 106. In addition, the CPU 106 includes a biological data detection unit 107 that detects each of a respiratory signal, a body motion signal, and a heartbeat signal from the biological signal detected by the sensor unit 102, a determination unit 108 that performs various determinations and calculations for sleep evaluation, A storage unit 109 that stores various conditional expressions, determination results, and calculation results for sleep stage determination and sleep evaluation, an evaluation unit 120 that evaluates sleep quality, and a power supply 110 that supplies power to the sleep evaluation device 101 And connected to. In this case, the CPU 106 includes a control unit that controls the sleep evaluation apparatus 101 and a timer unit that measures time. More specifically, the determination unit 108 includes an entering / leaving determination unit 111, a body movement determination unit 112, an arousal determination unit 113, a sleep determination unit 114, a deep sleep determination unit 115, a REM / light sleep determination unit 116, a midway An awakening determination unit 117 and a wakeup determination unit 118 (not shown) are included. Each of these determination units will be described later using a flowchart.

Further, the determination unit 108 includes a sleep determination data calculation unit 130 (not shown), a sleep evaluation score calculation unit 140 (not shown), a discrimination probability calculation unit 150 (not shown), and a sleep type determination unit 160 (not shown). And).

The sleep determination data calculation unit 130 calculates sleep determination data (a plurality of variable data) as a basis for calculating a sleep evaluation score (described later). Sleep determination data includes deep sleep rate (%), differential sleep cycle score, total bedtime (minutes), sleep cycle (minutes), deep sleep appearance amount (minutes), differential total bedtime score, mid-long time awakening It is preferable to use nine types of data such as the number of times (times), the number of times of short-term awakening (times), and sleep efficiency (%). Therefore, as the sleep determination data calculation unit 130, the deep sleep rate calculation unit, the differential sleep cycle score calculation unit, the total bedtime calculation unit, the sleep cycle calculation unit, the deep sleep appearance amount calculation unit, the differential total bedtime score calculation unit , A medium / long-time wake-up number calculating unit, a short-time wake-up number calculating unit, and a sleep efficiency calculating unit (all not shown). In the present embodiment, a sleep evaluation system and a sleep evaluation device using these nine types of sleep determination data will be described, but sleep determination data (variable data) other than these may be further added.

The sleep evaluation score calculation unit 140 calculates a sleep evaluation score that is a basis for evaluation of sleep quality. As an example, the sleep evaluation score includes four components: a sleep depth score as the first principal component score, a sleep cycle score as the second principal component score, a sleep time score as the third principal component score, and a midway awakening score as the fourth principal component score. It is preferable to compose the score. Furthermore, it is preferable to add a body motion frequency score as a score related to the occurrence frequency of body motion occurring during sleep as one of the sleep evaluation scores. Therefore, the sleep evaluation score calculation unit 140 includes a sleep depth score calculation unit, a sleep cycle score calculation unit, a sleep time score calculation unit, a midway awakening score calculation unit, and a body motion frequency score calculation unit (all not shown). When evaluating the quality of sleep more accurately, the depth of sleep, sleep cycle, sleep time, midway awakening, and body motion frequency are important indicators. Although the sleep evaluation apparatus using a component score will be described, a sleep evaluation score other than these may be further added.

The discrimination probability calculation unit 150 calculates a discrimination probability (described later) indicating the probability of being a sleep disorder person such as a SAS patient.

The sleep type determination unit 160 determines which sleep type corresponds to a sleep type among a plurality of preset sleep types (described later).

The evaluation unit 120 calculates a sleep score (sleep index) based on the sleep evaluation score calculated by the sleep evaluation score calculation unit 140 based on the sleep determination data and the discrimination probability calculated by the discrimination probability calculation unit 150. Moreover, the sleep type determination unit 160 determines which sleep type the subject's sleep corresponds to. The result of sleep quality evaluation including such sleep points and sleep type is displayed on the display unit 104. The processes executed by the sleep determination data calculation unit 130, the sleep evaluation score calculation unit 140, the discrimination probability calculation unit 150, the sleep type determination unit 160, and the evaluation unit 120 will be described later using each flowchart. The biometric data detection unit 107, the determination unit 108, the evaluation unit 120, the sleep determination data calculation unit 130, the sleep evaluation score calculation unit 140, the discrimination probability calculation unit 150, and the sleep type determination unit 160 are predetermined by the CPU 106. These functions may be realized by executing the program.

Next, the main operation of the sleep evaluation apparatus 101 will be described using the flowcharts of FIGS. FIG. 36 is a flowchart showing the main operation, and FIG. 37 is a flowchart showing the flow of sleep stage determination using the determination units 111 to 118.

First, as shown in FIG. 36, when the power of the sleep evaluation apparatus 101 is turned on by turning on the power of the operation unit 105, in step S1001, a guidance is given to take a sleeping posture and perform an operation to start the measurement of the operation unit 105. Is displayed on the display unit 104 to determine whether or not a measurement start operation has been performed. If measurement start operation is not performed, it will progress to NO and will continue displaying the said guidance in step S1001. If the measurement start operation is performed, the process proceeds to YES, and in step S1002, a biological signal is detected by the sensor unit 102, and is stored in the storage unit 109 as biological signal data together with the time measured by the timer unit built in the CPU 106.

In step S1003, it is determined whether or not a measurement end operation has been performed. If the measurement end operation has not been performed, the process proceeds to NO, the detection and storage of the biological signal in step S1002 is continued, and if the measurement end operation has been performed, the process proceeds to YES. In step S1004, each unit is controlled to process the biological signal detected by the control unit in the CPU. That is, the biological signal data stored in the storage unit 109 is read, and the biological data detection unit 107 detects the respiratory signal, the body motion signal, and the heartbeat signal, and the respective waveforms obtained from the respiratory signal, the body motion signal, and the heartbeat signal. Are calculated and stored in the storage unit 109 as respiration data, body motion data, and heartbeat data. At this time, it is assumed that respiratory data, body movement data, and heart rate data are stored for each unit section having a predetermined time, for example, 30 seconds as one unit (hereinafter, this unit section is referred to as “epoch”). Note that the calculation of the amplitude and period of the waveform of the respiratory signal, the body motion signal, and the heartbeat signal is already known and will be omitted. The length of the epoch is not limited to 30 seconds, and can be set to an arbitrary value within a range that does not impair the accuracy of determination.

When respiration data, body motion data, and heart rate data are detected and stored for all the biological signal data stored in the storage unit 109, in step S1005, the respiration data, body motion data, and heart rate data are stored. By using each of the determination units 111 to 118 in the determination unit 108, sleep stage determination (described later) is performed.

In step S1006, based on the result of sleep stage determination, sleep determination data (a plurality of variable data) is calculated, sleep evaluation score is calculated, and discrimination probability is calculated to calculate the sleep score. In step S1007, it is determined which sleep type corresponds to the sleep type of the plurality of sleep types set in advance. In step S 1008, the result of sleep quality evaluation including the sleep score and sleep type is displayed on the display unit 104. In step S1009, it is determined whether or not the power of the operation unit 105 has been turned off. If the power is not turned off, the process proceeds to NO, and the display in step S1008 is continued. If the power is turned off, the process proceeds to YES. Then, the sleep evaluation apparatus 101 is turned off and the process ends.

Next, the flow of sleep stage determination using the determination units 111 to 118 in the determination unit 108 will be described using the flowchart of FIG. The determination unit 108 is controlled by the CPU 106 and sequentially performs the following determination processing based on the respiratory data, body motion data, and heart rate data stored for each epoch in the storage unit 109 in step S1004 of FIG. .

In step S1011 (entrance / leaving determination step), the entrance / leaving determination unit 111 determines whether to enter or leave the floor from the start of measurement to the end of measurement based on changes in respiratory data, body movement data, and heart rate data. Make a decision. In step S1012 (body motion determination step), the body motion determination unit 112 performs coarse body motion that is a large motion such as turning over based on the amplitude or period of the waveform obtained from the respiratory data, body motion data, and heart rate data. Among the states of small body movements such as snoring and non-body movements obtained in a stable breathing / heartbeat / body movement state, it is determined which state each epoch is in. In step S1013 (awake determination step), the awake determination unit 113 determines whether the state is a clear awake state based on the determined body movement state. In step S1014 (sleep onset determination step), the sleep onset determination unit 114 determines in which epoch the sleep state transitioned from the awakened state immediately after entering the bed (hereinafter referred to as a sleep onset period or sleep onset latency). . In step S1015 (deep sleep determination step), the deep sleep determination unit 115 determines whether or not the patient is in a deep sleep state from the changes in the respiratory data and the heart rate data and the determined body movement state. In step S1016 (REM / shallow sleep determination step), the REM / shallow sleep determination unit 116 determines whether the deep sleep determination unit 115 determines that the sleep state is a REM sleep state or a shallow sleep state. Determine whether. In step S1017 (middle awakening determination step), the midway awakening determination unit 117 determines the presence or absence of a waking state during the sleep state based on the duration of body movement. In step S1018 (wake-up determination step), the wake-up determination unit 118 determines in which epoch the transition from the sleep state to the wake-up state (hereinafter referred to as the wake-up section) is made.

When all the above determinations are completed, the flow returns to the flowchart showing the main operation in FIG. 36. After the sleep score calculation processing in step 1006 and the sleep type determination in step 1007 are executed, in step S1008 the sleep score and sleep The result of sleep quality evaluation including type is displayed.

The processing of each of the determination units 111 to 118 will be described step by step using the flowcharts of FIGS. 38 to 49, respectively. However, it is assumed that the constants indicated by alphabets and the like are set based on the correlation between sleep stage determination based on polysomnographic examination data and actual measurement data obtained by the sleep evaluation apparatus 101.

The processing of the entering / leaving determination unit 111 will be described using the flowchart of FIG.
In step S1021, the total number of epochs set for the respiration data, body motion data, and heart rate data stored in the storage unit 9 is defined as an nmax interval, and for each epoch from the n = 1 interval to the n = nmax interval. Therefore, n = 0 is initialized. In step S1022, n = n + 1 is set, and the epoch respiration data of the corresponding epoch is read from the advance storage unit 109 for one epoch.

In step S1023, A is the minimum value of the respiratory amplitude that is recognized when a person is in the normal supine position, and the amplitude of the respiratory waveform in the epoch n is t (sec) or greater. It is determined whether or not it continues (see FIG. 39). Here, A and t are constants, and t <unit time. If this is the case, it is determined that respiration is detected, and the process proceeds to YES. In step S1024, it is determined that the subject is in the floor, and the epoch n is determined to be a floor section. In step S1025, the corresponding epoch is detected. The information is stored in the storage unit 109 in association with n. If the condition of step S1023 is not satisfied, it is determined that respiration is not detected, and the process proceeds to NO. In step S1027, the epoch n is determined to be a bed leaving section, assuming that the subject is in the bed leaving step. In S1025, it is stored in the same manner as described above. In step S1026, it is determined whether or not the entrance / leaving determination has been made for all epochs nmax, that is, n = nmax. If all epochs have not been determined, the process proceeds to NO, and in step S1022, n = n + 1 is set again. When entering / leaving determination is repeated and all epochs are determined, the process proceeds to YES, and the process returns to the flowchart of FIG. 37 to proceed to the next determination. In addition, although the example using respiration data was demonstrated, it cannot be overemphasized that body movement data and heart rate data may be used. Based on the processing result of the entrance / exit determination unit 111, the total bedtime calculation unit and the difference total bedtime score calculation unit are configured to calculate the total bedtime (minutes) and the difference total employment as the sleep determination data. The floor time score can be calculated. Further, based on the processing result of the entrance / leaving determination unit 111, the sleep efficiency calculation unit can calculate the “total bedtime” necessary for calculating the sleep efficiency (%).

The process of the body movement determination unit 112 will be described using the flowchart of FIG.
The process of the body movement determination unit first determines the magnitude of body movement from the amplitude of the waveform of the respiratory signal regardless of the epoch n, and then, depending on the presence or absence of the determined body movement in the epoch n, The body movement state of each epoch n is determined.

Therefore, in step S1031, the total respiration rate from the start of measurement to the end of measurement is set to imax, and i = 0 is initially set. Subsequently, in step S1032, i = i + 1 is set to advance by one respiration rate, and a respiration waveform corresponding to each respiration rate i from i = 1 to i = imax in the storage unit 109 is read.

In step S1032, the i = i + 2 and i = i + 3 respiration waveforms are further read, and the presence / absence of body motion is determined based on variations in the amplitudes of the three consecutive respiration waveforms. That is, in step S1033, it is determined whether or not the standard deviation of the amplitudes of the three respiratory waveforms ≧ B1 (where B1 is a constant indicating a threshold value indicating whether the respiratory waveform is stable). If the standard deviation is smaller than B1, it is determined that the respiration waveform is stable because the variation in respiration is small, and the process proceeds to NO. It is determined to indicate an inanimate state.

If the standard deviation is greater than or equal to B1 in step S1033, it is determined that there is body movement due to large variations in respiration, and the process proceeds to YES. In step S1034, the amplitude of the i = i + 1th respiration waveform is greater than or equal to It is determined whether or not B2 (where B2 is the maximum value of the respiration amplitude recognized when the person is in the normal supine position, and B2> A). If the magnitude of amplitude is greater than or equal to B2, the process proceeds to YES, and in step S1035, it is determined that i = i + 1-th breathing is in a rough body motion state (see FIG. 42). If the amplitude magnitude is smaller than B2, the process proceeds to NO. In step S1036, the magnitude of body movement is determined based on the period of the respiratory waveform. That is, it is determined whether or not i = i + 1th respiration cycle ≧ B3. If the breathing cycle ≧ B3, the process proceeds to YES, and it is determined that the body motion and state are present. If breathing cycle <B3, the process proceeds to NO, and in step S1037, it is determined that both the amplitude and cycle of the breathing waveform are small but the fluctuation is large, and it is determined that the body is in a thin body motion state (see FIG. 43).

As described above, when the states of the body movements of the coarse body movement, the thin body movement, and the non-body movement are determined, in step S1039, they are stored in the storage unit 109 in association with the corresponding respiration rate i. It is determined whether or not the body movement determination is made at the respiration rate imax, that is, whether i = imax. If all the epochs are not determined, the process proceeds to NO, and the body movement determination is repeated again from step S1032 as i = i + 1. When the determination of the total respiratory rate is made, the process proceeds to YES, and this time, the body movement determination for each epoch n after step S1041 is performed (see FIG. 11).

That is, similar to steps S1021 and S1022 of FIG. 38, epoch n = 0 is initially set in step S1041 of FIG. 40, and in step S1042, the respiratory data of the corresponding epoch is read as n = n + 1. In the following step S1043, it is determined whether or not the read epoch n includes the respiratory waveform determined to be the rough body movement state. If there is, the process proceeds to YES. In step S1044, the epoch n is subjected to the rough body movement. Judged as a section. If not, the process proceeds to NO. In step S1045, it is determined whether or not the epoch n has the respiratory waveform determined to be the thin body movement state. If there is, the process proceeds to YES. In step S1046, the process proceeds to step S1046. Epoch n is determined to be a thin body motion section. If not, the process proceeds to NO, and in step S1047, this epoch n is determined to be a non-movement section.

As described above, when the rough body motion section, the fine body motion section, and the non-body motion section are determined, the determination result is stored in the storage unit 109 in association with the corresponding epoch n in step S1048, and in step S1049, all the determination results are stored. It is determined whether or not the above determination has been made at epoch nmax. If all epochs have not been determined, the process proceeds to NO, and body movement determination for each epoch is repeated from step S1042 to n = n + 1, and all epochs are determined. The process proceeds to YES and returns to the flowchart of FIG. 37 to proceed to the next determination.

The processing of the awakening determination unit 113 will be described using the flowchart of FIG.
In order to make a determination for each epoch as described above, in step S1051, epoch n = 0 is initially set, and in step S1052, the respiration data of the corresponding epoch n is read as n = n + 1. Thereafter, as shown in FIG. 45, in the subsequent step S1053, for example, it is determined whether or not there are five epochs in total of ± 2 sections before and after the read epoch n in the storage unit 109. If it does not exist, the process proceeds to NO, and the process returns to step S1052 and proceeds as n = n + 1. If the five sections exist, the process proceeds to YES, and the five sections are read from the storage unit 109. In subsequent step S1055, the body motion value Z of each epoch in the five sections is obtained. Based on the determination of the body motion section of the body motion determination unit 112 described in detail with reference to FIG. 40, the body motion value Z is Z = 2 for a rough body motion section, Z = 1 for a thin body motion section, In the case of an inbody movement section, the value is defined as Z = 0. At the same time, based on the body motion value Z of each epoch, the sum of the body motion values Z of the five sections (here, 0 ≦ Z sum ≦ 10, hereinafter, the sum of Z may be referred to as ΣZ). )

In step S1056, it is determined whether or not the sum of the body motion values Z of the five sections is equal to 10. If the sum of Z = 10, the process proceeds to YES. In step S1057, since all the five sections are coarse body movement sections, the epoch n (the epoch at the center of the five sections) read in step S1052 is awakened. It is determined that the awakening section is in the state. If the sum of the body motion values Z is less than 10, the process proceeds to NO, and in step S1058, it is determined whether or not 5 ≦ the sum of the body motion values Z ≦ 9 as in step S1056. If the total sum of Z is within this range, the process proceeds to YES, and in step S1059, the epoch n is changed to an unstable interval in which the respiratory state is relatively unstable and the possibility of a REM sleep or a shallow sleep state is high. Is determined. If the sum of Z is not within the above range, that is, if the sum of Z ≦ 4, the process proceeds to NO, and the epoch n can be in a deep sleep state or a shallow sleep state where the respiratory state is relatively stable. Judged as a stable section with high characteristics.

As described above, when the determination of the awakening period, the unstable period, and the stable period is made, in step S1061, the epoch nmax is stored in the storage unit 109 in association with the corresponding epoch n. If it does not exist, the process proceeds to NO, returns to step S1052, and repeats the awakening determination for each epoch with n = n + 1. If it exists, the process proceeds to YES, and the flowchart of FIG. Return to the next determination.

The processing of the sleep determination unit 114 will be described using the flowchart of FIG.
45. In order to determine an epoch (hereinafter referred to as a sleep interval) that shifts from the initial awake state immediately after entering the bed to the sleep state, the awake determination unit 113 described in detail in FIG. In addition, the sleep interval is defined by determining the initial awake interval in more detail based on the person's sleep tendency.

In order to make a determination for each epoch as described above, in step S1071, epoch n = 0 is initially set, and in step S1072, the respiration data of the corresponding epoch n is read as n = n + 1. In a succeeding step S1073, it is determined whether or not the read epoch n is the unstable section detailed in FIG. However, this unstable section is an unstable section that appears for the first time after the continuation of the initial awakening section. Therefore, if it is not an unstable section, the process proceeds to NO, this epoch is replaced and stored as an awakening section, and the processing from step S1072 is repeated again until the unstable section is read. At this time, even in the case where the epoch is in a stable section, in a normal person's breathing, the stable state suddenly appears without going through the unstable state from the initial awakening state immediately after entering the bed. Very Hard to think. Therefore, it can be easily estimated that this stable interval is data with low reliability, and it can be said that it is appropriate to replace it as an awakening interval.

Further, if the epoch n is an unstable section, the process proceeds to YES, and it is determined whether or not there is an epoch determined to be a wake-up section between the epoch n and the predetermined number of sections C1. Here, assuming that the epoch n is a sleep interval, it is difficult to wake up immediately after falling asleep in human sleep. Therefore, the predetermined number of intervals C1 is assumed that the person does not normally wake up immediately after falling asleep. A constant with a range set. Therefore, if there is an awakening interval between the epoch n and the predetermined number of intervals C1, the process proceeds to NO, the epoch n is replaced with the awakening interval and stored, and the processing from step S1072 is repeated again. If there is no awakening section, the process proceeds to YES, and in step S1075, the epoch n is set as a sleep (temporary) section, and the sleep section is determined more strictly by the processing after step S1077.

The processing from step S1077 onward wakes up to which epoch of epochs determined to be unstable intervals after the sleep (temporary) interval based on the three respiratory fluctuation trends near the person's sleep, found by actual measurement. The epoch immediately after that is determined as a sleep interval by determining whether it should be replaced as an interval.

First, in step S1077, out of each epoch determined to be an entrance section by the entrance / leaving determination unit 111 described in detail with reference to FIG. Up to the epoch immediately before the (temporary) section is set as a reference range, and in this reference range, the variance σ ² with respect to the respiratory rate for each epoch is obtained. Further, including the reference range, the number of constant intervals α, β and γ from the sleep (provisional) interval (where α, β and γ are constants set as α <β <γ, and the three The time interval suitable for discriminating the respiratory fluctuation tendency is calculated and set from actual measurement.) The epochs that set each range are the α range, β range, and γ range up to the incremented range. Are defined as an α section, a β section, and a γ section, respectively, and in the same manner as the reference range, the variance of the respiration rate of each of these ranges is obtained and is set as σα ² , σβ ^2, and σγ ² . Based on these variances σ ² , σα ² , σβ ^2, and σγ ² , the three respiratory fluctuation tendencies are determined as Condition D, Condition E, and Condition F, respectively.

The first respiratory fluctuation tendency is that a subject's breathing variation is rapidly reduced to a sleep state. Therefore, in step S1078, the determination is made based on the condition D defined by the expression “σα ² > σβ ² (Expression 1)” and “σβ ² ≦ C2 (Expression 2)”. That is, as shown in the above equation 1, the variation in the respiratory rate rapidly decreases as the range increases, and as shown in the above equation 2, the variance accompanying the increase in the population becomes smaller than a certain number C2. It is. Here, C2 is a constant that can be determined to be significantly close to the variation in respiratory rate that appears after falling asleep. When this condition D is satisfied, it can be determined that the β section is already in the sleeping state.

Accordingly, if the condition D is satisfied, the process proceeds to YES. In step S1079, it is determined that at least the α section is the awakening section, and the epoch immediately after the α section is determined as the sleep period. If the condition D is not met, the process proceeds to NO, and in step S1080, the second respiratory fluctuation tendency is determined.

Accordingly, if the condition E is satisfied, the process proceeds to YES. In step S1079, the variation is very gradual but tends to decrease. The epoch immediately after is determined as the sleep interval. On the other hand, if the condition E is not satisfied, the process proceeds to NO, and a third respiratory fluctuation tendency is determined in step S1081.

Accordingly, if the condition F is satisfied, the process proceeds to YES, and in step S1079, it is determined that at least the β section in which the variation has increased is the awakening section, and the epoch immediately after the β section is determined. Determine as sleep interval.

If the condition F is not met, proceed to NO. This is a case where none of the conditions D, E, and F is satisfied. In step S1082, the number of sections α, β, and γ is increased by the number of sections δ, respectively, and an α range, β After resetting each of the range and the γ range, the process returns to step S1078, and the conditions D, E, and F are repeated until the sleep interval is determined.

When the sleep interval is determined in step S1079, the wake interval and sleep interval are stored in association with the corresponding epoch n in step S1083, and then the process returns to the flowchart of FIG. 37 and proceeds to the next determination. . Based on the processing results of the awakening determination unit 113 and the sleep onset determination unit 114, the deep sleep rate calculation unit calculates “sleep time” necessary for calculating the deep sleep rate (%), that is, “from sleep onset to final awakening”. Can be calculated. In addition, based on the processing results of the awakening determination unit 113 and the sleep onset determination unit 114, the sleep efficiency calculation unit calculates the “total amount of time awakened during sleep” necessary for calculating sleep efficiency (%). Is possible.

Processing of the deep sleep determination unit 115 will be described using the flowchart of FIG.
Here, in the deep sleep state, breathing has a gentle constant rhythm, and body movement hardly occurs, so the following determination is performed.

In order to make a determination for each epoch as described above, in step S1091, epoch n = 0 is initially set, and in step S1092, the respiration data of the corresponding epoch n is read as n = n + 1. In a succeeding step S1093, it is determined whether or not the read epoch n is the stable section detailed in FIG. If it is not the stable section, the process returns to step S1092 again, and repeats until n = n + 1 until it corresponds to the stable section. On the other hand, if it is a stable section, the process proceeds to YES, and in step S1094, a condition G that is a combination of a large number of determination conditions is determined.

The condition G is “respiration rate in epoch n ≦ H1”, “standard deviation of the period of the respiratory waveform in epoch n ≦ H2”, and “difference in respiratory rate between epoch n and ± 1 interval of epoch n ≦ H3” ”And“ Epoch n is a bodyless motion section ”, the epoch n is determined as a deep sleep section (where H1, H2, and H3 are constants obtained by actual measurement. .

Accordingly, if the read epoch n satisfies the condition G in step S1094, the process proceeds to YES. In step S1095, the epoch n is determined to be a deep sleep section, and the determination result is stored in the storage unit 109 in step S1096. . If the condition G is not satisfied, the process proceeds to NO. In step S1097, it is determined that it is an unstable section. In step S1096, the stable section is replaced as an unstable section and stored in the storage unit 109. In step S1098, it is determined whether or not the above determination has been made for all epochs nmax. If all epochs have not been determined, the process proceeds to NO, and the steps from step S1092 are repeated again. Then, the process returns to the flowchart of FIG. 37 and proceeds to the next determination. Based on the processing result of the deep sleep determination unit 115, the deep sleep appearance amount calculation unit can calculate the deep sleep appearance amount (minute) as the sleep determination data. Further, based on the processing result of the deep sleep determination unit 115, the deep sleep rate calculation unit can calculate “deep sleep time” necessary for calculating the deep sleep rate (%).

Processing of the REM / shallow sleep determination unit 116 will be described using the flowchart of FIG.
Here, in the REM sleep state, since the increase and fluctuation of the respiration rate occur continuously and the body movement increases, the following determination is performed.

In order to make a determination for each epoch in the same manner as described above, in step S1101, epoch n = 0 is initially set, and in step S1102, the respiratory data of the corresponding epoch n is read as n = n + 1.

In step S1103, it is determined whether or not the read epoch n is n ≠ nmax and is the unstable section detailed in FIG. If epoch n is n ≠ nmax and an unstable section, the process proceeds to YES, and in step S1104, the number of continuations of the unstable section is counted as j = j + 1. The condition I is determined as follows: “Average value of respiration rate in epoch ≦ respiration rate of epoch n”. That is, as described above, since an increase in respiratory rate is observed in REM sleep, it is determined whether the respiratory rate of the epoch n is higher than the average respiratory rate during sleep. .

If this condition I is not satisfied, the process proceeds to NO, and in step S1106, each epoch from the number of continuations j = 1 to j = j is determined as a shallow sleep section. If the condition I is satisfied, the process proceeds to YES, and in step S1102, n = n + 1 is set again to detect an unstable section.

In step S1103, if epoch n is n = nmax or not an unstable section, the process proceeds to NO. In step S1110, it is determined whether the number of continuations in the unstable section j = 0, and j = 0. If yes, the process proceeds to YES, returns to step S1102 again, and repeats until n = n + 1 and corresponds to the unstable section. If j is not 0, the process proceeds to NO. In step S1111, whether or not the continuation number j is equal to or greater than the predetermined number jx for the unstable section satisfying the condition I from continuation number j = 1 to j = j. Or j ≧ jx (where jx is the number of continuations suggesting the possibility of a REM sleep state). If not, the process proceeds to NO, and each epoch from j = 1 to j = j shown in step S1106 is determined as a shallow sleep section. If exceeded, the process proceeds to YES, and in step S1112, the epoch is determined to be a REM sleep (provisional) section, assuming that there is a high possibility that the continuation count j = 1 to j is in the REM sleep state.

Here, when there is an apnea state due to sleep apnea syndrome or the like, forced breathing occurs, and therefore, the “respiration rate of epoch n” of the condition I in step S1105 increases, and this abnormality The condition I is determined based on the value, and the section to be determined as the shallow sleep section is determined as the REM (provisional) section. Therefore, in step S1113, a stable section during sleep (that is, a deep sleep state or a shallow sleep state) is determined based on the determination of the condition K that “the number of stable sections in all bed sections / (the number of all bed sections−wakening sections) ≧ k”. ) Is determined whether or not a predetermined ratio k or more is present, thereby determining whether or not at least generally normal sleep is maintained. If the condition K is satisfied, the sleep is normal and the determination of the condition I is determined to be appropriate, and the process proceeds to YES. In step S1115, the epochs of the number of continuations j = 1 to j are defined as REM sleep intervals. decide. If the condition K is not satisfied, it is determined that there is an abnormal sleep state, the process proceeds to NO, and the determination based on the condition L is performed in step S1114.

Condition L is the determination of “(maximum respiratory rate in each segment−minimum respiratory rate in each segment) / REM sleep (provisional) segment number ≧ Lx” in the REM sleep (provisional) interval from the number of continuous times j = 1 to j. Therefore, even if there is a variation in the respiration rate, it is determined whether or not the respiration rate is within a normal range, and Lx is a minimum value that defines that respiration is abnormal. That is, it is assumed that an apnea state appears in any of the REM sleep (provisional) sections from the number of continuous times j = 1 to j. Therefore, if the condition L is satisfied, that is, if there is an abnormality in breathing, the process proceeds to YES, and in step S1106, each epoch from the continuation number j = 1 to j as the REM sleep (provisional) section is set as the shallow sleep section. Determine as. Further, if the condition L is not satisfied, that is, if the breathing is normal, the process proceeds to NO, and in step S1115, the epochs of the number of continuations j = 1 to j as the REM sleep (provisional) section are set as the REM sleep. It is determined as an interval.

When the REM sleep interval and the light sleep interval are determined, they are stored in the storage unit 109 in association with each epoch n in step S1107. In step S1108, the continuation number j is once reset to 0, and in step S1109, all epoch nmax 37, it is determined whether or not the above determination has been made. If all epochs have not been determined, the process proceeds to NO. The steps from step S1102 are repeated, and if all epochs have been determined, the process proceeds to YES, and the flowchart of FIG. Return to the next determination. Based on the processing result of the REM / shallow sleep determination unit 116, the sleep cycle calculation unit and the differential sleep cycle score calculation unit may calculate a sleep cycle (minutes) and a differential sleep cycle score as the sleep determination data. It becomes possible.

The process of the midway awakening determination unit 117 will be described using the flowchart of FIG.
Even in the sleep state, if the body movement continues for a certain time or more, it can be understood that the user has awakened in the middle, and the following determination is made.

In order to make a determination for each epoch in the same manner as described above, in step S1121, epoch n = 0 is initially set, and in step S1122, the respiration data of the corresponding epoch n is read as n = n + 1. In step S1123, the read epoch n is n ≠ nmax, and the coarse body motion, the fine body motion, and the non-body motion determined by the body motion determination unit 112 described in detail in FIG. It is determined whether it is either a section or a thin body movement section (hereinafter referred to as a body movement section).

If epoch n is n ≠ nmax and is a body motion section, the process proceeds to YES, and in step S1124, the number of continuations is counted as m = m + 1. If epoch n is n = nmax or is a body movement section, the process proceeds to NO, and in step S1125, the number of continuations m is m ≧ mx (where mx is the possibility of mid-wakefulness) And if m ≧ mx, the process proceeds to YES, and in step S1126, each epoch from m = 1 to m = m is in an awake state. Even if it is determined that each epoch is stored as a deep sleep interval, a shallow sleep interval, and a REM sleep interval, each epoch is replaced with an awakening interval and stored in the storage unit 109. In step S1127, The number of continuations m is once reset to zero.

If the number of continuations m does not exceed mx, the process proceeds to NO, and in step S1127, the number of continuations m = 0. In step S1128, it is determined whether or not the above determination has been made for all epochs nmax. If all epochs have not been determined, the process proceeds to NO. The steps from step S1102 are repeated again, and if all epochs have been determined, YES is determined. 50, in the determination of midway awakening condition described in detail in FIG. 50, the midway awakening is determined in detail according to each condition defined based on the tendency at the time of midway awakening found by the inventors. After this determination is made, the process returns to the flowchart of FIG. 37 and proceeds to the next determination. Based on the processing result of the midway awakening determination unit 117, the medium / long-time awakening frequency calculation unit and the short-time awakening frequency calculation unit calculate the middle / long-time awakening frequency (times) and the short-time awakening frequency as the sleep determination data ( Times) can be calculated.

Here, mid-wake condition determination will be described using the flowchart of FIG. In step S1131, the awakening condition determination is initially set to epoch n = 0, and in step S1132, the respiratory data of the corresponding epoch n is read as n = n + 1.

In step S1133, first, the state of body movement for each epoch n is obtained. That is, in the description of the body motion determination unit 112 described above, the coarse body motion, the fine body motion, and the non-body motion state stored in association with one breath i in step S1039 in the flowchart of FIG. In the same way, U = 2 is set for the moving state, U = 1 is set for the thin body moving state, U = 0 is set for the non-body moving state, and the body is set according to each breath i in the read epoch n. The movement state is obtained as the sum of U (hereinafter referred to as ΣU).

Next, it is determined whether or not ΣU ≧ 2 at the epoch n. If ΣU ≧ 2, the process proceeds to YES, and in step S1134, the number of continuations m = m + 1 is counted. If ΣU ≧ 2, the process proceeds to NO, and the number of continuations counted up to the previous time is m ≧ mp in step S1135 without counting the number of continuations (where mp includes the possibility of mid-wakefulness) It is a constant indicating the number of continuation of body motion sections, and it is determined whether or not mp <mx. If not m ≧ mp, it is determined that there is no possibility of mid-wakening, and the process proceeds to NO. In step S1140, the number of continuations is returned to m = 0. If m ≧ mp, it can be said that there is a possibility that each epoch n from the number of continuations m = 1 to m = m is in the awake state, so the process proceeds to YES and the next condition determination is performed.

That is, in step S1136, “(number of epochs n of ΣU ≧ 10) ≧ m1%” among the epochs from the continuation number m = 1 to m = m (where m1 is a ratio to the total continuation number m). It is determined whether or not it is a constant. If this condition is met, the process proceeds to YES, and in step S1139, each epoch n from the number of continuations m = 1 to m = m is replaced as an awakening section and stored in the storage unit 109. If the condition is not met, the process proceeds to NO and the next condition determination is performed.

That is, in step S1137, “(number of epochs n of ΣU ≧ 10) ≧ m2%” (where m2 is a constant indicating the ratio to the total number of continuations m, and m2 <m1). It is determined whether or not to do. If this condition is not met, the process proceeds to NO assuming that there is no possibility of awakening during m = 1 to m = m, and the number of continuations is returned to m = 0 in step S1140. If the above condition is met, it is determined that there is a possibility of awakening midway, the process proceeds to YES, and further conditions are added.

That is, whether or not “(average respiratory rate of all epochs from m = 1 to m = m) ≧ (average respiratory rate of all epochs from n = 1 to n = nmax) × mq” in step S1138. Determine whether. Here, mq is a constant of mq> 1, and since the respiratory rate in the awake state is generally higher than the respiratory rate in the sleep state, from n = 1 including the sleep state. If the average respiratory rate of epochs from m = 1 to m = m is larger than mq times the average respiratory rate of epochs up to n = nmax, it can be said that it is clearly determined that the patient is in an awake state.

If the above conditions are not satisfied, it is determined that there is no possibility of awakening during m = 1 to m = m, and the process proceeds to NO, and the number of continuations is returned to m = 0 in step S1140. If the condition is satisfied, the process proceeds to YES. In step S1139, each epoch n from the number of continuations m = 1 to m = m is replaced as an awakening interval and stored in the storage unit 109. In S1140, the number of continuations is returned to m = 0. In step S1141, it is determined whether or not the above determination has been made for all epochs nmax. If all epochs have not been determined, the process proceeds to NO, and the steps from step S1132 are repeated again. Proceed and return to the flowchart of FIG.

Processing of the wakeup determination unit 118 will be described with reference to the flowchart of FIG.
In step S1151, epoch n = nmax is set. In step S1152, n = n−1 is set back in time, and the corresponding epoch n is read. In step S1153, it is determined whether or not the read epoch n is determined to be a sleep state, that is, corresponds to any one of a deep sleep section, a shallow sleep section, and a REM sleep section (hereinafter referred to as a sleep section). Determine whether or not. If it does not correspond to the sleep section, the process proceeds to NO, and the detection of the sleep section is repeated with n = n−1 again in step S1152. If the epoch n is a sleep section, the process proceeds to YES, and this epoch n is defined as a wake-up (temporary) section in step S1154. In the following step S1155, it is determined whether or not there is a wake-up section in each epoch that goes back from the wake-up (temporary) section to a certain number of sections R. Here, in the normal sleep of a person, it can be considered that awakening does not occur a certain time before waking up. Therefore, the certain number of intervals R defines the certain time. If the awakening interval exists, the process proceeds to YES, and in step S1158, each epoch from the detected awakening interval to the wake-up (temporary) interval is defined as an awakening interval. In step S1154, the detected awakening is detected. The epoch immediately before the section is newly redefined as a wake-up (temporary) section, and the fixed section number R is set again in step S1155. In step S1155, if there is no awakening section up to a certain number of sections R, the process proceeds to NO. In step S1156, the wake-up (temporary) section is determined as the wake-up section. In step S1157, The information is stored in the storage unit 109 in association with the corresponding epoch n, and the process returns to the flowchart showing the main operation in FIG. Based on the processing result of the wake-up determination unit 118 described above, the deep sleep rate calculation unit calculates “sleep time” necessary for calculating the deep sleep rate (%), that is, “time from falling asleep to final awakening”. It becomes possible to do.

Subsequently, the CPU 106 proceeds to step S1006 in FIG. 36, and executes a sleep score calculation process. In the sleep score calculation process, a sleep score (sleep index) that comprehensively indicates the level of sleep quality is calculated by calculation. In one sleep from when the subject takes a sleeping posture to wake up, the above-described bed / bed determination, body movement determination, arousal determination, sleep determination, deep sleep determination, REM / light sleep determination, mid-wake determination, Sleep determination data (e.g. deep sleep time, number of mid-wakefulness) indicating sleep state can be obtained by wakeup determination. Although these sleep determination data can evaluate the quality of sleep to some extent by itself, the evaluation by itself only evaluates a part of the sleep state. Therefore, in the present embodiment, a plurality of predetermined items (variables) reflecting “depth”, “cycle”, “time”, and “halfway awakening” of sleep are extracted from PSG measurement data, and existing sleep is extracted. As variables correlated with the evaluation device, deep sleep rate (%), differential sleep cycle score, total bedtime (minutes), sleep cycle (minutes), deep sleep appearance amount (minutes), differential total bedtime score, The number of mid- and long-term awakenings (times), short-time awakenings (times), and sleep efficiency (%) were selected. Furthermore, in order to derive a sleep score that aggregates these predetermined items (variables), a principal component analysis is performed to develop a sleep evaluation score, and this sleep evaluation score and sleep apnea syndrome risk are reflected. A regression formula for sleep scores was developed. In addition, the object of the measurement data of PSG analyzed when creating the regression equation in the present embodiment was 49 healthy persons and 112 SAS patients.
As described above, in this embodiment, a sleep evaluation score is selected by a principal component analysis method in order to derive a sleep score that is an evaluation index for comprehensively evaluating the quality of sleep.

Second, a correlation coefficient between the plurality of predetermined items is calculated to obtain a correlation matrix. The correlation matrix based on the nine items is given by the determinant shown in the following equation (27). However, r11 to r99 are correlation coefficients.

Third, first to ninth principal components Z1 to Z9 and eigenvectors a11 to a99 are calculated based on the correlation matrix. These are given by the following equations (28) to (36).
Z1 = a11X1 + a12X2 + a13X3 + a14X4 + a15X5 + a16X6 + a17X7 + a18X8 + a19X9 (28)
Z2 = a21X1 + a22X2 + a23X3 + a24X4 + a25X5 + a26X6 + a27X7 + a28X8 + a29X9 (29)
Z3 = a31X1 + a32X2 + a33X3 + a34X4 + a35X5 + a36X6 + a37X7 + a38X8 + a39X9 (30)
Z4 = a41X1 + a42X2 + a43X3 + a44X4 + a45X5 + a46X6 + a47X7 + a48X8 + a49X9 (31)
Z5 = a51X1 + a52X2 + a53X3 + a54X4 + a55X5 + a56X6 + a57X7 + a58X8 + a59X9 (32)
Z6 = a61X1 + a62X2 + a63X3 + a64X4 + a65X5 + a66X6 + a67X7 + a68X8 + a69X9 (33)
Z7 = a71X1 + a72X2 + a73X3 + a74X4 + a75X5 + a76X6 + a77X7 + a78X8 + a79X9 Formula (34)
Z8 = a81X1 + a82X2 + a83X3 + a84X4 + a85X5 + a86X6 + a87X7 + a88X8 + a89X9 (35)
Z9 = a91X1 + a92X2 + a93X3 + a94X4 + a95X5 + a96X6 + a97X7 + a98X8 + a99X9 (36)
However, X1 to X9 are the nine items described above. Eigenvectors a11 to a99 are determined so that the first to ninth principal components Z1 to Z9 are orthogonal to each other. Orthogonal means that they are independent of each other and there is no correlation between them.

Eighth, the factor loading is calculated by multiplying the eigenvector by the square root of the eigenvalue, and those below a predetermined reference value (for example, 0.5) are deleted. Needless to say, it may be used without being deleted.
As described above, four sleep evaluation scores are selected through the first to eighth steps, and expressions (46) to (49) described later are derived.

Here, the four sleep evaluation scores are obtained by multiply-adding each of the nine items X1 to X9 and the first coefficients a11 to a99 shown in the equations (28) to (36). Since the first coefficients a11 to a99 are eigenvectors, the four sleep evaluation scores are in a linearly independent relationship with each other. That is, four sleep evaluation scores having smaller correlation coefficients than any two of the nine items are generated. Therefore, the sleep evaluation score is obtained by aggregating a plurality (nine in the present embodiment) of predetermined items (variables) from the viewpoint of sleep, and each expresses a characteristic of sleep. Therefore, by evaluating sleep using a sleep evaluation score, it is possible to obtain a more accurate evaluation index as compared to a single predetermined item or a combination of these appropriately. In the present embodiment, the predetermined items include deep sleep rate (%), differential sleep cycle score, total bedtime (minutes), sleep cycle (minutes), deep sleep appearance amount (minutes), differential total bedtime score, Select the number of mid- and long-term awakenings (times), short-time awakenings (times), and sleep efficiency (%). Based on these, four sleep evaluation scores are selected by the principal component analysis method. A “sleep depth score”, a “sleep cycle score” as the second principal component, a “sleep time score” as the third principal component, and a “halfway awakening score” as the fourth principal component could be obtained. “Sleep depth score” is an item indicating deep sleep, “sleep cycle score” is an item indicating sleep cycle, “sleep time score” is an item indicating sleep time, and “midway awakening score” is an item indicating midway awakening.

FIG. 52 is a flowchart showing the flow of each calculation in the sleep score calculation process. FIGS. 53 to 59 show the calculation processes in step S1165, step S1167, step S1168, step S1169, step S1170, step S1171, and step S1172, respectively. It is a flowchart which shows the detailed flow of. FIG. 61 is a time chart for explaining predetermined items calculated in the calculation process. In the following description, it is assumed that the CPU 106 executes these processes according to a predetermined program.

This sleep score calculation process is executed after the subject has completely awakened (that is, woken up). The state in which the subject is completely awakened may be determined to be a complete awake state when a state in which the subject's breathing is not detected continues for a predetermined period of time, or a measurement start / end button ( (Not shown) may be provided, and when the end button is pressed down, it may be determined that the state is completely awake. The storage unit 109 stores the state in each epoch from the start of measurement (time t0 in FIG. 61) to the end (time te) (that is, the result of the sleep stage determination process in step S1005). And is used for sleep score calculation processing.

52, in the sleep score calculation process, first, a deep sleep rate calculation process (step S1161) is executed. Deep sleep rate (%) means the ratio of deep sleep in sleep time, and “(deep sleep time / sleep time) × 100”, that is, “(deep sleep time / (time from falling asleep to final awakening)” ) × 100 ”. The “sleep time” necessary for calculating the deep sleep rate (%), that is, “time from sleep onset to final awakening” is determined by the sleep determination unit 114 as the sleep interval and the associated epoch is read to It is calculated by incrementing to an epoch that is determined to be an awakening section by the determination unit 113 and associated therewith. Further, as shown in FIG. 61, “deep sleep time” is the total number of epochs of deep sleep time from the start to the end of the measurement (time t0 to te). What is necessary is just to calculate similarly to quantity DT (min).

Subsequently, the CPU 106 executes a differential sleep cycle score calculation process in step S1162 of FIG. The differential sleep cycle score is a score representing how much the sleep cycle (minute) is different from the reference time (for example, 90 minutes). “− | Sleep cycle−predetermined reference time |” (|| represents an absolute value). Since the sleep cycle is an average value when one cycle is from the end of the REM sleep to the end of the next REM sleep, the REM / shallow sleep determination unit 116 determines the REM sleep interval and associates it with the epoch The average value may be calculated based on this. The reference time may be 90 minutes, for example, but is not particularly limited.

Subsequently, the CPU 106 executes a total bedtime calculation process in step S1163 of FIG. Total bedtime (minutes) means the time from bed to bed. What is necessary is just to calculate as a sum total of the epochs which are determined to be in the floor-entry state by the floor-in / bed-out determination unit 111 and are associated.

Subsequently, the CPU 106 executes a sleep cycle calculation process in step S1164 of FIG. As for the sleep cycle, the average value may be calculated based on the epoch that is determined as the REM sleep section by the REM / shallow sleep determination unit 116 and is associated with the difference sleep cycle score calculation process.

Subsequently, the CPU 106 proceeds to the deep sleep appearance amount calculation process in step S1165 of FIG. 52, and executes the deep sleep appearance amount calculation process shown in FIG. The deep sleep appearance amount DT (minutes) means the sum of deep sleep times, and more specifically, as shown in FIG. 61, from the start to the end of the measurement (time t0 to te). ) Is the total number of epochs of deep sleep appearance amount. Therefore, as shown in FIG. 61, when the deep sleep appearance amounts DT1 and DT2 are measured during the period from time t0 to te, the deep sleep appearance amount DT = DT1 + DT2.

As shown in FIG. 53, in the deep sleep appearance amount calculation process, first, the process proceeds to the epoch next to the first epoch (step S1221). Next, in step S1222, it is determined whether or not the epoch is the final epoch Ee. If this determination condition is negative, it is determined in step S1223 whether or not the epoch is in a deep sleep state. If this determination condition is affirmed, the deep sleep appearance amount DT is incremented in step S1224 (however, the initial value DT of DT = 0), and the process returns to step S1221. On the other hand, if the determination condition in step S1223 is negative, the deep sleep appearance amount DT is not incremented, and the process returns to step S1221. The processing from step S1221 to S1224 or the processing from step S1221 to S1223 is repeated until the determination condition of step S1222 is affirmed. If the determination condition in step S1222 is affirmed, the process returns to the flowchart of FIG.

Subsequently, the CPU 106 executes a difference total bedtime score calculation process in step S1166 of FIG. The difference total bedtime score is a score indicating how much the total bedtime (minutes) is different from the reference time (for example, 6.5 hours (390 minutes)). Floor time−reference time | ”(|| represents an absolute value). The total bedtime (minutes) may be calculated as the total number of epochs that are determined to be in the bedded state by the bed-and-bed determination unit 111 and associated with each other, as in the above-mentioned total bedtime calculation process.

Subsequently, the CPU 106 proceeds to the medium / long-time awakening number calculation process in step S1167 of FIG. 52, and executes the medium / long-time awakening number calculation process shown in FIG. The number of awakening times (times) means the number of times of awakening over a reference time (for example, 2 minutes 30 seconds) that appears during sleep. As shown in FIG. 54, in the middle / long-term awakening count calculation process, first, the process proceeds to the epoch next to the first epoch (step S1191). Next, in step S1192, it is determined whether or not the epoch is the final epoch Ee. If this determination condition is negative, it is determined in step S1193 whether or not the awakening continues for T minutes or more. As described above, since 1 epoch = 30 seconds in this embodiment, when T = 2.5, if 5 or more awakening epochs continue, it is considered that the state is awakening. Therefore, the CPU 106 affirms the determination in step S 1193 only when a predetermined number (for example, five or more) of epochs in the awake state continues. Subsequently, the routine proceeds to step S1194, and increments the number of wake-up times (WN; where initial value WN = 0). Subsequently, in step S1195, the epoch is advanced by the number of continued awakenings, and the process returns to step S1191. On the other hand, if the determination condition in step S1193 is negative, the process returns to step S1191. The processing of steps S1191 to S1195 or steps S1191 to S1193 is repeated until it is determined in step S1192 that the epoch to be determined is the final epoch Ee. If the determination condition in step S1192 is affirmed, the medium / long-term awakening count calculation process ends, and the routine returns to the flowchart of FIG.

Subsequently, the CPU 106 proceeds to the short-time awakening number calculation process in step S1168 of FIG. 52, and executes the short-time awakening number calculation process shown in FIG. The number of short-time awakenings (times) means the number of times of awakening within a reference time (for example, 2 minutes) that appears during sleep. As shown in FIG. 55, in the short-time awakening count calculation process, first, the process proceeds to the epoch next to the first epoch (step S1231). Next, in step S1232, it is determined whether or not the epoch is the final epoch Ee. If this determination condition is negative, it is determined in step S1233 whether or not the awakening continues for less than T minutes. As described above, since 1 epoch = 30 seconds in the present embodiment, when T = 2.5, if there are less than 5 continuous awakening epochs, it is regarded as a short-time awakening. Therefore, CPU 106 affirms the determination in step S1233 only when the number of consecutive awakening epochs is less than a predetermined number (for example, 5). Subsequently, the routine proceeds to step S1234, and increments the number of short-time awakenings (MN; provided that the initial value MN = 0). Subsequently, in step S1235, the epoch is advanced by the number of continued awakenings, and the process returns to step S1231. On the other hand, if the determination in step S1233 is negative, the process returns to step S1231. The processing in steps S1231 to S1235 or steps S1231 to S1233 is repeated until it is determined in step S1232 that the epoch to be determined is the final epoch Ee. If the determination in step S1232 is affirmative, the short-time awakening count calculation process ends, and the routine returns to the flowchart of FIG.

Subsequently, the CPU 106 proceeds to the sleep efficiency calculation process in step S1169 of FIG. 52, and executes the sleep efficiency calculation process shown in FIG. The sleep efficiency (%) means the ratio of the actual sleep time to the total bedtime, which is “(total sleep time / total bedtime) × 100”, that is, “((total bedtime−sleeping time) (Total sum of hours awakened) / total bedtime) × 100 ”. That is, the sleep efficiency SE is the total number of epochs from the start of measurement (time t0 in FIG. 61) to the end (time te), IA, and the epoch determined to be in an awake state in determination step S1183 described later. This is a value obtained by (IA-I) / IA where I is the number of wakefulness (number of awakening epochs). Therefore, in the sleep efficiency calculation process, since the first epoch is in the awake state, the initial value of the number of awakening epochs (I) is set to 1, and it is determined in step S1183 that the epoch to be determined is awake. It is incremented every time (step S1184).

As shown in FIG. 56, in step S1181, the CPU 106 first proceeds to the next epoch. Subsequently, in step S1182, it is determined whether or not the epoch is the epoch immediately before transitioning to the complete awake state (final epoch Ee in FIG. 61). If this determination is negative, it is determined whether or not the epoch is in an awake state (step S1183). If the determination result is affirmative, the process proceeds to step S1184, the value of the awakening epoch number (I) is incremented, the process returns to step S1181, and the process proceeds to the next epoch. On the other hand, if the determination condition in step S1183 is negative, the process returns to step S1181 without incrementing the value of the number of awakening epochs (I). CPU 106 repeats the processing of steps S1181 to S1183 or steps S1181 to S1184 unless the determination result of step S1182 becomes affirmative.

On the other hand, if the determination in step S1182 is affirmative, the routine proceeds to step S1185 and calculates sleep efficiency SE. That is, the final value of the awakening epoch number I and the numerical value of the total epoch number IA are substituted into the above formula SE = (IA−I) / IA to obtain the sleep efficiency SE, and the sleep efficiency calculation process ends. Returning to the flowchart of FIG.

Subsequently, the CPU 106 executes each data standardization process in step S1170 of FIG. As shown in FIG. 57, in each data standardization process, the standardization process of the value of the sleep determination data acquired in steps S1161 to S1169 described above is executed. First, in step S1241, the standard value Za (st) of the deep sleep rate Za is obtained by Za (st) = (Za−average Za) / standard deviation Za. Here, each average Za and standard deviation Za are each average value and standard deviation value (fixed values) of deep sleep rate Za in the population based on the measurement data of PSG. The population is, for example, a group of X people in their 20s when the subject's age is in their 20s. The test subject inputs his / her parameters (for example, age and sex) in advance using the operation unit 105, so that data regarding an appropriate population is selected and used for the standardization process. Data regarding this population is stored in the storage unit 109 in advance. CPU 106 reads out the average value and the standard deviation from storage unit 109 and executes the calculation of step S1241. The same applies to the processing in steps S1242 to S1249.

Similarly, in each step S1242 to S1249, each sleep determination data is standardized. The standard value of each sleep determination data is obtained by the following formulas (38) to (45) (steps S1242 to S1249).
Difference sleep cycle score Zb (st) = (Zb−average Zb) / standard deviation Zb (38)
Total bedtime Zc (st) = (Zc−average Zc) / standard deviation Zc (39)
Sleep cycle Zd (st) = (Zd−average Zd) / standard deviation Zd Equation (40)
Deep sleep appearance amount Ze (st) = (Ze−average Ze) / standard deviation Ze (formula 41)
Difference total bedtime score Zf (st) = (Zf−average Zf) / standard deviation Zf Equation (42)
Number of middle and long awakenings Zg (st) = (Zg−average Zg) / standard deviation Zg (43)
Number of short-time awakenings Zh (st) = (Zh−average Zh) / standard deviation Zh (formula 44)
Sleep efficiency Zi (st) = (Zi−average Zi) / standard deviation Zi (formula 45)
By this standardization process, the deep sleep rate Za, the differential sleep cycle score Zb, the total bedtime Zc, the sleep cycle Zd, the deep sleep appearance amount Ze, the differential total bedtime score Zf, and the number of mid- and long-term awakenings Zg of different scales. It becomes possible to handle the number of short-time awakenings Zh and sleep efficiency Zi by the same process. When the process of step S1249 ends, the routine returns to the flowchart of FIG.

Subsequently, the CPU 106 executes each principal component score calculation process in step S1171 in FIG. As shown in FIG. 58, in each principal component score calculation process, the standard value of each sleep determination data acquired in step S1170 described above is used for the calculation. In the principal component score calculation process, a plurality of types of predetermined items including at least items related to sleep depth, items related to sleep rhythm, and items related to mid-wake awakening extracted based on PSG measurement data Sleep evaluation by multiplying the principal component coefficient for each predetermined item of the sleep evaluation score obtained by performing the principal component analysis on the sleep determination data corresponding to the predetermined item calculated from the biological signal of the subject Calculate the score. More specifically, the nine sleep determination data are deep sleep rate Za, differential sleep cycle score Zb, total bedtime Zc, sleep cycle Zd, deep sleep appearance amount Ze, differential total bedtime score Zf, medium From the long-time awakening count Zg, the short-time awakening count Zh, and the sleep efficiency Zi, four sleep evaluation scores, that is, a “sleep depth score”, a “sleep cycle score”, a “sleep time score”, and a “halfway awake score” calculate. The CPU 106 calculates “sleep depth score”, “sleep cycle score”, “sleep time score”, and “halfway awakening score” according to the following equations (46) to (49) (steps S1251 to S1254).

First principal component score (sleep depth score)
= Coefficient C1a * Standard value Za (st) + Coefficient C1b * Standard value Zb (st) + Coefficient C1c * Standard value Zc (st) + Coefficient C1d * Standard value Zd (st) + Coefficient C1e * Standard value Ze (st) + Coefficient C1f * Standard value Zf (st) + Coefficient C1g * Standard value Zg (st) + Coefficient C1h * Standard value Zh (st) + Coefficient C1i * Standard value Zi (st) Equation (46)

Second principal component score (sleep cycle score)
= Coefficient C2a * Standard value Za (st) + Coefficient C2b * Standard value Zb (st) + Coefficient C2c * Standard value Zc (st) + Coefficient C2d * Standard value Zd (st) + Coefficient C2e * Standard value Ze (st) + Coefficient C2f * Standard value Zf (st) + Coefficient C2g * Standard value Zg (st) + Coefficient C2h * Standard value Zh (st) + Coefficient C2i * Standard value Zi (st) Equation (47)

Third principal component score (sleep time score)
= Coefficient C3a * Standard value Za (st) + Coefficient C3b * Standard value Zb (st) + Coefficient C3c * Standard value Zc (st) + Coefficient C3d * Standard value Zd (st) + Coefficient C3e * Standard value Ze (st) + Coefficient C3f * Standard value Zf (st) + Coefficient C3g * Standard value Zg (st) + Coefficient C3h * Standard value Zh (st) + Coefficient C3i * Standard value Zi (st) Equation (48)

Fourth principal component score (midway awakening score)
= Coefficient C4a * Standard value Za (st) + Coefficient C4b * Standard value Zb (st) + Coefficient C4c * Standard value Zc (st) + Coefficient C4d * Standard value Zd (st) + Coefficient C4e * Standard value Ze (st) + Coefficient C4f * Standard value Zf (st) + Coefficient C4g * Standard value Zg (st) + Coefficient C4h * Standard value Zh (st) + Coefficient C4i * Standard value Zi (st) Equation (49)

Here, each coefficient in the equations (46) to (49) is a constant obtained from a principal component analysis method based on nine predetermined items, and is stored in the storage unit 109. The CPU 106 reads out each coefficient from the storage unit 109 and executes arithmetic processing. A principal component score coefficient matrix of nine predetermined items and a principal component score is as shown in Table 24.

When the process of step S1254 ends, the routine returns to the flowchart of FIG.

Subsequently, the CPU 106 executes sleep disorder determination probability calculation processing in step S1172 of FIG. Here, the sleep disorder discrimination probability refers to the probability of being a sleep disorder person such as a SAS patient. As shown in FIG. 59, in the sleep disorder determination probability calculation process, the sleep disorder determination probability obtained by performing logistic regression analysis on the sleep evaluation score is calculated.
Logistic regression analysis (multiple regression analysis) of multiple sleep evaluation scores (explanatory variables) and sleep disorder dummy variables (objective variables) allows the probability of falling into sleep disorder from multiple sleep evaluation scores (sleep disorder) Discrimination probability).
In the logistic regression analysis in this embodiment, the objective variable is represented by a dummy variable of 0/1 (the presence or absence of SAS: SAS group 1 / non-SAS group 0). The explanatory variables are a first principal component score (sleep depth score), a second principal component score (sleep cycle score), a third principal component score (sleep time score), which are sleep evaluation scores calculated from sleep determination data, At least any three of the fourth principal component scores (halfway awakening scores) are used. This makes it possible to create a regression equation of sleep disorder discrimination probability. Here, an example using four sleep evaluation scores will be described with reference to FIG.
The CPU 106 calculates the first principal component score (sleep depth score), the second principal component score (sleep cycle score), the third principal component score (sleep time score), and the fourth principal component score (intermediate awakening) calculated in step S1171. Using the score, the variable P is calculated by the following equation (50) (step S1261). At this time, a process of inverting the signs of the third principal component score (sleep time score) and the fourth principal component score (halfway awakening score) may be appropriately performed.
Variable P
= Fixed value F1 * 1st principal component score + Fixed value F2 * 2nd principal component score + Fixed value F3 * 3rd principal component score + Fixed value F4 * 4th principal component score ... Formula (50)
Here, the fixed values F1 to F4 are fixed values obtained from a principal component analysis method of a certain population. These fixed values are stored in the storage unit 109 and read out by the CPU 106 and used for calculation.

Next, the CPU 106 calculates the sleep disorder determination probability by the following equation (51) using the variable P calculated in step S1261 (step S1262).
Sleep disorder discrimination probability = 1 / (1+ (exp- (P))) (51)
When the process of step S1262 ends, the routine returns to the flowchart of FIG.

Thereafter, the CPU 106 executes the sleep score calculation process of step S1173 in FIG. As shown in FIG. 60, in the sleep score calculation process, the CPU 106 calculates the sleep score (sleep index) by the following formula (52) using the sleep disorder determination probability calculated in step S1262 (step S1263). ).
Number of sleep points = 100-(Sleep disorder discrimination probability * 100) ... Formula (52)
CPU6 memorize | stores the sleep score obtained by step S1263 in the memory | storage part 9 (step S1174, refer FIG. 52).
Thus, from equation (52), when the sleep disorder determination probability is high, the sleep score is calculated low, and when the sleep disorder determination probability is low, the sleep score is calculated high. This makes it possible to predict whether or not SAS will develop before SAS develops.

Thus, the regression equations for obtaining the number of sleep points in the present invention are summarized in the following three equations described above.
Variable P = fixed value F1 * first principal component score + fixed value F2 * second principal component score + fixed value F3 * third principal component score + fixed value F4 * fourth principal component score Equation (50)
Discrimination probability = 1 / (1+ (exp- (P))) (51)
Number of sleep points = 100-(Sleep disorder discrimination probability * 100) ... Formula (52)

Next, sleep type determination processing will be described.
The sleep type determination process executed by the sleep evaluation apparatus 101 according to the present invention determines which sleep type the subject's sleep corresponds to among a plurality of preset sleep types. is there. The sleep type is a type classified according to the feature of sleep content based on the level of each value of the sleep evaluation score. The sleep type is a sleep evaluation score calculated as described above, a sleep depth score (first score), a sleep cycle score (second score), a sleep time score (third score), and a midway awakening score (fourth score). In addition, it is preferable to set and store in advance based on the degree of the body movement frequency score (fifth score). The body motion frequency score is a score related to the frequency of body motion that occurs during sleep.

As an example, there are five types of sleep types as shown in FIGS. 62 to 66 in consideration of the values of the sleep depth score, the sleep cycle score, the sleep time score, the midway awakening score, and the body motion frequency score. Set sleep type. 62 to 66 are radar charts showing examples of the first sleep type to the fifth sleep type. The first sleep type shown in FIG. 62 is a sleep type in which sleep time is standard, but there are many body movements and no deep sleep / rhythm. The second sleep type shown in FIG. 63 is a sleep type in which the sleep time is short, there are many midway awakenings, and the sleep rhythm is bad. The third sleep type shown in FIG. 64 is a sleep type in which sleep time and rhythm are almost standard, but there are many body movements and awakening during the middle, and little deep sleep. The fourth sleep type shown in FIG. 65 is a sleep type in which all scores are standard. The fifth sleep type shown in FIG. 66 is a sleep type in which there is little sleep time / deep sleep, and there is a lot of awakening / body movement. In the present embodiment, five types of sleep are described as a plurality of preset sleep types, but it goes without saying that the content and type of the sleep type may be changed as appropriate.

First, as shown in FIG. 67, the body motion frequency score is calculated. FIG. 67 is a flowchart showing the body motion frequency score calculation. This process will be described assuming that the CPU 106 as the body motion frequency score calculation unit executes according to a predetermined program. As shown in FIG. 67, the CPU 106 calculates a body movement frequency score by the following equation (53) and stores it in the storage unit 109 (step S1264).
Body motion frequency score = number of body motions / total bedtime (53)
Here, as described above, as the total bedtime, a value that has already been calculated by executing the total bedtime calculation process in step S1163 of FIG. 52 can be used.

The number of body movements may be obtained by detecting a part where the output from the sensor unit 102 is larger than a predetermined threshold due to body movement. Although the method of obtaining is not particularly limited, an example is as follows.
(1) Folding is performed with reference to the median value of the signal detected from the sensor unit 102 (full-wave rectification).
(2) A predetermined filter process is performed on the waveform of (1). For example, an infinite impulse response (IIR) filter is used, and a fourth-order band pass filter (BPF) of 0.01 Hz on the low frequency side and 0.1 Hz on the high frequency side is used in the forward and reverse directions. Filter on both.
(3) A predetermined threshold is set for the waveform of (2). For example, an average value and a standard deviation of all data of the waveform of (2) are obtained, and the sum of both is set as a threshold value.
(4) For the waveform of (2), the number of epochs where points exceeding the threshold of (3) exist is counted, and this is taken as the number of body movements.

Subsequently, the CPU 106 executes sleep type determination processing in step S1007 in FIG. In addition, this process demonstrates as what CPU106 as the sleep type determination part 160 performs according to a predetermined program. An example of the sleep type determination process will be described with reference to FIG. FIG. 68 is a flowchart showing sleep type determination.
As shown in FIG. 68, in the sleep type determination process, the sleep depth score (first principal component score) and sleep cycle score (second score), which are the four sleep evaluation scores acquired in steps S1251 to S1254 described above. The principal component score), the sleep time score (third principal component score), and the midway awakening score (fourth principal component score) are substituted into the following formulas (54) to (58), and the first judgment value to the first judgment value 5 Determination values are calculated (steps S1265 to S1270). The first determination value is a value for determining the suitability of the first sleep type, the second determination value is a value for determining the suitability of the second sleep type, and the third determination value is the third sleep type. The value for determining the appropriateness, the fourth determination value are the values for determining the appropriateness of the fourth sleep type, and the fifth determination value is the value for determining the appropriateness of the fifth sleep type.

First judgment value = coefficient V1a * first principal component score + coefficient V2a * second principal component score + coefficient V3a * third principal component score + coefficient V4a * fourth principal component score (54)

Second determination value = coefficient V1b * first principal component score + coefficient V2b * second principal component score + coefficient V3b * third principal component score + coefficient V4b * fourth principal component score Equation (55)

Third judgment value = coefficient V1c * first principal component score + coefficient V2c * second principal component score + coefficient V3c * third principal component score + coefficient V4c * fourth principal component score (56)

Fourth determination value = coefficient V1d * first principal component score + coefficient V2d * second principal component score + coefficient V3d * third principal component score + coefficient V4d * fourth principal component score Equation (57)

Fifth judgment value = coefficient V1e * first principal component score + coefficient V2e * second principal component score + coefficient V3e * third principal component score + coefficient V4e * fourth principal component score (58)

Here, each coefficient in the formulas (54) to (58) is a constant set in advance to determine which of the five types of sleep types is closest, and is stored in the storage unit 109. ing. The CPU 106 reads out each coefficient from the storage unit 109 and executes arithmetic processing. The coefficient matrix of the first to fifth determination values and the principal component score is as shown in Table 5.

After calculating the first determination value to the fifth determination value (steps S1265 to S1269), the determination value that is the largest value is selected, and the sleep type corresponding to the determination value is specified, whereby the sleep type of the subject is determined. Which of the first sleep type to the fifth sleep type is determined is determined (step S1270) and stored in the storage unit 9. When the process of step S1270 ends, the routine returns to the flowchart of FIG.

Next, the evaluation result display process (step 1008, see FIG. 36) will be described. 69 is a flowchart showing the contents of the evaluation result display process executed by the CPU 106, FIG. 70 is an example of a sleep evaluation screen, FIG. 71 is an example of a display screen of sleep type and graph display, 72 is an example of a sleep score transition screen.

First, the CPU 106 executes a sleep score comparison process (step S1271). In this sleep score comparison process, the sleep score Score obtained by the sleep score calculation process is compared with the first reference value W and the second reference value Y, and the sleep score Score is divided into three stages. Specifically, the sleep point average value + dispersion value of the sleep abnormal group included in a certain population is the first reference value W, and the sleep point average value + dispersion value of the healthy sleep group is the second reference value Y. When the sleep score Score is less than or equal to the first reference value W, the first division, and when the sleep score Score exceeds the first reference value W and is less than or equal to the second reference value Y, the second category and the sleep score Score When the reference value Y is exceeded, the sleep score Score is classified into the third category. Here, the first reference value W is the sleep point average value + dispersion value of the abnormal sleep group included in a certain population, and the second reference value Y is the sleep point average value + dispersion value of the healthy sleep group. is there. These fixed values W and Y are stored in the storage unit 109 and read out in the sleep score comparison process.

Next, the CPU 106 displays “bad sleep” on the display unit 104 when the sleep score Score is classified into the first category, and “normal” when the sleep score Score is classified as the second category. If the sleep score is classified into the third category, “Good sleep” is displayed (step S1272). For example, in the case of good sleep, the sleep evaluation screen shown in FIG. In this case, it is preferable that the CPU 106 displays the transition of one sleep stage together using the processing results of the step S1005 and step S1006 in addition to the processing result of the sleep score comparison processing. By displaying the category in addition to the sleep score, the user can know the quality of sleep quality. Furthermore, since the time course of the sleep stage such as awakening, shallowness, and deepness is displayed, it is possible to make use of it for self-condition management.

Next, the CPU 106 determines whether or not an operation for displaying the next screen has been performed (step S1273). If the operation has been performed, the sleep evaluation score calculated and stored in the storage unit 109 is as described above. That is, graphs of a sleep depth score (first score), a sleep cycle score (second score), a sleep time score (third score), an awakening score (fourth score), and a body movement frequency score (fifth score) The display and the corresponding sleep type are displayed on the display unit 104 (step S1274). The graph display is not particularly limited. As an example, as shown in FIG. 71, the sleep depth score (first score), the sleep cycle score (second score), and the sleep time score (third) are radiated from the center. A radar chart having axes of (score), midway awakening score (fourth score), and body movement frequency score (fifth score). Thus, the state of a some score can be visualized and a sleep score can be analyzed and self-evaluation can be easily performed. That is, it is possible to determine whether the number of sleep points is low (or high) due to various factors such as sleep depth, sleep cycle, sleep time, mid-wake awakening, body movement frequency, etc. It becomes easier to find improvements such as what will lead to an improvement in the number of sleep points. In addition, by displaying an evaluation of which sleep type the subject's sleep corresponds to among the representative sleep types that are set in advance, the subject himself / herself is what his / her sleep is. It will be very easy to understand. In addition, it is preferable to adopt and display an intuitively easy-to-understand name as the sleep type name because the subject can easily imagine his / her sleep state. For example, “Gira Gira” for sleep types with a high midway awakening score, “Grogro” for sleep types with a high body movement frequency score, “Good” or “Utoto” depending on the level of sleep depth score, Depending on the level of sleep time score Thus, “long time” or “short time” may be used, or a combination of these may be adopted as appropriate.

Next, the CPU 106 determines whether or not an operation for displaying the next screen has been performed (step S1276). If the operation has been performed, the CPU 106 determines the average value based on the sleep score Score stored in the storage unit 109. The variance value is calculated (step S1276), and a sleep score transition screen is displayed on the display unit 104 (step S1277). For example, as shown in FIG. 72, it is assumed that the vertical axis represents a sleep score Score and the horizontal axis represents a date. In this case, if the sleep score Score is equal to or less than the average value + dispersion value calculated in step S1276, the bar graph is colored and displayed. Thereby, the user can know the day when the quality of sleep was bad, and can use it for physical condition management.

Next, the CPU 106 determines whether or not an operation for displaying the next screen has been performed (step S1278), and when the operation has been performed, the processing ends.

According to the present invention, since the deep sleep rate (%) is selected as one of the predetermined items, the deep sleep rate is reflected in the regression equation for evaluating the quality of sleep, and the evaluation ability related to the sleep depth Is improved. FIG. 73 is a graph showing a comparison between the related art regarding the sleep score and the present invention, wherein (a) is a graph showing the correlation between the conventional sleep score and the deep sleep rate, and (b) is a sleep according to the present invention. It is a graph which shows the correlation with a score and a deep sleep rate. The sleep evaluation device (Sleep Scan SL-501) manufactured by the applicant was used as the sleep score according to the prior art. Further, the deep sleep rate (%) on the vertical axis uses the measurement results obtained by the sleep evaluation device of the applicant's product for both (a) and (b). It is clear that the correlation between the sleep score of the present invention and the deep sleep rate (FIG. 73 (b)) is better than the correlation between the conventional sleep score and the deep sleep rate (FIG. 73 (a)), Among items related to sleep quality, items related to sleep depth, items related to sleep rhythm, and items related to awakening during sleep, the ability to evaluate items related to sleep depth can be improved.

According to the present invention, the probability of being a sleep disorder person such as a SAS patient (sleep disorder discrimination probability) is calculated, and the sleep score is calculated by reflecting this, so the sleep disorder person and the healthy person can sleep. Differences in quality assessment results can be made. FIG. 74 is a graph showing a comparison between the related art relating to the sleep score and the present invention. As a prior art, a sleep evaluation device (Sleep Scan SL-501) manufactured by the applicant is used. According to the present invention, it can be seen that the difference between the sleep score of a SAS patient and the sleep score of a healthy person appears more markedly than in the case of the prior art.

[Modification 1 of the third embodiment]
In the third embodiment, the sleep depth score (first score), the sleep cycle score (second score), the sleep time score (third score), the midway awakening score (fourth score), and the body movement frequency score (first score) 5 scores) were determined, the sleep type was determined based on them, and a pentagonal radar chart and the corresponding sleep type were displayed as an example (FIG. 71). On the other hand, four scores of a sleep depth score (first score), a sleep cycle score (second score), a sleep time score (third score), and a midway awakening score (fourth score) are obtained and based on them. Then, the sleep type may be determined, and the rhombus radar chart and the corresponding sleep type may be displayed.

[Modification 2 of the third embodiment]
In addition to the sleep depth score (first score), sleep cycle score (second score), sleep time score (third score), mid-wake awakening score (fourth score), body movement frequency score (fifth score), Six scores including the sleep habit score may be obtained as the sixth score, and a hexagonal radar chart may be displayed. This will be described below.

The sleep habit score is a score calculated based on the bedtime and wake-up time of each day in the sleep record for a predetermined number of days. More specifically, the sleep habit score is obtained as an average value of a bedtime score and a wake-up time score determined as described below as an example. Based on FIG. 75 thru | or FIG. 83, a sleep habit score, a bedtime score, and a wake-up time score are demonstrated. 75, FIG. 78, FIG. 80, and FIG. 82 are diagrams showing examples of sleep diaries, FIG. 76, FIG. 79, FIG. 81, and FIG. It is a figure which shows the example of a sleep habit pattern table.

The sleep evaluation apparatus 101 according to the present invention aggregates data related to sleep measured for a subject as a sleep diary (sleep record) and stores it in the storage unit 109. As an example of the sleep diary, there is a sleep diary SD ₁ as shown in FIG. Referring to the row of No. 1 which is the record of the first day, this subject shows that he went to sleep from 23:00 to 24:00 and got up from 6 to 7 on the next day. Similarly, it can be seen that the bedtime and the wake-up time remain constant thereafter. Sleep evaluation device 101, along such sleep diary SD _1, to calculate the bedtime score and wake-up time score.

To calculate the bedtime score and wake-up time scores, in a manner shown in bedtime table ST ₁ and the rising time table WT 1 _(see FIG. 76) to reflect the result of sleep diary SD _1, it is observed during the predetermined number of days Find the most frequent bedtime and wake-up time. In this embodiment, an example is a last seven days as a predetermined number of days to determine the sleep habits score, therefore, as shown in FIG. 76, bedtime table ST ₁ and the rising time table WT _1, both, 24 hours 24 columns indicating 7 days and 7 rows indicating 7 days.

In the sleep diary SD ₁ , after the latest seven days are recorded, the sleep evaluation apparatus 101 calculates a bedtime time score and a wake-up time score along the record for the latest seven days. That is, after the items No. 1 to No. 7 of the sleep diary SD ₁ are recorded, the sleep evaluation apparatus 101 performs the bedtime table ST ₁ and the wake-up time according to the records of the items No. 1 to No. 7. performing an input to the table WT _1. Note that, on the next day, items up to item number 8 are recorded in the sleep diary SD ₁ , so the sleep evaluation apparatus 101 excludes the record of item number 1 that is the oldest, and item numbers 2 through item number 8 are excluded. The bedtime score and the wake-up time score are calculated along with the record.

According to item 1 sleep diary SD _1, so bedtime is a 24 o'clock starting at 23, as shown in FIG. 76, the sleep evaluation device 101, the item number 1 bedtime table ST _1, 23:00 “1” is input to the column 12 corresponding to the time zone of 24:00. Similarly, according to item No. 1 of the sleep diary SD ₁ , the wake-up time is between 6 o'clock and 7 o'clock, so that the sleep evaluation apparatus 101 has the item number 1 of the wake-up time table WT ₁ from 6 o'clock to 7 o'clock. Enter “1” in column 19 corresponding to the time zone. Similarly, the sleep evaluation apparatus 101 corresponds to the records from No. 2 to No. 7 of the sleep diary SD ₁ and from No. 2 to No. 7 of the bedtime table ST ₁ and the wake-up time table WT _1. Make input.

After the input to bedtime table ST ₁ and the rising time table WT ₁ of No. 1 to No. 7, sleep evaluation device 101, the total value SumST of each column, calculates a SumWT, also, their SumST, the SumWT The maximum value is detected from the inside. According to the example shown in FIG. 76, the maximum value of sumST is detected as “7” in column 12, and the maximum value of sumWT is detected as “7” in column 19. Furthermore, the sleep evaluation apparatus 101 also detects sumST and sumWT in the column immediately before and the column immediately after the column indicating the detected maximum value. As a result, the bedtime table ST ₁ in FIG. 76, "0-7-0" than

columns

11, 12 and 13 (sum of 3 values "7") SumST pattern that is detected, similarly, waking In the time table WT ₁ , a sumWT having a pattern of “0-7-0” (the total value of the three values is “7”) is detected from the

columns

18, 19, and 20. Here, by detecting the maximum values of sumST and sumWT, it is possible to grasp the bedtime and the wake-up time that were most common in a predetermined number of days (for example, 7 days). Also, the difference between the bedtime and the wake-up time within one hour is the basis of the sleep habit score by detecting sumST and sumWT in the column immediately before and after the column indicating the maximum value of sumST and sumWT. This is because it is an acceptable range in examining sleep habits. In the case of a subject who regularly goes to bed and wakes up every day, each maximum value of sumST and sumWT is a number corresponding to a predetermined number of days (“7” in FIG. 76). However, in the case of subjects sleeping irregularly and getting up, variation occurs, so that the maximum values of sumST and sumWT are less than the number corresponding to the predetermined number of days (“7” in FIG. 76). Become.

The sleep evaluation apparatus 101 detects the pattern including the maximum value of sumST, the sumST in the immediately preceding column, and the sumST in the immediately following column (“0-7-0” (according to the above example) ( The total value of the three values is “7”)) against the sleep habit pattern table stored in the storage unit 109 in advance, and the score corresponding to the matching pattern is determined as the “sleeping time score”. Similarly, the sleep evaluation apparatus 101 uses the pattern of the maximum value of sumWT, the sumWT in the immediately preceding column, and the sumWT in the immediately following column (“0-7-0” according to the above example) Is compared with the sleep habit pattern table and the score corresponding to the matching pattern is determined as the “wake-up time score”.

Here, the sleep habit pattern table will be described with reference to FIG. As shown in FIG. 77, the sleep habit pattern table is associated with all the patterns including the maximum value of sumST (sumWT), sumST (sumWT) in the immediately preceding column, and sumST (sumWT) in the immediately following column. Thus, the score that is the “wake-up time score” is set. Patterns located at the top of the table can be evaluated as regular sleeping habits with a constant bedtime (wake-up time), and are scored high for these. On the other hand, the patterns located in the lower part of the table can be evaluated as irregular sleeping habits with variations in bedtime (wake-up time), and the score is set low for these patterns.

The sleep evaluation apparatus 101 has a pattern consisting of the maximum sumST value, the sumST in the immediately preceding column, and the sumST in the immediately following column (“0-7-0” according to the above example (the total value of the three values is “7 ))) Is compared with the sleep habit pattern table and matches the pattern number 1, and the score “10” corresponding to this is determined as the “sleeping time score”. Similarly, the sleep evaluation apparatus 101 uses the pattern of the maximum value of sumWT, the sumWT in the immediately preceding column, and the sumWT in the immediately following column (“0-7-0” according to the above example) "7")) is compared with the sleep habit pattern table and matches the pattern number 1, so the score "10" corresponding to this is determined as the "wake-up time score".

The sleep evaluation apparatus 101 calculates the average value of the bedtime score (“10” in the above example) and the wake-up time score (“10” in the above example) determined as described above, and the average value (in the above example) Then, “10”) is determined as the “sleep habit score” and stored in the storage unit 109.

The example described above based on the sleep diary SD ₁ in FIG. 76 and the bedtime table ST ₁ and the wake-up time table WT ₁ in FIG. 76 is a case of a subject who practices the most regular sleep habits. Hereinafter, the example of the test subject of another sleep habit is demonstrated.

Sleep diary SD ₂ shown in FIG. 78 is generally going to bed and getting up at the Saturday and Sunday it is often a holiday is, shows an example of a slower subjects than going to bed and wake up in Monday through Friday. In such a sleep diary SD ₂ , after the recording up to item number 8 is made, the bedtime time score and the wake-up time score are calculated according to the records for the most recent 7 days (ie, item numbers 2 to 8). An example will be described.

According to item 2 of sleep diary SD _2, so bedtime is a 24 o'clock starting at 23, as shown in FIG. 79, the sleep evaluation device 101, the item number 1 bedtime table ST _2, 23:00 “1” is input to the column 12 corresponding to the time zone of 24:00. Similarly, according to item No. 2 of the sleep diary SD ₂ , the wake-up time is between 6 o'clock and 7 o'clock, so the sleep evaluation apparatus 101 can be used for the item No. 1 of the wake-up time table WT ₂ from 6 o'clock to 7 o'clock. Enter “1” in column 19 corresponding to the time zone. Similarly, sleep evaluation device 101, in correspondence with the recording from No. 3 sleep diary SD ₂ to No. 5, to No. 4 from No. 2 bedtime table ST ₂ and the rising time table WT ₂ Make input. According to No. 6 of sleep diary SD _2, so bedtime is 2 o'clock starting at 1, sleep evaluation device 101, the item number 5 bedtime table ST _2, the time zone of the 2 o'clock 1 Enter “1” in the corresponding column 14. Similarly, according to item number 6 of the sleep diary SD ₂ , the wake-up time is between 8 o'clock and 9 o'clock, so the sleep evaluation device 101 is the number of items 5 in the wake-up time table WT ₂ from 8 o'clock to 9 o'clock. Enter “1” in the column 21 corresponding to the time zone. Similarly, sleep evaluation device 101, No. 7 sleep diary SD _2, in correspondence with the recording of item No. 8, the input No. 6. bedtime table ST ₂ and the rising time table WT _2, the No. 7 I do.

After the input to bedtime table ST ₂ and the rising time table WT ₂ of No. 1 to No. 7, sleep evaluation device 101, the total value SumST of each column, calculates a SumWT, also, their SumST, the SumWT The maximum value is detected from the inside. According to the example shown in FIG. 79, the maximum value of sumST is detected as “5” in column 12, and the maximum value of sumWT is detected as “5” in column 19. Furthermore, the sleep evaluation apparatus 101 also detects sumST and sumWT in the column immediately before and the column immediately after the column indicating the detected maximum value. As a result, the bedtime table ST ₂ of FIG. 79, "0-5-0" than

columns

11, 12 and 13 (sum of 3 values "5") SumST pattern that is detected, similarly, waking In the time table WT ₂ , a sumWT having a pattern of “0-5-0” (the total value of the three values is “5”) is detected from the

columns

18, 19, and 20.

The sleep evaluation apparatus 101 displays a pattern “0-5-0” (sum of three values is “5”) including the maximum value of sumST, sumST in the immediately preceding column, and sumST in the immediately following column, Since it matches with the pattern number 25 by matching with the pattern table, the score “6” corresponding to this is determined as the “sleeping time score”. Similarly, the sleep evaluation apparatus 101 displays the pattern “0-5-0” (the total value of the three values is “5”) including the maximum value of sumWT, the sumWT in the immediately preceding column, and the sumWT in the immediately following column. Since it matches with the sleep habit pattern table and matches the pattern number 25, the score “6” corresponding thereto is determined as the “wake-up time score”. The sleep evaluation apparatus 101 calculates the average value of the bedtime score “6” and the wake-up time score “6” determined as described above, determines the average value “6” as the “sleep habit score”, Store in the storage unit 109.

Sleep Diary SD ₃ shown in FIG. 80 shows an example of a subject sleeping and waking Slows daily. In such a sleep diary SD ₃ , after the recording up to item No. 7 is made, the bedtime time score and the wake-up time score are calculated along the records for the most recent 7 days (ie, item Nos. 1 to 7). An example will be described.

Along No. 1 to No. 7 sleep diary SD _3, as shown in FIG. 81, the sleep evaluation device 101, No. 1 to No. 7 bedtime table ST _3, section rising time table WT ₃ Enter in No. 1 to No. 7.

In the example shown in FIG. 81, the maximum value of sumST is detected as “1” in columns 12 to 18, and the maximum value of sumWT is detected as “1” in columns 19 to 24 and column 1. Furthermore, the sleep evaluation apparatus 101 also detects sumST and sumWT in the column immediately before and the column immediately after the column indicating the detected maximum value. As a result, in the bedtime table ST ₃ and the wake-up time table WT ₃ in FIG. 81, “0-1-1” (total value of three values is “2”), “1-1-1” (total of three values) A sumST or sumWT of any pattern of “3”) or “1-1-0” (the total value of the three values is “2”) is detected.

The sleep evaluation apparatus 101 includes patterns “0-1-1” and “1-1-1-” including the maximum value of sumST (sumWT), sumST (sumWT) in the immediately preceding column, and sumST (sumWT) in the immediately following column. “1” and “1-1-0” are checked against the sleep habit pattern table and determined to match at least one of

pattern numbers

42, 44, and 45. As described above, in the case of matching with a plurality of pattern numbers, a condition is given that a higher-order condition (a pattern having a high ternary total value) is preferentially applied in the sleep habit pattern table. Thereby, the pattern number 42 is applied, and the score “0” corresponding to this is determined as the “sleeping time score” (“wake-up time score”). The sleep evaluation apparatus 101 calculates the average value of the bedtime time score “0” and the wake-up time score “0” determined as described above, determines the average value “0” as the “sleep habit score”, Store in the storage unit 109.

Sleep diary SD ₄ shown in FIG. 82 shows an example of a random subject bedtime and wake-up is daily. In such a sleep diary SD ₄ , after the records up to item number 8 are made, the bedtime time score and the wake-up time score are calculated according to the records for the most recent 7 days (that is, item numbers 2 to 8). An example will be described.

Along No. 2 to No. 8 of the sleep diary SD _4, as shown in FIG. 83, the sleep evaluation device 101, No. 1 to No. 7 bedtime table ST _4, section rising time table WT ₄ Enter in No. 1 to No. 7.

According to the example shown in FIG. 83, the maximum value of sumST is detected as “2” in

columns

11, 12, and 14, and the maximum value of sumWT is detected as “4” in column 19. Furthermore, the sleep evaluation apparatus 101 also detects sumST and sumWT in the column immediately before and the column immediately after the column indicating the detected maximum value. As a result, the bedtime table ST ₄ in FIG. 81, "0-2-2" (the sum of 3 values "4"), "2-2-1" (the sum of 3 values "5") "1-2-0" (the sum of 3 values "3") SumST of any pattern are detected, the wake-up time table ST _4, sumWT the pattern of "0-4-1" is detected Will be.

The sleep evaluation apparatus 101 includes patterns “0-2-2”, “2-2-1”, “1-2-0” including the maximum value of sumST, the sumST in the immediately preceding column, and the sumST in the immediately following column. "Is compared with the sleep habit pattern table and it is determined that at least one of the pattern numbers 36, 32, and 41 matches. In the case of coincidence with a plurality of pattern numbers, the pattern number 32 is applied under the condition that the higher-order condition of the sleep habit pattern table is preferentially applied.
Here, in the case of a pattern such as “2-2-1,” in the bedtime table ST, the sumST is “2” in a column immediately before the column of the maximum value of the sumST or in a column far from the column immediately after. There is a possibility that there is a column that is. Therefore, depending on whether there is a column in which the sumST is “2” in such a distant column, the score is set to “(a) 3”, “(b) 2”, etc. It is preferable to provide a difference. More specifically, in the bedtime table ST (or wake-up time table WT), sumST is “2” in addition to the three columns of (a) the column of the maximum value of sumST, the column immediately before it, and the column immediately after it. If there is no column that is, the score is "3". (B) If there is a column that has a sumST of "2" in a column that is 2 columns or more away from the column of the maximum value of sumST, the score is "2". Add the following condition. It is preferable to add the same conditions for the pattern numbers 31, 34 to 41.
In the above example, the pattern of "2-2-1" to pattern number 32 is applied, the column 11 (immediately before the column) bedtime table ST _4, (column of the maximum value) row 12, column 13 (immediately after Column)). Since the sumST after two columns from the column 12 of the maximum value of sumST (that is, column 14) is “2”, “(b) 2” is applied as the score of the pattern number 32. On the other hand, the sleep evaluation apparatus 101 compares the pattern “0-4-1” including the maximum value of sumWT, the sumWT in the immediately preceding column, and the sumWT in the immediately following column with the sleep habit pattern table, and the pattern number 26 Therefore, the score “5” corresponding to this is determined as the “wake-up time score”. The sleep evaluation apparatus 101 calculates the average value of the bedtime score “2” and the wake-up time score “5” determined as described above, and determines the average value “3.5” as the “sleep habit score”. And stored in the storage unit 109.

The sleep habit score (sixth score) thus determined is the sleep evaluation score described in the third embodiment, that is, the sleep depth score (first score), the sleep cycle score (second score), and the sleep time score. The graph may be displayed together with the (third score), the midway awakening score (fourth score), and the body movement frequency score (fifth score) (see step S1274). Although the graph display is not particularly limited, as an example, as shown in FIG. 84, the sleep depth score (first score), the sleep cycle score (second score), and the sleep time score (first score) radiate from the center. 3 score), midway awakening score (fourth score), body movement frequency score (fifth score), and sleep habit score (sixth score).

As a second modification of the third embodiment, the five types of scores (sleep depth score (first score), sleep cycle score (second score), sleep time score (third score) described in the third embodiment are used. ), Sleep awakening score (fourth score), body movement frequency score (fifth score)) plus sleep habit score (sixth score), all six types of scores can be calculated and displayed For example, instead of the body movement frequency score (fifth score)) among the scores described in the third embodiment, a sleep habit score (sixth score) is adopted, and all five types of scores are obtained. It is good also as a sleep evaluation apparatus which can be calculated and displayed.

[Fourth Embodiment]
Hereinafter, the form for implementing the sleep evaluation system which is 4th Embodiment by this invention is demonstrated. As shown in the external view of FIG. 134, the sleep evaluation device 101 of the third embodiment is established as one device including the sensor unit 102 and the control box 103, and the control box 103 includes the device according to the present invention. Since a series of processing programs including a regression equation for obtaining the sleep score is already incorporated, the sleep evaluation data can be obtained and the sleep score can be calculated only by the sleep evaluation device 101.
On the other hand, the sleep evaluation system according to the fourth embodiment is for executing a series of processing programs including a measuring device that acquires a biological signal of a subject and a regression equation for obtaining a sleep score (sleep index) in the present invention. And an information processing terminal. The determination unit (equivalent to the determination unit 108 of the third embodiment) that performs sleep stage determination based on the biological signal may be configured in any one of the measurement device and the information processing terminal. The output of the data measured by the measuring device to the information processing terminal is not particularly limited, for example, using a wired or wireless connection means.
According to the present invention, since the regression equation for evaluating the quality of sleep is created based on the measurement data of PSG, a series of processing programs including such a regression equation is stored in the information processing terminal (for example, a personal computer). In addition, the measurement apparatus calculates sleep determination data, which is a plurality of variable data, based on the biological signal detected from the subject, thereby enabling evaluation of the sleep quality of the subject, that is, calculation of the sleep score. . In addition, since the regression equation for evaluating the quality of sleep according to the present invention is created based on the measurement data of PSG, the acquisition device can measure biological information capable of calculating a predetermined item (sleep determination data). If it is, it will not specifically limit. Therefore, for example, using a PSG measurement device as a measurement device, the sleep determination data can be directly substituted for use in the evaluation of sleep quality, which is also useful in the evaluation of sleep quality in medical institutions. Since the specific process flow is the same as that of the sleep measurement apparatus 101 of the third embodiment, a detailed description thereof will be omitted.

In the third and fourth embodiments described above, four sleep evaluation components are selected from nine sleep determination data. However, the present invention is not limited to this, and n (n is 2 indicating the sleep state). From the above natural number) sleep determination data, m (m is a natural number satisfying n> m) sleep evaluation scores having independent relationships are calculated, and the sleep score is calculated based on the m sleep evaluation scores. It may be calculated. In this case, sleep determination data includes sleep onset latency, sleep efficiency, number of mid- and long-term awakenings, deep sleep latency, deep sleep time, number of short-time awakenings, deep sleep rate, differential sleep cycle score, differential total bedtime Score, total bedtime, bed rest latency, sleep time, total sleep time, mid-wake time, REM sleep latency, shallow sleep time, REM sleep time, number of transitions to sleep stage, number of shallow sleep appearances, number of REM sleep appearances, Number of deep sleep appearances, REM sleep duration, REM sleep interval time, REM sleep cycle, sleep cycle, ratio of shallow sleep in the first and second half, ratio of REM sleep in the first and second half, ratio of deep sleep in the first and second half , Items related to sleep depth (for example, at least one of deep sleep rate or deep sleep appearance amount), items related to sleep rhythm (for example, at least one of sleep cycle or differential sleep cycle score), and midway awakening Items related to At least one) and the may be arbitrarily selected in sleep efficiency or medium long awakening times. If items relating to sleep depth, sleep rhythm, and awakening that are important indicators in the evaluation of sleep quality are selected, it is possible to appropriately derive an indicator that indicates the overall level of sleep quality. The sleep evaluation score may include any one or more of a sleep depth score, a sleep cycle score, a sleep time score, and a midway awakening score.

In the third embodiment, as the sleep evaluation apparatus 101, the detection of a respiratory signal by a mattress and a condenser microphone sensor is taken as an example. A piezoelectric element such as a cable, a capacitive sensor, a film sensor, a strain gauge, or the like may be used, and a known device may be used as long as it can detect a respiratory signal, a body motion signal, and a heartbeat signal.

Further, in step S1138 of the midway awakening condition determination described with reference to the flowchart of FIG. 50, “(average respiration rate of all epochs from m = 1 to m = m) ≧ (n = 1 to n = nmax The epoch average respiration rate) × mq ”was used to determine midway arousal based on the respiration rate. The sleep stage is corrected using a heartbeat signal detecting means for detecting a heartbeat-related index and the heartbeat-related index. By further providing a correcting means, for example, “(average heart rate of all epochs from m = 1 to m = m) ≧ (average heart rate of all epochs from n = 1 to n = nmax) × mv” (Here, mv is a constant such that mv> 1), and a condition that satisfies this condition may be determined as an arousal state, and a more accurate arousal determination is possible.

Furthermore, the determination result may be corrected by taking a known correlation using the transition of the determination result of the sleep evaluation apparatus 101 and the transition of the index related to the heartbeat detected by the heartbeat signal detecting means.

In the third and fourth embodiments described above, nine predetermined items and four sleep evaluation scores have been described as examples. However, the present invention is not limited to this, and n (a natural number of 2 or more) ) The sleep score may be calculated using m (n ≧ m, where m is a natural number) sleep evaluation scores obtained by collecting the predetermined items.

In the above-described third and fourth embodiments, the regression equation for evaluating the quality of sleep is described based on the PSG measurement data. For example, the sleep evaluation apparatus 101 according to the present invention is described below. Collect and analyze sample data of unspecified number of subjects related to sleep determination data (eg, length of sleep, number of awakenings, amount of deep sleep, body movement, etc.) Then, a regression equation for determining the quality of sleep may be created and used.

Also, sleep evaluation scores (sleep depth score (first score), sleep cycle score (second score), sleep time score (third score), mid-wake awakening score (fourth score), body movement frequency score (fifth score) ), And the graph display of the sleep habit score (sixth score)), a bar graph or other graph display may be applied in addition to the radar chart as in the third and fourth embodiments described above. Moreover, it may replace with a graph display and may represent with the icon set to each score. For example, the icon of a character displays the level of the score according to the different facial expressions and poses of the character, and the hand icon shows a folded state of the finger (an OK pose with the index finger and thumb fingertip, thumb up The level of the score may be displayed according to a GOOD pose etc.

In the above-described third and fourth embodiments, the example of displaying the sleep score, the transition of the sleep stage (see FIG. 70), and the sleep score transition (see FIG. 72) has been described. It does not have to be displayed.

DESCRIPTION OF SYMBOLS 1 Sleep evaluation apparatus 2 Sensor part (biological information detection means)
3 Control box 4 Display unit 5 Operation unit 6 CPU
7 Biometric Data Detection Unit 8 Judgment Unit 9 Storage Unit 10 Power Supply 20 Evaluation Unit 101 Sleep Evaluation Device 102 Sensor Unit (Biological Information Detection Unit)
103 Control Box 104 Display Unit 105 Operation Unit 106 CPU
107 Biometric data detection unit 108 Judgment unit 109 Storage unit 110 Power supply 120 Evaluation unit

Claims

A sleep evaluation system comprising: a measurement device including biological information detection means for detecting biological information of a subject and outputting it as a biological signal; and an information processing terminal that calculates a sleep index of the subject based on the biological signal. ,
The information processing terminal
Principal component analysis is performed on a plurality of types of predetermined items including at least items related to sleep depth, items related to sleep rhythm, and items related to mid-wakefulness extracted based on PSG measurement data The sleep evaluation score is calculated by multiplying the principal component coefficient for each predetermined item of the obtained sleep evaluation score by the sleep determination data corresponding to the predetermined item calculated from the biological signal of the subject,
Calculating a sleep disorder discrimination probability obtained by performing logistic regression analysis on the sleep evaluation score;
A sleep evaluation system, wherein the sleep index of the subject is calculated based on the sleep disorder discrimination probability.
The item related to the sleep depth includes at least one of a deep sleep rate or a deep sleep appearance amount, the item related to the sleep rhythm includes at least one of a sleep cycle or a differential sleep cycle score, and the midway The sleep evaluation system according to claim 1, wherein the item related to awakening includes at least one of sleep efficiency and the number of times of awakening in the middle and long period.
The sleep evaluation system according to claim 1, wherein the sleep evaluation score includes one or more of a sleep depth score, a sleep cycle score, a sleep time score, and a midway awakening score.
The predetermined items of the plurality of types are deep sleep rate, differential sleep cycle score, total bedtime, sleep cycle, deep sleep appearance amount, differential total bedtime score, number of middle and long awakening times, number of short time awakenings, and The sleep evaluation system according to any one of claims 1 to 3, wherein the sleep evaluation system includes sleep efficiency.
A sleep evaluation device comprising: biological information detection means for detecting biological information of a subject and outputting it as a biological signal; and a determination unit that calculates a sleep index of the subject based on the biological signal,
The determination unit
Principal component analysis is performed on a plurality of types of predetermined items including at least items related to sleep depth, items related to sleep rhythm, and items related to mid-wakefulness extracted based on PSG measurement data The sleep evaluation score is calculated by multiplying the principal component coefficient for each predetermined item of the obtained sleep evaluation score by the sleep determination data corresponding to the predetermined item calculated from the biological signal of the subject,
Calculating a sleep disorder discrimination probability obtained by performing logistic regression analysis on the sleep evaluation score;
The sleep evaluation apparatus, wherein the sleep index of the subject is calculated based on the sleep disorder discrimination probability.
The item related to the sleep depth includes at least one of a deep sleep rate or a deep sleep appearance amount, the item related to the sleep rhythm includes at least one of a sleep cycle or a differential sleep cycle score, and the midway The sleep evaluation apparatus according to claim 5, wherein the item relating to awakening includes at least one of sleep efficiency or the number of times of awakening in the middle and long period.
The sleep evaluation apparatus according to claim 5 or 6, wherein the sleep evaluation score includes any one or more of a sleep depth score, a sleep cycle score, a sleep time score, and a midway awakening score.
The predetermined items of the plurality of types are deep sleep rate, differential sleep cycle score, total bedtime, sleep cycle, deep sleep appearance amount, differential total bedtime score, number of middle and long awakening times, number of short time awakenings, and The sleep evaluation apparatus according to any one of claims 5 to 7, comprising sleep efficiency.
A sleep evaluation apparatus comprising: biological information detection means for detecting biological information of a subject and outputting it as a biological signal; and a determination unit that determines the sleep state of the subject based on the biological signal,
The determination unit
At least, a sleep evaluation score is calculated for each of a plurality of types of predetermined items including an item related to sleep depth, an item related to sleep rhythm, and an item related to mid-wake awakening,
A sleep evaluation device that determines which of the predetermined sleep types the sleep state of the subject corresponds to based on the sleep evaluation score.
The sleep type is a type stored in advance according to the characteristics of the sleep content,
The determination unit calculates a determination value for each sleep type using the calculated sleep evaluation score, and determines which one of the predetermined sleep types corresponds to the determination value. The sleep evaluation apparatus according to claim 9.
The sleep evaluation score used to determine which of the predetermined sleep types includes at least a sleep depth score, a sleep cycle score, and a midway awakening score, or in addition to these, a sleep time score is further included. The sleep evaluation apparatus according to claim 9 or 10, wherein:
The sleep evaluation apparatus according to any one of claims 9 to 11, further comprising a display unit capable of collectively displaying the sleep evaluation scores calculated for the plurality of types of predetermined items.
The sleep evaluation score collectively displayed on the display unit includes a sleep evaluation score used to determine which of the predetermined sleep types corresponds to, or in addition to these, a body movement frequency score and / or sleep The sleep evaluation apparatus according to claim 12, further comprising a habit score.
The sleep evaluation according to any one of claims 9 to 13, further comprising a display unit capable of collectively displaying the sleep evaluation score calculated for each of the plurality of types of predetermined items on a radar chart. apparatus.
A sleep evaluation system comprising: a measurement device including biological information detection means for detecting biological information of a subject and outputting it as a biological signal; and an information processing terminal that determines the sleep state of the subject based on the biological signal. ,
The information processing terminal
Calculating a sleep evaluation score for each of a plurality of types of predetermined items including at least an item relating to sleep depth, an item relating to sleep rhythm, and an item relating to mid-wakening,
A sleep evaluation system that determines which of the predetermined sleep types the sleep state of the subject corresponds to based on the sleep evaluation score.