JP2009297455A - Sleeping state estimating device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a sleeping state estimating device, improving estimating precision of the sleeping state of a sleeping person. <P>SOLUTION: The device is provided with a sensor part 1 to generate electric charges in accordance with vibration caused by heartbeats and breaths and the like of the sleeping person C on a mattress B, a conversion part 2 to convert the electric charges generated by the sensor part 1 to be output as electric signals, a bioinformation extracting part 3 to extract vibration components of the frequency zone of the heartbeats out of the output signals from the conversion part 2, and a sleeping state estimating part 4 to estimate the sleeping state to be one of a plurality of sleeping states in accordance with the depth of sleeping of the sleeping person C based on the bioinformation obtained by the bioinformation extracting part 3. The sleeping state estimating part 4 estimates body motion of the sleeping person C based on the bioinformation, and estimates the sleeping state of the sleeping person C as the body motion transitive state when detecting the state where body motion of the sleeping person C is occurring, or any of the other sleeping states is applicable. The body motion transitive state is thus included in the sleeping states of the sleeping person C in estimation. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a sleep state estimation device that estimates a sleep state of a sleeping person such as REM sleep and non-REM sleep.

In recent years, techniques for estimating a sleep state without deteriorating the quality of sleep of a sleeper have been developed, and such a technique notifies the sleep state of the sleeper or uses the estimation result for controlling the bedroom environment. It is a very important technology. As a method for estimating the sleeping state of a sleeping person, a non-contact sensor that is not attached to the sleeping person's body is attached to a bed to detect a biological signal such as breathing or heartbeat, and the sleeping state is estimated based on the biological signal. Such a method is generally disclosed in, for example, Patent Documents 1 and 2. The conventional example described in Patent Document 1 includes a sleeper's heartbeat, breathing, the frequency of turning over, the sleep slow wave power density distribution, awake state, REM sleep state, non-REM sleep, or shallow sleep state of the first and second stages, It is estimated which of the deep sleep states of the third and fourth stages. In addition, the conventional example described in Patent Document 2 determines the sleeping state of the sleeping person based on the pulse interval data that is the time interval data of one cycle of the sleeping person's pulse wave and the body movement data indicating the sleeping person's body movement. If the body motion determination unit determines that there is body motion, the pulse interval data measured in parallel with the body motion data is excluded from the series of pulse interval data, and data processing is performed. The sleep state is estimated based on the autonomic nerve index acquired from
JP 2000-215 A JP-A-2005-279113

  However, in the former conventional example, when the sleeper's body movement occurs or the sleep state transitions, the distribution of the sleep state parameter temporarily changes greatly, or the biological signal is extracted due to the influence of the sleeper's body movement There was a case where the sleep state was erroneously estimated due to problems such as inability to do so. On the other hand, in the latter conventional example, since the sleep state is estimated by excluding data in a period determined to have body movement, the above problem can be solved, but until the sleeping person settles down after the body movement occurs, 20 It may take about ˜40 seconds, and an error occurs in the estimation of the sleep state during this period. As a result, there is a possibility that an error occurs in the estimation of the sleep state of the next state.

  The present invention has been made in view of the above points, and an object of the present invention is to provide a sleep state estimation device capable of improving the estimation accuracy of a sleeper's sleep state.

  In order to achieve the above object, the invention according to claim 1 is arranged between a bed table and a mattress placed on the bed table, and responds to vibrations caused by a sleeper's heartbeat and breathing on the mattress. A sensor unit that generates electric charge, a conversion unit that converts the electric charge generated in the sensor unit into an electrical signal and outputs the signal, and a vibration component in a heart rate frequency band and a vibration component in a respiration frequency band of the output signal of the conversion unit A biological information extraction unit for extracting a sleep state, a sleep state estimation unit for estimating one sleep state among a plurality of sleep states according to the sleep depth of the sleeper from the biological information obtained by the biological information extraction unit, The sleep state estimation unit estimates the body movement of the sleeping person based on the biological information, and detects the movement of the sleeping person or a state that does not correspond to any of the sleeping states. The body movement / transient state It was characterized by estimating the sleep state of the sleeping person is.

  According to a second aspect of the present invention, in the first aspect of the invention, the sleep state estimating unit estimates a sleep cycle of the sleeper based on a temporal change in the sleep state, and sleeps the sleeper together with the obtained cycle. The state is estimated.

  The invention of claim 3 is the sleep invention according to claim 1 or 2, further comprising an autonomic nerve activity detection unit that detects an index of an autonomic nerve by analyzing a heartbeat frequency obtained by the biological information extraction unit. The estimation unit estimates the sleep state of the sleeper together with the autonomic nerve index obtained by the autonomic nerve activity detection unit.

  According to a fourth aspect of the present invention, in the invention according to any one of the first to third aspects, the sleep state estimating unit estimates the transition from the body movement / transient state to another sleep state. A transition condition is set in accordance with the state before.

  The invention of claim 5 is the invention of any one of claims 1 to 4, wherein the sleep state estimation unit sets the transition condition of the sleep state and body movement / transient state according to the characteristics of the sleeper's biological information. It is characterized by doing.

  According to the first aspect of the present invention, the body movement / transient state that has been excluded in the past is set as the body movement / transient state, and the sleep state is estimated including the body movement / transient state. Errors in estimation of dynamic / transient states and sleep states can be reduced, and the sleep state estimation accuracy of a sleeping person can be improved.

  According to the invention of claim 2, by estimating the sleep cycle, it is possible to exclude the estimation result of the sleep state that is clearly deviated from the sleep cycle, and to reduce the sleep state estimation error.

  According to the invention of claim 3, by using the index of the autonomic nerve, it is possible to reduce the estimation error of the sleep state at the time of sleep or REM sleep. In addition, the degree of relaxation and tension of the sleeping person can be estimated.

  According to the invention of claim 4, since it is possible to set an appropriate sleep state transition condition according to each of awakening and sleep, compared with the case where the sleep state is estimated only by biological information such as heartbeat and respiration. Sleep state estimation accuracy can be improved.

  According to the invention of claim 5, since it is possible to set an appropriate sleep state transition condition in consideration of differences in biological information such as heart rate for each user, the sleep state is estimated using an average threshold value. The estimation accuracy of the sleep state can be improved compared to the case.

  Hereinafter, embodiments of a sleep state estimation apparatus according to the present invention will be described with reference to the drawings. In the present embodiment, as shown in FIG. 1, vibration caused by the heartbeat and breathing of a sleeper C on the mattress B disposed between the bed table A and the mattress B placed on the bed table A. Sensor unit 1 having one or a plurality of sensor elements (not shown) that generate charges in response to the signal, a conversion unit 2 that converts the charges generated by the sensor unit 1 into an electrical signal, and an output signal of the conversion unit 2 Among them, the biological information extraction unit 3 that extracts vibration components in the frequency band of the heartbeat, and a plurality (four in this embodiment) according to the sleep depth of the sleeping person C from the biological information obtained by the biological information extraction unit 3 It is comprised from the sleep state estimation part 4 which estimates any one sleep state among these sleep states.

  The sensor unit 1 is configured by fixing a sensor element on a flat sensor table (not shown) made of a resin material such as ABS (acrylonitrile-butadiene-styrene resin) or wood such as MDF (medium density fiberboard). , Between the bed table A and the mattress B. The type of the bed table A is not particularly limited, and may be any of a saw board type, a wooden board type, and a mattress type. The sensor element is composed of a piezoelectric element made of a polymer piezoelectric material such as PVDF (polyvinylidene fluoride), for example, and generates an electric charge in response to minute biological vibration caused by the sleeper C's heartbeat or respiration. The electric charge generated in the sensor unit 1 is input to the conversion unit 2 and converted into an electric signal. In the present embodiment, the sensor element is composed of a piezoelectric element. However, for example, it may be composed of an element that generates an electric charge by bending fluctuation. In this case, it is desirable that the sensor base has elasticity enough to bend the sensor element.

  The biological information extraction unit 3 is composed of a band-pass filter that extracts only vibration components in the heartbeat frequency band of 0.5 to 1.5 Hz, for example, and extracts the vibration components in the heartbeat frequency band from the output signal of the conversion unit 2. To do. At this time, when body movement such as turning over occurs, an excessive signal that masks the vibration component of the heartbeat is output from the conversion unit 2, and therefore the body of the sleeping person C is output from the excessive signal output of the conversion unit 2. Estimate movement. The extracted output signal is output to the sleep state estimation unit 4 and is also output to the autonomic nerve activity detection unit 7 described later via the frequency analysis unit 5.

  The sleep state estimation unit 4 calculates the heart rate from the vibration component of the heart beat obtained by the biological information extraction unit 3 and sleeps together with the estimation results of the time series distribution detection unit 6 and the autonomic nerve activity detection unit 7 described later. The sleep state of the person C is estimated. In the present embodiment, the state in which the sleeper C is awakened is the awake state S1, the sleep of the sleeper C in the sleep stages 1 and 2 is shallow, the shallow sleep state S2, and the sleep of the sleeper C in the sleep stages 3 and 4 is sleep. The deep state is the deep sleep state S3, the sleeper C is a REM sleep state S4 in which the sleeper C is a REM sleep, the sleeper C is in a state of body movement such as turning over, or the above-described states S1 to S4. The state where the sleeper C is not present on the bed table A is defined as the body movement / transient state S5, and the state where the sleeper C is in the absent floor state S6.

  The time series distribution detection unit 6 detects the heart rate time series distribution from the output signal of the biological information extraction unit 3 and calculates the heart rate variance from this time series distribution. The obtained heart rate variance is used by the sleep state estimation unit 4 to estimate each sleep state. In general, in the awake state S1 and the REM sleep state S4, the fluctuation of the output signal of the biological information extraction unit 3 becomes large, so that the dispersion of the heart rate increases. In the deep sleep state S3, the output signal of the biological information extraction unit 3 increases. Since the fluctuation is small, the heart rate variance is small. Therefore, each sleep state can be estimated using the variance of the heart rate as an index.

  The frequency analysis unit 5 converts the output signal of the biological information extraction unit 3 from the time domain to the frequency domain by a frequency analysis method such as FFT (Fast Fourier Transform). And the autonomic nerve activity detection part 7 carries out the vibration component of the low frequency band LF (about 0.04-0.15 Hz) among the obtained frequency spectrum distribution, and the high frequency band HF (about 0.15-0. The power spectrum of the vibration component of 4 Hz) is calculated. Here, the ratio of the power spectrum of the low frequency band LF to the sum of the power spectra of the low frequency band LF and the high frequency band HF is used as an index of the sympathetic activity N1, and the power spectrum of the high frequency band HF is used as the activity of the parasympathetic nerve. It is an index of degree. In general, it is considered to be tense when sympathetic nerve is dominant and relaxed when parasympathetic nerve is dominant. Thus, the sympathetic activity N1 is dominant in the awake state S1, the power spectrum of the high frequency band HF is dominant in the deep sleep state S3, and the sympathetic activity N1 is the value at the time of wake in the REM sleep state S4. Each sleep state can be estimated using the activity of the sympathetic nerve and the parasympathetic nerve as an index.

  Hereinafter, a method for estimating the sleep state of the sleeper C in the present embodiment will be described with reference to FIGS. First, when the operation of the present embodiment is started, the state transits unconditionally to the awake state S1 (No. 1 in FIGS. 2 and 3). In the arousal state S1, the period in which body movement does not occur continues for 5 minutes and the heart rate decreases by 10% or more than the heart rate at the time of awakening, or the period in which body movement does not occur continues for 5 minutes and sympathetic nerves When the activity N1 falls by 10% or more than the activity N1 of the sympathetic nerve at the time of awakening, or when the period in which body movement does not occur continues for 10 minutes, the transition to the shallow sleep state S2 occurs (FIG. No. 2 in 2, 3). When body motion occurs, the state transits to body motion / transient state S5 (No. 3 in FIGS. 2 and 3), and in other cases, the awake state S1 is maintained (No. 4 in FIGS. 2 and 3). .

  In the shallow sleep state S2, when a body motion occurs, the body transition / transition state S5 is reached (No. 5 in FIGS. 2 and 3). Also, the period in which no body movement occurs continues for 3 minutes, and the heart rate is 95% or more of the heart rate at awakening and the heart rate variance is 400 or more, or the period in which body movement does not occur is 3 minutes. If the sympathetic nerve activity N1 is within ± 10% of the sympathetic nerve activity N1 during awakening, the state transitions to the REM sleep state S4 (No. 6 in FIGS. 2 and 3). In addition, when the period in which body movement does not occur continues for 10 minutes, the state transits to the deep sleep state S3 (No. 8 in FIGS. 2 and 3), and in other cases, the shallow sleep state S2 is maintained (7 in FIGS. 2 and 3). Number).

  In the deep sleep state S3, when body movement occurs, the body transition / transition state S5 is reached (No. 9 in FIGS. 2 and 3). In addition, when the heart rate becomes 95% or more of the heart rate at the time of awakening and the variance of the heart rate becomes 400 or more, the state transits to the REM sleep state S4 (No. 10 in FIGS. 2 and 3). The sleep state S3 is maintained (No. 11 in FIGS. 2 and 3).

  In the REM sleep state S4, when body motion occurs, the state transits to the body motion / transient state S5 (No. 12 in FIGS. 2 and 3), and in other cases, the REM sleep state S4 is maintained (FIGS. 2 and 2). No. 13 in 3).

  In the body movement / transient state S5, when the period of body movement continues for 3 minutes and the heart rate is within ± 10% of the heart rate at the time of awakening, the state transitions to the awakening state S1 (FIGS. 2 and 3). No. 14). Further, when the heart rate is 95% or more of the heart rate at the time of awakening and the variance of the heart rate is 300 or more or the variance of the high frequency band component HF is 1000 or more, the transition is made to the REM sleep state S4 (FIG. 2, FIG. No. 15 in 3). If body motion occurs, the body motion / transient state S5 is maintained for 20 seconds after the occurrence (No. 17 in FIGS. 2 and 3), and if no heartbeat pulse is detected for 1 second, the absent floor state is maintained. The process proceeds to S6 (No. 18 in FIGS. 2 and 3). In cases other than the above, the state transitions to the shallow sleep state S2 (No. 16 in FIGS. 2 and 3).

  In the absent floor state S6, when body movement occurs, the state transitions to the body movement / transient state S5 (No. 19 in FIGS. 2 and 3), and in the other cases, the absent floor state S6 is maintained (FIGS. 2 and 2). No. 20 in 3).

  As described above, the period of body movement that has been excluded in the past is the body movement / transient state S5 and the sleep state is estimated including the body movement / transient state S5. -Errors in estimation of the transient state S5 and the sleep state can be reduced, and the estimation accuracy of the sleep state of the sleeper C can be improved. In addition, by using an autonomic nerve index such as the sympathetic nerve activity N1 and the power spectrum of the high frequency band HF, it is possible to reduce an estimation error of the sleep state during sleep onset or REM sleep. In addition, the degree of relaxation and tension of the sleeping person C can be estimated.

  By the way, biometric information such as heart rate obtained from the sleeping person C varies depending on the user, and when the sleep state is estimated using an average threshold as a transition condition, the sleep state estimation result varies. This is because the maximum and minimum widths of the power spectrum of the high frequency band HF, which is an index of sympathetic nerve activity N1 and parasympathetic nerve activity, vary between individuals or the same person. Even if there is, it depends on the physical condition and the daytime life, and it is caused by the largeness.

  Therefore, in the present embodiment, when the operation of the present embodiment is started, the sleep state estimation unit 4 measures and stores the heart rate during awakening / resting for several days, and calculates the average value of the heart rate. Based on this, the threshold value of the heart rate in the transition condition is changed. For example, in the transition condition No. 2 in FIG. 3, the transition is made when the heart rate is “decreased by 10% or more” than the heart rate at the time of awakening, but depending on the average value of the heart rate at the time of awakening and resting. If it is 50 beats / minute or less, it shall be “decrease by 8% or more”, if it is 50-60 beats / minute, it will be “10% or more reduction”, and if it is 60 beats / minute or more, it shall be “12% or more reduction”. Thus, it is possible to reduce the variation in the sleep state estimation result for each user. In the case of a user who has little fluctuation in the power spectrum of the sympathetic nerve N1 and the high frequency band HF while sleeping, the sympathetic nerve activity N1 and the power spectrum of the high frequency band HF are measured overnight. The variation of the sleep state estimation result can be reduced by changing the transition condition according to the ratio.

  Further, as shown in FIG. 2, the body movement / transient state S5 may transition from all other sleep states, and the body movement / transient state S5 during awakening and the body movement / transient state S5 during sleep The transition conditions may be different. Therefore, the state immediately before the transition to the body movement / transition state S5 is divided into a sleep state group (shallow sleep state S2, deep sleep state S3, REM sleep state S4) and awake state group (wake state S1, absent bed state S6). The transition condition from the body motion / transient state S5 to the next sleep state may be changed according to which group has transitioned to the body motion / transient state S5. For example, in the sleep state group, the condition for transition from the body motion / transient state S5 to the awake state S1 is “the body motion is occurring for 3 minutes and the heart rate is within ± 10% of the heart rate at the time of awakening. In the wakefulness group, “the period during which body motion occurs continues for 1 minute”. As described above, it is possible to set appropriate sleep state transition conditions according to awakening time and sleep time, so sleep state estimation is compared with the case where sleep state is estimated only from biological information such as heartbeat and respiration. Accuracy can be improved.

  In general, although there are individual differences and physical condition differences, human sleep repeats a shallow sleep state S2, a deep sleep state S3, and a REM sleep state S4 in a cycle of about 90 minutes. Therefore, the sleep state estimation unit 4 may estimate the sleep period of the sleeper C from the temporal change of the sleep state, and may estimate the sleep state of the sleeper C together with the obtained period. Specifically, when the sleep state estimation unit 4 first estimates the appearance timing of each sleep state with the sleep cycle of the sleeper C being approximately 90 minutes, and a sleep state that is significantly different from the cycle is estimated. The sleep state is estimated while excluding the sleep state. Here, the sleep cycle is estimated by accumulating several hours of data estimating each sleep state, and the sleep state is estimated together with the cycle, so that the sleep cycle corresponding to the user is calculated. It is possible to improve the estimation accuracy of the sleep state. However, if the sleep cycle cannot be detected due to ingestion of antiallergic drugs or alcohol acting on the autonomic nervous system, autonomic ataxia, menopause, etc., the cycle is set to about 90 minutes.

  In the present embodiment, the sleep state is estimated based on the heart rate of the sleeper C. However, the sleep state may be estimated by extracting a respiratory component from the electrical signal of the conversion unit 2. Moreover, you may estimate a sleep state using both a heartbeat component and a respiratory component.

It is the schematic which shows embodiment of the sleep state estimation apparatus which concerns on this invention. It is a state transition diagram same as the above. It is a figure which shows the transition conditions of each sleep state same as the above, and a body movement and a transient state.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Sensor part 2 Conversion part 3 Biological information extraction part 4 Sleep state estimation part 5 Frequency analysis part 6 Time series distribution detection part 7 Autonomic nerve activity detection part A Bed bed B Mattress C Sleeper

Claims (5)

  A sensor unit that is disposed between the bed table and the mattress placed on the bed table and generates charges in response to vibration caused by a sleeper's heartbeat and breathing on the mattress, and the charges generated by the sensor unit Obtained from the conversion unit that converts the signal into an electrical signal, the biological information extraction unit that extracts the vibration component in the heart rate frequency band and the vibration component in the respiration frequency band from the output signal of the conversion unit, and the biological information extraction unit A sleep state estimation unit that estimates any one sleep state among a plurality of sleep states according to the sleep depth of the sleeper from the obtained biological information, and the sleep state estimation unit sleeps based on the biological information Estimating the body movement of the person who is sleeping, or the state that does not correspond to any of the sleep states is a body movement / transient state, together with the body movement / transient state of the sleeper Characterized by estimating sleep state That sleep state estimation device.   The sleep state estimating unit estimates a sleep period of the sleeper based on a temporal change of the sleep state, and estimates the sleep state of the sleeper together with the obtained period. Sleep state estimation device.   An autonomic nerve activity detection unit that detects an index of the autonomic nerve by analyzing the frequency of the heartbeat obtained by the biological information extraction unit is provided, and the sleep state estimation unit is an autonomy obtained by the autonomic nerve activity detection unit The sleep state estimation apparatus according to claim 1, wherein the sleep state of the sleeping person is estimated together with a nerve index.   The sleep state estimating unit sets a transition condition according to a state before the body motion / transient state when estimating a transition from the body motion / transient state to another sleep state. The sleep state estimation apparatus according to any one of 1 to 3.   The sleep according to any one of claims 1 to 4, wherein the sleep state estimation unit sets a transition condition of a sleep state and a body movement / transient state according to a feature of a living person's biological information. State estimation device.

Images