JP2023115691A - Sleep disorder discrimination device and sleep disorder discrimination method - Google Patents

Abstract

To provide a sleep disorder discrimination device and a sleep disorder discrimination method capable of discriminating comparatively accurately respiratory disturbance of a subject at a sleeping time.SOLUTION: An amplitude envelope of a respiratory signal is modeled, to thereby build algorithm for retrieving successively a peak which is a local maximum value of amplitude fluctuation, and it is estimated based on the algorithm that the respiratory condition of a subject is apnea or low respiratory.SELECTED DRAWING: Figure 1

Description

TECHNICAL FIELD The present invention relates to a sleep disorder discriminating device and a sleep disorder discriminating method, and more particularly to propose a sleep disorder discriminating device and a sleep disorder discriminating method for discriminating an apnea syndrome during sleep of a subject.

Sleep apnea syndrome (SAS) is characterized by loud snoring and recurrent apnea (stopping of airflow for 10 seconds or more) and hypopnea (airflow cessation for 10 seconds or more) during sleep. It is a disease characterized by

Sleep apnea includes obstructive apnea due to mechanical obstruction of the upper airway, central apnea induced by impaired breathing control function of the central nervous system, and a mixture of obstructive and central apnea. It is roughly divided into three types: mixed apnea and mixed apnea.

Although sleep apnea syndrome (hereinafter referred to as SAS) is a common disease that can cause serious health problems such as nervous system and cardiovascular, in addition to chronic fatigue, At present, many people remain undiagnosed because they are asymptomatic or simply have not seen a sleep specialist.

Conventionally, polysomnography has been performed as a method for diagnosing SAS (see Patent Document 1). Polysomnography uses multiple sensors to record brain and muscle activity, track eye movements, and analyze breathing throughout the night to help identify apnea and hypopnea episodes. A hypopnea index (AHI) can be calculated. The apnea-hypopnea index corresponds to the number of apnea/hypopnea episodes per hour of sleep and is used to evaluate the severity of apnea syndrome.

However, polysomnography is an advanced medical test that takes a long time and a high cost. It was also inconvenient for patients who had to spend the night in

For this reason, in recent years, by wearing a small wireless device (microphone) in the subject's throat to capture the subject's breathing sounds and acoustic sounds from blood vessels, the condition associated with sleep apnea has been investigated. A method for wirelessly monitoring has been proposed (see Patent Document 2).

Japanese translation of PCT publication No. 2009-519802 Japanese Patent Publication No. 2016-517324

By the way, in the wireless monitoring method described in US Pat. It is made like this.

However, this wireless monitoring method only counts the number of apnea events that occur within a given time window after denoising the respiratory signal to distinguish between levels of apnea (mild, moderate and severe). . Therefore, there is a problem that the detection accuracy of apnea, which is important for calculating the apnea hypopnea index (AHI) and diagnosing SAS, is not sufficient for practical use.

The present invention has been made in consideration of the above points, and intends to propose a sleep disorder discriminating apparatus and a sleep disorder discriminating method capable of discriminating breathing disorders during sleep of a subject with relatively high accuracy. It is.

In order to solve this problem, in the present invention, it is detachably fixed to the body surface of the subject centered on the suprasternal fossa, and is generated along with the biological activity of the respiratory system and the circulatory system of the subject. and a respiratory signal that extracts only the frequency band corresponding to the subject's breathing sound as a respiratory signal from the electrical signal obtained by converting the vibration detected by the biological vibration detection unit. an amplitude peak searcher for sequentially searching for amplitude fluctuation peaks by envelope analysis of the signal waveform of the respiratory signal extracted by the respiratory signal extracting section; Based on the appearance interval of consecutive peaks, the respiratory state of the subject is detected when the respiratory cycle is not detected for a predetermined period of time or more while the respiratory cycles of the subject's expiration and inspiration are detected in cycle units. Based on the respiratory state estimating unit estimating that the subject is apnea or hypopnea and the frequency of occurrence of apnea or hypopnea estimated by the respiratory state estimating unit, determine whether the subject has respiratory disorders during sleep and a sleep disorder determination unit.

As a result, the sleep disorder discriminating apparatus models the amplitude envelope of the respiratory signal and constructs an algorithm that sequentially searches for peaks that are the local maximum values of amplitude fluctuations. By estimating apnea or hypopnea, respiratory disorder during sleep of the subject can be determined with much higher accuracy than in the past.

Further, in the present invention, the respiratory signal extractor removes signal levels of the electrical signal that are equal to or higher than a predetermined threshold as a pre-stage for extracting the respiratory signal. As a result, the sleep disorder discriminating apparatus obtains a respiratory signal in which an excessively high amplitude level is removed as an abnormal value from the electric signal as vibration generated by the biological activity of the respiratory system and the circulatory system of the subject. be able to.

Furthermore, in the present invention, the respiratory state estimator sets the minimum duration between two consecutive peaks for the peaks sequentially searched by the amplitude peak searcher. As a result, the sleep disorder determination device can remove peaks that do not correspond to respiratory events among peaks that are local maximum values of vibration levels that are sequentially retrieved.

Further, in the present invention, the respiratory state estimating unit uses the average value of the amplitude levels of the past predetermined number of peaks among the peaks sequentially searched by the amplitude peak searching unit as a reference, and estimates only half or less of the average value. , and three times or more of the average value was determined to be noise. As a result, the sleep disorder discriminating device extracts only the apnea event targets among the peaks, which are the local maximum values of the vibration levels sequentially retrieved, and thus estimates the respiratory state of the subject with higher accuracy. It becomes possible to

Furthermore, in the present invention, the bio-vibration detector is attached to the subject's body surface by inserting a double-sided adhesive tape having relatively high adhesiveness. As a result, the apparatus for determining sleep disorders can almost directly detect biological vibrations from the subject's body surface through the double-sided adhesive tape. In particular, the body surface centered on the suprasternal fossa of the subject is the part where the vibration of the tracheal wall can be captured at the shortest distance from the skin, and the vibration accompanying heart and breathing that cannot be heard by the ear (inaudible) area) can also be detected with very high sensitivity.

Furthermore, in the present invention, the sensor is detachably fixed to the body surface of the subject centered on the suprasternal fossa, and detects vibrations generated by biological activities of the respiratory and circulatory systems of the subject. a first step; a second step of extracting, as a respiratory signal, only the frequency band corresponding to the respiratory sound of the subject from the electrical signal obtained by converting the vibration detected by the biological vibration detection unit; and a respiratory signal extraction. A third step of sequentially searching for peaks of amplitude fluctuations by analyzing the envelope of the signal waveform for the respiratory signal extracted by the unit, and based on the amplitude level of the peaks sequentially searched by the amplitude peak search unit and the appearance interval of consecutive peaks. The respiratory state of the subject is apnea or hypopnea when the respiratory cycle is not detected for a predetermined period of time or more while sequentially detecting the respiratory cycle in which the subject's expiration and inspiration are cycle units. and a fifth step of determining the presence or absence of respiratory disorder during sleep of the subject based on the frequency of occurrence of apnea or hypopnea estimated by the respiratory state estimation unit. made it

As a result, in the sleep disorder determination method, an algorithm is constructed to sequentially search for peaks that are local maximum values of amplitude fluctuations by modeling the waveform envelope of the respiratory signal, and the respiratory state of the subject is determined based on the algorithm. By estimating apnea or hypopnea, respiratory disorder during sleep of the subject can be determined with much higher accuracy than in the past.

Advantageous Effects of Invention According to the present invention, it is possible to realize a sleep disorder determination device and a sleep disorder determination method capable of determining breathing disorders during sleep of a subject with relatively high accuracy.

BRIEF DESCRIPTION OF THE DRAWINGS It is a conceptual diagram with which it uses for description of the sleep disorder determination apparatus which concerns on this embodiment. FIG. 2 is a schematic diagram for explaining a state in which the sleep disorder determination device shown in FIG. 1 is attached to a subject; 3 is a block diagram showing the internal configuration of the sleep disorder determination device shown in FIG. 2. FIG. Fig. 3 is a graph representing the waveform and amplitude envelope of a respiratory signal; FIG. 11 is a graph representing sequentially retrieved peak detection results and estimated apnea events. FIG. FIG. 11 is a graph showing the number of apnea and hypopnea events as a function; FIG. FIG. 10 is a chart showing a confusion matrix reflecting estimated results of apnea and hypopnea and SAS diagnosis results calculated from AHI. FIG. 8 is a chart showing performance indicators that can be read from the confusion matrix shown in FIG. 7;

One embodiment of the present invention will be described in detail below with reference to the drawings.

(1) Configuration of Sleep Disorder Determining Apparatus According to Present Embodiment FIG. 1 shows the configuration of a sleep disorder determining apparatus 1 according to the present invention. The sleep disorder determination device 1 includes a wristwatch-type wearable terminal device 2 and an acoustic vibration sensor (biological vibration detection unit) 4 connected via a cable 3 to a connection terminal provided on the side surface of the wearable terminal device 2. It is configured.

The acoustic vibration sensor 4 is detachably fixed to the body surface of the subject centered on the suprasternal fossa, and detects vibrations generated by the biological activities of the respiratory and circulatory systems of the subject. .

Specifically, the acoustic vibration sensor 4 has predetermined performance characteristics (sensitivity: -35 [dB], frequency band: 25 [Hz] to 1 [kHz]), and passes through the respiratory tract during the subject's breathing. At the same time as detecting pressure fluctuations due to turbulence in the airflow, it also detects vibrations (non-audible range) associated with the heart and breathing that cannot be heard by the ear.

The wearable terminal device 2 is a multi-purpose sensor device with a low weight (about 20 [g]) in which the device main body 2X has a small size (56 [mm] x 37 [mm] x 12 [mm]) and a substantially rectangular parallelepiped shape. The belt 5 is wound around the wrist of the subject and worn. The wearable terminal device 2 can also be used as means for measuring biological information of the subject (at least one or more of electrocardiogram, pulse, pulse pressure, oxygen saturation, body temperature, electroencephalogram, blood sugar level, blood pressure and myoelectric potential). It is configured.

In addition, in the wearable terminal device 2, a display unit 6 made of liquid crystal or the like is provided on the surface of the device main body, and according to the control of the control unit 10 (FIG. 3) described later, the detection result by the acoustic vibration sensor 4, the current time, etc. is made to be displayed.

As shown in FIG. 2, when the sleep disorder determination device 1 is used, the wearable terminal device 2 is attached to the wrist (not shown) of the subject using a belt 5, and the wearable terminal device 2 is connected to the cable 3 Acoustic vibration sensor 4 connected via is attached to the subject's suprasternal fossa.

The subject's suprasternal fossa is the body site that provides the most suitable amplitude-to-noise ratio for recording tracheal sounds. In this embodiment, the sensor surface of the acoustic vibration sensor 4 is attached to the body surface of the subject's suprasternal fossa by inserting a medical adhesive tape having relatively high adhesiveness.

As a result, the sleep disorder determination device 1 can almost directly detect biological vibrations from the subject's body surface through the medical adhesive tape. In particular, the body surface centered on the suprasternal fossa of the subject has the most suitable amplitude-to-noise ratio for recording tracheal sounds, which is the part where the vibration of the tracheal wall can be captured at the shortest distance from the skin. It is possible to detect, with extremely high sensitivity, vibrations associated with the heart and breathing (inaudible range) that cannot be heard by the human ear.

When several commercially available medical adhesive tapes were tested, it was found that the double-sided medical tape (product number: 414125) manufactured by Nichiban Medical Co., Ltd. was the most effective.

(2) Internal System Configuration in Wearable Terminal Device FIG. 3 is a block diagram showing the configuration of the control system of the wearable terminal device 2 in the sleep disorder determination device 1. As shown in FIG. As shown in FIG. 3, the control system of the wearable terminal device 2 includes a control unit 10 consisting of a CPU (Central Processing Unit) that controls the entire system, and various data reading and writing according to commands from the control unit 10. and a data storage unit 11 that can be stored in a database.

The control unit 10 comprises a respiratory signal extraction unit 20 , an amplitude peak search unit 21 , a respiratory state estimation unit 22 and a sleep disorder determination unit 23 . The respiratory signal extractor 20 extracts only the frequency band corresponding to the subject's respiratory sound as a respiratory signal from the electric signal obtained by converting the vibration detected by the acoustic vibration sensor (biological vibration detector) 4. .

Specifically, the respiratory signal extraction unit 20 digitally converts the vibration detected by the acoustic vibration sensor (biological vibration detection unit) into an electrical signal with a sampling frequency of 8 [kHz] and a quantization bit number of 16 [bit]. It is transferred to the data storage unit as an MP4 audio file created by oversampling at 16 [kHz].

Subsequently, the respiratory signal extraction unit 20 removes signal levels above a predetermined threshold value from the digital signal read from the data storage unit 11, thereby obtaining a signal from which excessively high amplitude levels are removed as abnormal values. .

Then, the respiratory signal extraction unit 20 filters the digital signal after removing the abnormal value using a 10th-order Butterworth bandpass filter (cutoff frequency: 300 [Hz] and 1,000 [Hz]), Only the frequency band corresponding to the person's breathing sound is extracted as a breathing signal.

Actually, the respiratory signal extraction unit 20 removes amplitude spike noise by amplifying and smoothing the respiratory signal extracted by frequency separation five times using a third-order median filter. FIG. 4A shows the waveform of the actual respiratory signal.

The amplitude peak search unit 21 sequentially searches for amplitude fluctuation peaks in the respiratory signal extracted by the respiratory signal extraction unit 20 by envelope analysis of the signal waveform. That is, the amplitude peak search unit 21 constructs an algorithm that models the amplitude envelope of the respiratory signal and sequentially searches for peaks that are local maximum values of amplitude fluctuations. FIG. 4B shows the amplitude envelope of the respiratory signal in FIG. 4A.

Specifically, the amplitude peak search unit 21 sets the frame size to 2,048 points and sets the hop size, which is the number of samples between consecutive frames, to 1,024 points in order to calculate the local maximum value of the respiratory signal. did.

The respiratory state estimator 22 sequentially detects respiratory cycles in units of cycles of exhalation and inhalation of the subject, based on the amplitude levels of the peaks sequentially retrieved by the amplitude peak retrieving unit 21 and the appearance intervals of successive peaks. . FIG. 5A shows the detection results of peaks detected for 5 minutes.

The respiratory state estimation unit 22 can calculate the total number of respiratory cycles (one cycle corresponds to inspiration and expiration) by dividing the number of sequentially searched peaks by two. The instantaneous respiration rate can also be calculated by setting a temporal window size (10 seconds, 30 seconds, 1 minute, etc.) and counting the number of peaks in each window of the respiratory signal.

Respiratory state estimator 22 can remove peaks that do not correspond to respiratory events by setting the minimum duration between two consecutive peaks sequentially searched by amplitude peak searcher 21 .

Subsequently, the respiratory state estimator 22 uses the average value of the amplitude levels of the past predetermined number of peaks among the peaks sequentially searched by the amplitude peak searcher 21 as a reference, and determines that only half or less of the average value is to be estimated. and judges that three times or more of the average value is noise.

Here, apnea refers to a state in which breathing stops for 10 seconds or longer. Assuming no peaks between apneas, two peaks separated by more than 10 seconds mark an apnea break. The AHI score, which corresponds to the total number of apneas and hypopneas per hour of sleep, is classified as mild if 5 or more and less than 15, moderate if 15 or more and less than 30, and severe if 30 or more. be done.

In this manner, the respiratory state estimating unit 22 estimates that the respiratory state of the subject is apnea or hypopnea when the respiratory cycle is not detected for a predetermined time (for example, 10 seconds). By extracting only the respiratory event target, it becomes possible to estimate the respiratory state of the subject with higher accuracy. FIG. 5(B) shows an apnea event estimated from the peak detection results in FIG. 5(A).

Based on the occurrence frequency of apnea or hypopnea estimated by the respiratory state estimation unit 22, the sleep disorder determination unit 23 determines the presence or absence of respiratory disorder during sleep of the subject.

In this way, the sleep disorder determination device 1 models the amplitude envelope of the respiratory signal, constructs an algorithm for sequentially searching for peaks that are local maximum values of amplitude fluctuations, and based on the algorithm, the respiratory state of the subject. By estimating that the subject is apnea or hypopnea, it is possible to determine the breathing disorder during sleep of the subject with much higher accuracy than in the past.

(3) Effectiveness verification experiment of the sleep disorder discriminating device according to the present embodiment Actually using the sleep disorder discriminating device 1, seven healthy subjects between the ages of 22 and 27 were used as subjects, and breathing disorders during sleep The effectiveness of the discrimination method was verified.

As shown in FIG. 2 described above, each subject is seated on a chair, wears the wearable terminal device on the left wrist, and attaches the acoustic vibration sensor 4 above the sternum so that the medical adhesive tape is inserted. Affix to the fossa.

Five minutes of respiratory status were recorded for each subject. After breathing normally for the first minute, they were instructed to hold their breath periodically for 5 seconds after 1 minute, 10 seconds after 2 minutes, 15 seconds after 3 minutes, and 20 seconds after 4 minutes. Three central apneas (interruptions in airflow for at least 10 seconds) were simulated. Outside of these periods, breathing was regular and did not limit respiratory rate.

FIG. 6 shows the number of events of apnea and hypopnea calculated using the sleep disorder determination device 1 according to the present embodiment, and the actual number of events that could be obtained directly from the signal waveform of the breath sound of the subject. is shown as a function of In FIG. 6, the calculated number of apnea and hypopnea events is plotted on the Y axis, and the measured number of apnea and hypopnea events is plotted on the X axis. Added trend curve. According to this graph, it can be seen that the number of apnea and hypopnea events calculated using the sleep disorder determination device 1 according to the present embodiment is more realistic.

The estimation results of apnea and hypopnea and the SAS diagnosis results calculated from AHI were arranged using a confusion matrix as shown in FIG. 7(A). The estimation results of apnea and hypopnea are, as shown in FIG. was 3, and false negatives (FN) were 5. In addition, the SAS diagnosis results calculated from AHI are, as shown in FIG. 7 (C), 19 true positives (TP), 9 true negatives (TN), 0 false positives (FP), was 0.

Subsequently, accuracy, sensitivity, and specificity, which are performance indicators that can be read from the confusion matrix, were determined. The accuracy rate represents (TP+TN)/(TP+FP+TN+FN), that is, the number of correct evaluations/the number of all evaluations. Sensitivity also represents TP/(TP+FN), the percentage of correctly identified positives (with apnea and hypopnea events). Further, the specificity represents TN/(TN+FP), the proportion of correctly identified negatives (no apnea and hypopnea events).

As shown in FIG. 8, the estimation results of apnea and hypopnea had an accuracy rate of 95.9 [%], a sensitivity of 93.7 [%], and a specificity of 97.6 [%]. Ta. The SAS diagnosis results calculated from AHI were 100 [%] in accuracy, 100 [%] in sensitivity, and 100 [%] in specificity.

In the sleep disorder determination device according to the present embodiment, in the confusion matrix shown in FIGS. It was set to minimize the number of positives (FP), ie to maximize the specificity shown in FIG.

As a result, it is better to detect apneic events in normal subjects than to miss the detection of one or two apneic events when subjects who are SAS patients are regularly apnea. This is because we thought that

In addition to the specificity, the accuracy rate is about 96 [%] for the estimation results of apnea and hypopnea, and 100 [%] for the SAS diagnosis result calculated from AHI. did it.

(4) Other Embodiments As described above, in the present embodiment, the case where the acoustic vibration sensor 4 is applied as the biological vibration detection unit in the sleep disorder determination device 1 was described. Detects, but is not limited to, vibrations generated in the respiratory and circulatory system through the body surface centering on the suprasternal fossa of the subject, up to the inaudible range, which is inaudible to the human ear. If possible, other various vibration detection means may be applied.

Further, in the present embodiment, a specific commercial product is applied as the medical adhesive tape for fixing the acoustic vibration sensor (biological vibration detection unit) 4 to the body surface of the subject. However, the present invention is not limited to this, and various types of double-sided adhesive tapes may be applied as long as they do not affect the skin and have relatively high adhesiveness.

1... sleep disorder determination device, 2... wearable terminal device, 3... cable, 4... acoustic vibration sensor, 5... belt, 6... display unit, 10... control unit, 11... data storage unit, 20... respiratory signal extraction unit, 21... Amplitude peak search unit, 22... Respiratory state estimation unit, 23... Sleep disorder determination unit.

Claims (9)

a bio-vibration detection unit that is detachably fixed to the body surface of a subject centered on the suprasternal fossa and that detects vibrations generated in accordance with biological activities of the respiratory system and the circulatory system of the subject;
a respiratory signal extraction unit that extracts, as a respiratory signal, only a frequency band corresponding to the respiratory sound of the subject from an electrical signal obtained by converting the vibration detected by the biological vibration detection unit;
an amplitude peak search unit for sequentially searching for amplitude fluctuation peaks by envelope analysis of a signal waveform of the respiratory signal extracted by the respiratory signal extraction unit;
Based on the amplitude level of the peak and the appearance interval of the continuous peaks sequentially searched by the amplitude peak search unit, while sequentially detecting the respiratory cycle in units of cycles of exhalation and inspiration of the subject, a respiratory state estimating unit that estimates that the respiratory state of the subject is apnea or hypopnea when a cycle has not been detected for a predetermined time or longer;
a sleep disorder determination unit that determines whether the subject has respiratory disorders during sleep based on the occurrence frequency of apnea or hypopnea estimated by the respiratory state estimation unit. Discriminator. 2. The sleep disorder determination device according to claim 1, wherein the respiratory signal extractor removes signal levels of the electrical signal equal to or higher than a predetermined threshold as a pre-stage of extracting the respiratory signal. 3. The sleep disorder determination according to claim 1 or 2, wherein the respiratory state estimation unit sets a minimum duration between two consecutive peaks for the peaks sequentially searched by the amplitude peak search unit. Device. The respiratory state estimating unit determines, as a reference, only half or less of the average value of amplitude levels of a predetermined number of past peaks among the peaks sequentially searched by the amplitude peak searching unit to be estimated. 4. The sleep disorder discriminating device according to claim 3, wherein three times or more of the average value is determined as noise. 5. The bio-vibration detector according to any one of claims 1 to 4, wherein the body surface of the subject is attached to the body surface of the subject by inserting a double-sided adhesive tape having relatively high adhesiveness. Sleep disorder discriminator. a first step of detachably fixing to the body surface of the subject centered on the suprasternal fossa and detecting vibrations generated in accordance with biological activities of the respiratory system and the circulatory system of the subject;
a second step of extracting, as a respiratory signal, only a frequency band corresponding to the respiratory sound of the subject from an electrical signal obtained by converting the vibration detected by the biological vibration detection unit;
a third step of sequentially searching for amplitude fluctuation peaks by envelope analysis of a signal waveform of the respiratory signal extracted by the respiratory signal extraction unit;
Based on the amplitude level of the peak and the appearance interval of the continuous peaks sequentially searched by the amplitude peak search unit, while sequentially detecting the respiratory cycle in units of cycles of exhalation and inspiration of the subject, a fourth step of estimating that the respiratory state of the subject is apnea or hypopnea if no cycle has been detected for a predetermined period of time;
and a fifth step of determining the presence or absence of respiratory disorder during sleep of the subject based on the occurrence frequency of apnea or hypopnea estimated by the respiratory state estimation unit. Method. 7. The sleep disorder determination method according to claim 6, wherein said second step removes a signal level equal to or higher than a predetermined threshold from said electrical signal as a step prior to extracting said respiratory signal. The sleep disorder determination method according to claim 6 or 7, wherein the fourth step sets a minimum duration between two consecutive peaks for the peaks sequentially searched by the third step. . In the fourth step, of the peaks sequentially retrieved in the third step, the average value of the amplitude levels of the peaks for a predetermined number of times in the past is used as a reference, and only half or less of the average value is determined to be an object to be estimated; 9. The sleep disorder determination method according to claim 8, wherein three times or more of the average value is determined as noise.

Images