JP4434122B2 - Apnea estimation device - Google Patents

Description

  The present invention relates to an apnea estimation device that detects an occurrence time of an apnea state of a sleeping person while sleeping.

  Sleep Apnea Syndrome is one of respiratory disorders that repeatedly repeats apnea for several tens of seconds during sleep, and is classified into three types: central type, obstructive type, and mixed type.

  Central sleep apnea syndrome is a problem that occurs when the respiratory control system is damaged due to diseases of the central nervous system or when the respiratory center functions abnormally. This stops the stimulation of the human body, and there is no breathing movement. Abnormal breathing (Chain Stokes breathing) seen in patients with brain disease and heart failure falls into this central type.

  On the other hand, obstructive sleep apnea syndrome is that the respiratory center works normally and breathing exercises in the chest and abdomen are performed, but obstruction of the upper airway due to tongue root depression etc. makes breathing in the nose and mouth impossible It is. This obstructive type exhibits a strange movement in which the movements of the abdomen and chest are opposite to each other, and is accompanied by a large amount of snoring when resuming breathing, and the breathing level decreases during apnea. Note that. Mixed sleep apnea syndrome, which begins with central apnea and shifts to obstruction, is considered a type of obstruction.

  By the way, the apnea estimation apparatus which estimates the apnea state which generate | occur | produces during sleep conventionally is known as what is used for the diagnosis etc. of sleep apnea syndrome. Patent Document 1 discloses a sleep respiratory information measurement device that can estimate an apnea state without attaching a sensor or the like to a human body.

The sleep breathing information measuring device disclosed in Patent Document 1 includes a breathing sensor embedded in a bed pad and a control device that detects an apnea state based on a signal from the breathing sensor. In this sleep respiratory information measuring device, the control device determines whether or not the sleep of the sleeper is stable using the autocorrelation coefficient of the signal of the respiratory sensor. A threshold for determining the presence or absence of is determined. Then, the control device compares this threshold value with the signal level of the respiration sensor to detect a respiration cycle, and detects a time when the respiration cycle is 10 seconds or more as an apnea state.
JP 2003-265439 A

  By the way, conventionally, if the state of stable breathing continues, the threshold value is not updated and it is used for detection of the respiratory cycle, so the sleep phase of the sleeper changes slightly while the breathing is stable Thus, there is a problem that an apnea state cannot be accurately estimated. That is, in such a case, although the signal level of the respiration sensor changes with the change of the sleep phase, the threshold value is not updated, and thus there is a possibility that an error occurs in the detection of the respiration cycle.

  The present invention has been made in view of such a point, and an object thereof is an apnea estimation device capable of accurately estimating an apnea state without being affected by a change in sleep phase of a sleeper or a rough body motion. Is to provide.

  The first invention includes body motion detection means (20) for detecting a body motion of a sleeping person and outputting a body motion signal, and based on time series data of the body motion signal from the body motion detection means (20). Assumes an apnea estimation device (10) for estimating the sleep apnea state. The apnea estimation device (10) includes a breathing signal extraction means (50) for extracting breathing signal time series data generated by the sleep motion of the sleeper from the time series data of the body motion signal, In the time-series data, each of the forward direction in which the time advances and the reverse direction in which the time goes back is set as the processing direction, and the respiratory signal increases at a higher rate of change than when the respiratory signal increases in each of the forward direction and the reverse direction. Respiration signal processing means (60) for generating time-series data for forward processing and time-series data for backward processing by performing processing to follow the waveform of the signal time-series data, and time-series data for the forward processing And an apnea estimation means (61) for comparing the smaller one of the time series data of the reverse processing and the time series data of the respiratory signal at each time point to estimate the apnea state.

  In the first invention, an apnea state is estimated by comparing the smaller of the time-series data of the forward processing and the time-series data of the backward processing with the time-series data of the respiratory signal at each time point. In other words, the time-series data consisting of the smaller one of the time-series data of the forward direction processing and the time-series data of the reverse direction processing at each time point compared with the time series data of the body motion signal is used for estimating the apnea state. It becomes the reference line.

  Here, in the time-series data of the respiratory signal, the signal level may change greatly, for example, when the sleeping phase of the sleeping person changes. In the first aspect of the invention, a process of following the waveform of the respiratory signal time-series data with a larger rate of change than when the respiratory signal increases in each of the forward direction and the reverse direction is performed. Yes. Therefore, as shown in FIG. 8 (A), at the point where the signal level increases greatly, the smaller forward time series data of the forward time series data and the reverse time series data becomes the reference line. The time-series data of the respiratory signal has almost no difference and follows the waveform. When the signal level drops significantly, the time-series data in the reverse direction, which is the smaller of the time-series data in the forward direction and the time-series data in the reverse direction, is the reference. It becomes a line with almost no difference in the time-series data of respiratory signals. In this way, the reference line has a waveform that follows almost no difference even at a location where the signal level of the respiratory signal changes greatly, and therefore always has a waveform that substantially follows the time-series data of the respiratory signal.

  Further, the time-series data of the respiratory signal may include noise data that is not caused by respiratory motion such as vibration caused by opening and closing of a door. Of the noise data, data that has a relatively steep mountain-shaped waveform compared to before and after as shown in FIG. 8B may affect the estimation of the apnea state. In this case, the time-series data of the forward processing and the time-series data of the backward processing increase rapidly to the vicinity of the extreme value of the respiratory signal when viewed from each processing direction, but gradually decrease when exceeding the peak. For this reason, as shown in FIG. 8B, the extreme value of the noise data portion of the waveform of the reference line composed of both smaller data is greatly reduced.

  Further, it is known that the respiratory signal is hardly detected in the central sleep apnea syndrome, and the signal level of the respiratory signal is significantly reduced in the obstructive sleep apnea syndrome. For this reason, as shown in FIG. 8C, the data of the apnea state in the time series data of the respiratory signal has a valley waveform that is relatively steep compared to before and after. In this case, both the time-series data for the forward direction processing and the time-series data for the backward direction processing hardly change because the rate of change when the respiratory signal decreases in each processing direction is small. For this reason, as shown in FIG. 8C, the valley-shaped waveform resulting from the apnea state is almost deleted from the reference line waveform. Therefore, the apnea is compared with the time-series data of the respiratory signal. The state is inferred.

  According to a second aspect of the present invention, in the first aspect, the apnea estimation means (61) performs the time series data of the respiratory signal for the smaller of the time series data of the forward processing and the time series data of the backward processing. Estimate apnea based on the ratio of

  In the second invention, in order to estimate the apnea state, the ratio of the time series data of the respiratory signal to the smaller one of the time series data of the forward processing and the time series data of the backward processing is used. That is, the apnea state is estimated based on how much the portion where the signal level of the respiratory signal is lowered is lower than the reference line.

  According to a third aspect of the present invention, in the first or second aspect, the rough body motion detecting means (40) for detecting the rough body motion of the sleeping person based on the body motion signal detected by the body motion detecting means (20) is provided. The apnea estimation means (61) determines that the time when the coarse body motion detection means (40) detects the coarse body motion of the sleeper is not an apnea state.

  In the third aspect of the invention, the coarse body motion detection means (at the time when the smaller one of the time series data of the forward direction processing and the time series data of the backward direction processing is compared with the time series data of the respiratory signal to estimate the apnea state ( If 40) detects a sleeper's coarse body movement, it is determined that the time is not apnea. That is, since it is known that a sleeper in an apnea state does not cause coarse body movement (relatively large body movement), it is determined that the time during which the rough body movement occurs is not an apnea state.

  In the present invention, when the respiration signal increases in each of the forward direction and the reverse direction, the process of following the waveform of the time-series data of the respiration signal is performed at a larger rate of change than when the respiration signal is decreased. Create series data and time series data for backward processing, and compare the time series data for forward processing and the time series data for backward processing with the time series data for respiratory signals at each time point. If an apnea condition occurs during sleep, the interval between the reference line consisting of the smaller one of the time-series data for forward processing and the time-series data for reverse processing and the time-series data of the respiratory signal There is a difference in the apnea state.

  By performing the above-described processing, the reference line for estimating the apnea state compared with the time-series data of the respiratory signal is almost free from changes in the signal level of the respiratory signal due to changes in the sleep phase of the sleeper. There is no difference and the waveform follows, and the waveform of noise data not caused by respiratory motion is relaxed. Therefore, it is possible to accurately estimate the apnea state by avoiding misjudgment due to a change in sleep phase or coarse body movement, and it is possible to improve the estimation accuracy of the apnea state and the reliability of the estimation result.

  Further, according to the third aspect, since it is known that a sleeper who is in an apnea state does not cause coarse body movement, it is determined that the time when the rough body movement is detected is not in an apnea state. ing. Accordingly, the apnea state can be estimated more accurately, and the apnea state estimation accuracy and the reliability of the estimation result can be improved.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  The apnea estimation device (10) of the embodiment extracts a signal due to respiratory motion in a body motion signal of a sleeper, and processes and analyzes the time-series data of the extracted signal after waking up, and is generated during sleep An apnea state is estimated. The apnea estimation device (10) includes a sleep sensor (20) and a circuit unit (30).

  The sleep sensor (20) constitutes a body motion detection means for detecting a body motion of a sleeping person and outputting a body motion signal. The sleep sensor (20) includes a pressure sensing part (21) and a pressure receiving part (22).

  As shown in FIG.1 and FIG.2, the pressure sensitive part (21) of this embodiment is laid on the mattress laid on the bed of bedclothes, such as a bed. This pressure-sensitive part (21) is comprised by the elongate and hollow tube, and the space is formed in the inner side. When the sleeping person lies on the bed, pressure / vibration is transmitted to the pressure sensing part (21) along with the body movement of the sleeping person, and the internal pressure of the pressure sensing part (21) acts on the pressure receiving part (22).

  The pressure receiving part (22) includes a casing (23) and a sensor part (24) accommodated in the casing (23). A sensor part (24) is comprised by a microphone, a pressure sensor, etc. The sensor unit (24) receives the internal pressure applied from the pressure sensing unit (21), and outputs the internal pressure as a body motion signal to the circuit unit (30) via the lead wire (25). In the present embodiment, a minute leak groove (26) is formed at the connection position between the pressure sensitive part (21) and the pressure receiving part (22). For this reason, for example, when a sleeping person gives a strong impact to the bed, a sudden rise in internal pressure acts on the sensor unit (24), resulting in a malfunction of the sensor unit (24) or body movement. This prevents the signal from becoming saturated.

  As shown in FIG. 3, the circuit unit (30) includes a rough body motion detection means (40), a respiratory signal extraction means (50), a respiratory signal processing means (60), and an apnea estimation means (61). Yes.

  The coarse body motion detection means (40) creates time series data of body motion and time series data of the body motion threshold based on the body motion signal output from the sleep sensor (20), and the time series of the body motion. The data and the time series data of the body motion threshold are compared to detect the occurrence time of the rough body motion while sleeping. The coarse body motion detection means (40) includes a band pass filter unit (41), an envelope detection unit (42), a low-pass filter unit (43), a body motion data creation unit (44), and a body motion threshold creation unit ( 45).

  The band pass filter unit (41) extracts a signal of a predetermined frequency band (7.5 ± 2.5 Hz) from the body motion signal output from the sleep sensor (20). The envelope detector (42) rectifies (converts to an absolute value) the filtered signal and then performs peak hold for a predetermined time. The low-pass filter unit (43) extracts a signal in a relatively low frequency vibration band (0.5 Hz or less) from the signal from the envelope detection unit (42). The body motion data creation unit (44) creates time series data of body motion from the signal extracted by the low-pass filter unit (43). The body motion threshold value creation unit (45) creates time series data of the body motion threshold value from the signal extracted by the low-pass filter unit (43).

  The respiration signal extraction means (50) extracts respiration intensity representing the respiration strength of the sleeper based on the body motion signal output from the sleep sensor (20). The respiratory intensity corresponds to the respiratory signal according to the present invention. The respiration signal extraction means (50) includes a band pass filter unit (51), a respiration interval extraction unit (52), and a respiration intensity calculation unit (53).

  The band pass filter unit (51) extracts a signal of a predetermined frequency band (0.25 ± 0.15 Hz) as a respiratory waveform from the body motion signal output from the sleep sensor (20). The respiration interval extraction unit (52) calculates a respiration cycle in each time segment divided every 10 seconds from the start of measurement using a respiration waveform. The respiratory intensity calculation unit (53) calculates the respiratory intensity in each time segment divided every 10 seconds from the start of measurement by creating a frequency spectrum of the body motion signal.

  The respiratory signal processing means (32) creates time-series data for forward processing and time-series data for backward processing, which will be described later, based on the time-series data of the respiratory intensity extracted by the respiratory signal extraction means (31), The smaller one of the processing values of each time series data is extracted as a processing value for comparison with the respiratory intensity. The apnea estimation means (61) compares the respiratory intensity and the comparison processing value, detects the occurrence time of an apnea when sleeping, and displays it on a display unit such as a monitor.

-Sleep judgment operation-
The operation of the apnea estimation device (10) of the embodiment will be described.

  When the sleeper is lying on the bed of the bedding and the power is set to the ON state, the apnea estimation device (10) is connected to the circuit unit (30 ) Outputs a body motion signal. In the circuit unit (30), the body motion signal from the sleep sensor (20) is input to both the coarse body motion detection means (40) and the respiratory signal extraction means (50).

  First, the operation of the coarse body motion detection means (40) will be described in detail with reference to the flowchart of FIG. In the rough body motion detection means (40), the body motion signal from the sleep sensor (20) is input to the band pass filter unit (41). The band pass filter unit (41) filters the body motion signal in a band (7.5 ± 2.5 Hz) corresponding to the natural vibration band of the human body (step ST1). Next, the envelope detection unit (42) rectifies (converts to an absolute value) the filtered signal, performs peak hold for a predetermined time (for example, 0.1 second), and performs so-called envelope detection processing ( Step ST2). Next, the low-pass filter unit (43) extracts a signal in a relatively low frequency vibration band (0.5 Hz or less) (step ST3).

  Next, the body motion data creation unit (44) performs integration and averaging of the signals extracted by the low-pass filter unit (43) at 10-second intervals, and the body motion time series as shown in FIG. Data is created (step ST4). As a result of the signal integration / averaging process, instantaneous fluctuations in the body motion signal are alleviated. The time series data of body movement may be created by obtaining the standard deviation of the signal extracted by the low-pass filter unit (43).

  In addition, the interval of the integration / average processing by the body motion data creation unit (44) is a normal respiratory cycle (generally 2.5 to 5 seconds) to reflect a signal resulting from sleep of the sleeper. A larger interval is preferred, and is less than the sleeper's convulsive cycle (typically 20-30 seconds) to suppress fluctuations in body motion signals due to involuntary limb spasms during sleep of the sleeper An interval is preferred.

  Next, the body motion threshold value creation unit (45) creates a body motion threshold value based on the signal extracted by the low-pass filter unit (43). The body movement threshold value creation unit (45) calculates the minimum value from the signal extracted by the low-pass filter unit (43) for each 10-second time segment, and alleviates the time series change of the minimum value with an exponential function. The body motion threshold is created by calculating the signal level and multiplying the signal level by a predetermined constant.

  Specifically, as the time constant used in the exponential function, different values are used when the minimum value increases and decreases. For example, when the minimum value (value relaxed by the exponential function) in a certain time segment is Min, and the minimum value (value before relaxation by the exponential function) in the next time segment is increased to X1 (X1> Min) When the time constant τ1 is 1200 [sec], the minimum value Min ′ that is relaxed by the exponential function of the next time section is calculated by Equation 1.

Formula 1: Min ′ = Min × α + X1 (1−α)
Here, α = exp (−to / τ1), τ1 = 1200 [sec]
On the other hand, for example, when the minimum value (value relaxed by an exponential function) in a certain time segment is Min, and the minimum value (value before relaxation by the exponential function) in the next time segment decreases to X1 (X1) <In the case of Min), the time constant τ2 is set to 60 [sec], and the minimum value Min ′ that is relaxed by the exponential function of the next time section is calculated by Equation 2.

Formula 2: Min ′ = Min × β + X1 (1−β)
Here, β = exp (−to / τ2), τ2 = 60 [sec]
As shown in FIG. 6, the body movement threshold value creation unit (45) multiplies the signal level (minimum value Min ′) relaxed by the exponential function by a predetermined constant (for example, 4) and traces the multiplied value. A time series data of a proper body motion threshold is created (step ST5). The coarse body motion detecting means (40) compares the time series data of the body motion and the time series data of the body motion threshold value, and determines that there is a rough body motion at a location where the body motion is larger.

  As the time constant of the exponential function, a value larger at the time of increasing change than at the time of decreasing change of the minimum value is used. For this reason, the signal level relaxed by the exponential function follows the change of the minimum value with a large change rate while the signal level relaxed by the exponential function follows the change of the minimum value with a small change rate. To do. In addition to the exponential function (1-hour delay system step function), for example, a 2-hour delay system step function or other functions may be used as the function for relaxing the change in the minimum value of the body motion signal.

  The operation of the respiratory signal extraction means (50) will be described in detail with reference to the flowchart of FIG. In the respiratory signal extraction means (50), the body motion signal output from the sleep sensor (20) is input to the bandpass filter unit (51). The band pass filter unit (51) filters the signal in a band (0.25 ± 0.15 Hz) corresponding to the natural vibration band of the respiratory motion (step ST6). Thereby, the signal resulting from the sleep motion of the sleeper is extracted as a respiratory waveform from the body motion signal.

  Next, the breath interval extraction unit (52) has a zero crossing time (a time when the breathing waveform and the zero amplitude line intersect) in the process of changing from the valley to the mountain in the breathing waveform extracted by the bandpass filter unit (51). The respiratory cycle is calculated from the interval (step ST7). The respiratory cycle is calculated as an average value in each time segment divided every 10 seconds from the start of measurement.

Next, the respiratory intensity calculation unit (53) creates the frequency spectrum of the body motion signal of each time segment divided every 10 seconds from the start of measurement based on the body motion signal from the sleep sensor (20). The frequency spectrum of the body motion signal is created by a process such as fast Fourier transform (FFT). Then, the respiratory intensity calculation unit (53) sets the magnitude of the spectrum corresponding to the respiratory cycle of the same time segment in the frequency spectrum of the body motion signal of each time segment as the respiratory intensity (step ST8). Thereby, the respiratory intensity of each time segment divided every 10 seconds from the start of measurement is calculated as time series data.

  In this embodiment, since it is generally said that apnea or hypopnea occurs for 10 to 120 seconds in an apnea state in sleep apnea syndrome, it is divided into time intervals of 10 seconds. Processing in the breath interval extraction unit (52) and the breathing intensity calculation unit (53) is performed.

  The processing by the coarse body motion detection means (40) and the respiratory signal extraction means (50) may be performed during measurement of the body motion of the sleeper or after the measurement is completed.

  The respiratory signal processing means (60) and apnea estimation means (61) will be described in detail with reference to the flowchart of FIG.

  The respiratory signal processing means (60) receives the respiratory intensity time-series data created by the respiratory intensity calculator (53). In the respiratory signal processing means (60), in the respiratory intensity time-series data, the forward direction in which the time advances and the backward direction in which the time goes back are set as the processing directions, and the respiratory intensity time-series data in each of the forward direction and the reverse direction are used. A process for following the waveform is performed to create time-series data for forward processing and time-series data for backward processing. The time-series data for forward processing and the time-series data for reverse processing are converted into waveforms of respiratory intensity time-series data at a greater rate of change than when the respiratory intensity increases with respect to the processing direction. Created to follow.

  Specifically, the time-series data of the forward direction process is composed of forward direction process values calculated corresponding to changes in the respiratory intensity in the forward direction with time. The time-series data of the forward processing is created by calculating the forward processing value for each time segment divided every 10 seconds in the forward direction in which the time advances from the start of measurement (step ST11). The calculation of the forward processing value uses Equation 3 when the respiratory intensity increases with time (X2> L1), and uses Equation 4 when the respiratory intensity decreases (X2 <L1).

Formula 3: L2 = L1 × α + X2 (1−α)
Formula 4: L2 = L1 × β + X2 (1−β)
Where α = exp (−to / τ1), β = exp (−to / τ2)
In the above formulas 3 and 4, L1 is the forward processing value of the previous time segment with respect to the processing direction, L2 is the forward processing value calculated by formula 3 or 4, and X2 is the previous processing value with respect to the processing direction. Respiratory intensity for each time segment is shown. Further, to is 10 [sec], which is the same as the length of the time segment in which the respiratory intensity is calculated, τ1 is an upward correction time constant 10 [sec], and τ2 is a downward correction time constant 60 [sec].

  On the other hand, the time-series data of the backward processing is composed of backward processing values calculated corresponding to changes in the respiratory strength in the backward direction over time. The time-series data of the backward process is created by calculating the backward process value in the backward direction in which the time is increased for the same time segment as the forward process value (step ST12). In calculating the backward processing value, the above formula 5 is used when the respiratory intensity increases with time (X3> L3), and the above formula 6 is used when the respiratory intensity decreases (X3 <L3).

Formula 5: L4 = L3 × α + X3 (1−α)
Formula 6: L4 = L3 × β + X3 (1-β)
Where α = exp (−to / τ1), β = exp (−to / τ2)
In the above formulas 5 and 6, L3 is the backward processing value of the previous time segment with respect to the processing direction, L4 is the backward processing value calculated by formula 5 or 6, and X3 is the previous processing value with respect to the processing direction. Respiratory intensity for each time segment is shown. Further, to is 10 [sec], which is the same as the length of the time segment in which the respiratory intensity is calculated, τ1 is an upward correction time constant 10 [sec], and τ2 is a downward correction time constant 60 [sec]. In addition to the above exponential function (1-time delay step function), the functions for calculating the processing values of Equation 3, Equation 4, Equation 5, and Equation 6 are, for example, a 2-hour delay step function and others. The function may be used.

  In Expressions 3, 4, 5, and 6, the upward correction time constant is made smaller than the downward correction time constant. For this reason, the time-series data for forward processing and the time-series data for backward processing calculated using these equations are reduced when the respiratory intensity increases in the processing direction, as shown in FIG. A waveform that follows the waveform of the time-series data of the respiratory intensity is drawn at a larger rate of change than the case.

  Next, the time-series data of the forward processing and the time-series data of the backward processing created by the respiratory signal processing means (60) are input to the apnea estimation means (61). The apnea estimation means (61) selects the smaller one of the forward processing value of the time-series data of the forward processing and the backward processing value of the time-series data of the backward processing in each time section, and the selected smaller one Is extracted as a processing value for comparison with the respiratory intensity (step ST13).

  Next, the apnea estimation means (61) compares the value obtained by dividing the respiration intensity by the comparison processing value in each time section with a determination threshold value (0.25 in this embodiment) (step ST14). Then, for a time segment in which the divided value is larger than the determination threshold value, the process proceeds to step ST18 to determine the normal breathing state, and for a time segment in which the divided value is smaller than the determination threshold value, there is a possibility of an apnea state. It is determined that there is one and the process proceeds to step ST15.

  Here, the determination threshold is set to 0.25 in order to generally determine that the case where the amplitude of the respiratory waveform is 0.5 times or less than that of normal breathing is hypopnea in obstructive sleep apnea syndrome. is there. In order to make the amplitude correspond to the spectrum value, a value of 0.5 square is used. When the threshold value is below the threshold for judgment, not only hypopnea of obstructive sleep apnea syndrome but also apnea of central sleep apnea syndrome is included.

  Here, in the respiratory intensity time series data, for example, when the sleeping phase of the sleeping person changes, the signal level of the respiratory intensity may change greatly. On the other hand, in this embodiment, as described above, the process of following the waveform of the time-series data of the respiratory signal with a larger change rate than when the respiratory signal increases in each of the forward direction and the reverse direction decreases. In each time segment, the smaller one of the time-series data for the forward process and the time-series data for the backward process is used as the comparison process value. For this reason, as shown in FIG. 8A, the waveform of the time-series data of the comparative processing values is a waveform that follows with little difference even at locations where the signal level of the respiratory intensity changes greatly. The waveform almost follows the time series data. Therefore, the apnea state can be accurately estimated even immediately after the sleep phase has changed greatly.

  In addition, the respiratory intensity time-series data may include noise data that is not caused by respiration, such as door opening and closing. Of the noise data, data having a relatively steep mountain-shaped waveform compared to before and after as shown in FIG. 8B may affect the estimation of the apnea state. In this case, the time-series data for the forward processing and the time-series data for the backward processing increase rapidly to the vicinity of the extreme value of the respiratory intensity when viewed from each processing direction, but gradually decrease when the peak is exceeded. For this reason, as shown in FIG. 8B, the extreme value of the noise data portion of the waveform of the time-series data of the comparison processing values composed of the smaller data is greatly reduced.

  Further, it is known that the respiratory signal is hardly detected in the central sleep apnea syndrome, and the signal level of the respiratory signal is significantly reduced in the obstructive sleep apnea syndrome. For this reason, as shown in FIG. 8C, the data of the apnea state portion in the respiratory intensity time-series data has a valley waveform that is relatively steep compared to before and after. In this case, the time-series data for the forward direction processing and the time-series data for the backward direction processing are hardly changed because the rate of change when the respiratory intensity decreases in each processing direction is small. For this reason, as shown in FIG. 8C, since the valley-shaped waveform resulting from the apnea state is almost deleted from the waveform of the time-series data of the comparative processing value, By comparing, an apnea state is estimated.

  Next, the apnea estimation means (61) determines whether or not there is a sleeping person on the bedding for the time segment determined that there is a possibility of an apnea state. The presence determination is performed, for example, by comparing a signal related to body movement created by the body movement data creation unit (44) with a threshold for floor presence determination. And when it determines with being in bed, it transfers to step ST16 as what may be an apnea state, and when it determines with being absent, it transfers to step ST19. Thereby, it can be avoided that the sleeper is determined to be in an apnea state although the sleeper is not on the bedding.

  Next, the apnea estimation means (61) uses the detection result of the rough body motion detection means (40) to check whether or not there is a rough body motion for a time segment in which there is a possibility of an apnea state. The apnea estimation means (61) determines that it is not an apnea state when there is a rough body movement, and moves to step ST19. If there is no rough body movement, it moves to step ST17. Thereby, it can be avoided that the time segment in which the rough body motion has occurred is determined to be an apnea state.

  Next, the apnea estimation means (61) confirms whether or not the time during which there is a possibility of an apnea state continues for 120 seconds or more. Apnea estimation means (61), when there is a possibility of an apnea for 120 seconds (when it is determined that there is a possibility of an apnea continuously in 12 or more time segments) ) Goes to step ST19, and if it is less than 120 seconds, it goes to step ST18 and finally determines an apnea state. Since it is said that the apnea state in sleep apnea syndrome is less than 120 seconds, it can be avoided that the time determined as the apnea state due to a factor other than the apnea state is determined as the apnea state. .

  The apnea estimation means (61) displays the time determined to be an apnea state on a display unit such as a monitor. Thereby, the sleeper can confirm the generation | occurrence | production time, if it is an apnea state after waking up.

-Effect of the embodiment-
In the above embodiment, the process of following the waveform of the time series data of the respiratory intensity at a larger rate of change than the case where the respiratory intensity increases in each of the forward direction and the reverse direction is performed to perform the forward direction process. Create time-series data and time-series data for reverse processing, and compare the time-series data for respiration intensity at the same time for the smaller time-series data for forward-direction processing and time-series data for reverse-direction processing Therefore, if an apnea condition occurs during sleep, there is no difference between the comparative processing value of the smaller time-series data in the forward processing and the time-series data in the reverse processing and the respiratory intensity. A respiratory condition is inferred.

  Then, by performing the processing as described above, the comparison processing value for estimating the apnea state compared with the time-series data of the respiratory intensity is a change in the signal level of the respiratory intensity due to the change in the sleep phase of the sleeper. There is almost no difference in the waveform, and the waveform follows, and the waveform of noise data not caused by respiratory motion is relaxed. Therefore, it is possible to accurately estimate the apnea state by avoiding misjudgment due to a change in sleep phase or coarse body movement, and it is possible to improve the estimation accuracy of the apnea state and the reliability of the estimation result.

  Moreover, in the said embodiment, since it is known that the sleeper of an apnea state does not produce a rough body motion, it is determined that the time when the rough body motion is detected is not an apnea state. Accordingly, the apnea state can be estimated more accurately, and the apnea state estimation accuracy and the reliability of the estimation result can be improved.

-Modification of the embodiment-
A modification of the embodiment will be described. The apnea estimation device (10) of this modification is different from the above embodiment in the configuration of the respiratory signal extraction means (50).

  Specifically, the respiration signal extraction means (50) is configured only by a respiration intensity calculation unit (53). The respiratory intensity calculation unit (53) creates the frequency spectrum of the body motion signal of each time segment divided every 10 seconds from the start of measurement based on the body motion signal from the sleep sensor (20), and the natural vibration of the respiratory motion The maximum spectral value in the band (0.25 ± 0.15 Hz) corresponding to the band is extracted as the respiratory intensity.

<< Other Embodiments >>
About the said embodiment, it is good also as the following structures.

  In the above embodiment, the spectrum value converted from the time-series data of the body motion signal is extracted as the respiratory intensity, but the voltage value (amplitude) corresponding to the respiratory signal in the body motion signal without being converted into the spectrum value. May be extracted as the respiratory intensity. In this case, the threshold for determination is the square root of the value when the spectrum value is the respiratory intensity.

  In addition, the above embodiment is an essentially preferable illustration, Comprising: It does not intend restrict | limiting the range of this invention, its application thing, or its use.

  As described above, the present invention is useful for an apnea estimation device that detects an occurrence time of an apnea state of a sleeper while sleeping.

1 is an overall configuration diagram of an apnea estimation device of an embodiment. It is a schematic sectional drawing of the sleep sensor of embodiment. It is a block diagram of the circuit unit of an embodiment. It is a flowchart which shows operation | movement of the rough body motion detection means and the respiration signal extraction means of the apnea estimation apparatus of embodiment. It is a flowchart which shows the operation | movement of the respiration signal processing means of the apnea estimation apparatus of embodiment, and the apnea estimation means. It is a chart showing the time series data of the body movement detected by the rough body movement detection means of the embodiment and the time series data of the body movement threshold. It is a chart showing the time series data of the forward direction process and the time series data of the backward process created by the respiratory signal processing means of the embodiment. It is a figure for demonstrating the reference line which consists of the smaller data of the time series data of a forward direction process, and the time series data of a reverse direction process.

10 Apnea estimation device
20 Sleep sensor (body motion detection means)
40 Coarse body motion detection means
50 Respiratory signal extraction means
60 Respiratory signal processing means
61 Apnea estimation

Claims (3)

Body motion detecting means (20) for detecting a body motion of the sleeping person and outputting a body motion signal is provided, and the sleeper's apnea is based on the time series data of the body motion signal from the body motion detecting means (20). An apnea estimation device for estimating a state,
Breathing signal extraction means (50) for extracting the time series data of the respiratory signal generated by the sleep movement of the sleeping person from the time series data of the body motion signal;
In the time series data of the respiratory signal, the forward direction in which the time advances and the reverse direction in which the time goes back are set as the processing directions, and the case where the respiratory signal increases in each of the forward direction and the reverse direction is larger than the case where it decreases. Respiration signal processing means (60) for generating time-series data for forward processing and time-series data for backward processing by performing processing to follow the waveform of the time-series data of the respiratory signal at a rate of change;
Apnea estimation means for comparing the time series data of the forward processing and the time series data of the backward processing with the time series data of the respiratory signal at each time point to estimate the apnea state (61) And an apnea estimation device. In claim 1,
The apnea estimation means (61) estimates an apnea state based on a ratio of time series data of the respiratory signal to a smaller one of the time series data of the forward processing and the time series data of the backward processing. An apnea estimation device characterized by. In claim 1 or 2,
The body motion detection means (20) includes a rough body motion detection means (40) for detecting the rough body motion of the sleeping person based on the body motion signal detected by the body motion detection means (20),
The apnea estimation unit (61) determines that the time when the coarse body motion detection unit (40) detects the coarse body motion of the sleeper is not in an apnea state.

Images