Biological information monitoring system, biological information monitoring method, and program

Abstract

To more accurately measure subject's biological information.SOLUTION: A biological information monitoring system 1 includes a data acquisition unit 11a and a signal analysis unit 11c. The data acquisition unit 11a is provided at a member for supporting the subject and, from a plurality of sensors for detecting dynamic physical quantities (for example, weight), acquires signals showing the physical quantities. The signal analysis unit 11c acquires biological information with respect to a heartbeat of the subject supported by the member on the basis of peaks of the signals showing the physical quantities acquired by the data acquisition unit 11a.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a biological information monitoring system, a biological information monitoring method, and a program for monitoring biological information of a subject. [Background technology]

[0002] Heart disease, which occurs in the heart and is accompanied by abnormal fluctuations in heart rate, is the second leading cause of death in Japan, and is characterized by its unconscious progression due to poor lifestyle habits and a high likelihood of occurring while sleeping at night. Due to these characteristics, approximately 70% of heart disease cases occur at home. It is known that if 10 minutes pass before treatment after cardiac arrest due to heart disease, the chances of survival drop to approximately 20%. In contrast, statistics show that the average arrival time for an ambulance in Japan is about seven minutes, so early detection of cardiac arrest is essential to saving lives. Furthermore, because signs such as abnormal heartbeat intervals often appear before cardiac arrest, it is important to accurately measure heartbeats in order to prevent serious conditions caused by heart disease. A technique for measuring biological information (heart rate, etc.) of a subject is described in, for example, Patent Document 1. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 2018-126511 Summary of the Invention [Problem to be solved by the invention]

[0004] However, in conventional technologies for measuring biometric information (such as heart rate), including the technology described in Patent Document 1, devices are placed on the subject's body (e.g., arm or torso) to measure biometric information such as heart rate, and there is a possibility that measurement accuracy may decrease if the subject moves their body, etc.

[0005] An object of the present invention is to measure biological information of a subject with higher accuracy. [Means for solving the problem]

[0006] In order to achieve the above object, a biological information monitoring system according to one aspect of the present invention comprises: a physical quantity acquiring means that is installed on a member supporting the subject and that acquires signals indicating mechanical physical quantities from a plurality of sensors that detect the physical quantities; a signal analysis means for acquiring biological information relating to the heartbeat of the subject supported by the member based on peaks of the signal indicating the physical quantity acquired by the physical quantity acquisition means; The present invention is characterized by comprising: [Effects of the Invention]

[0007] According to the present invention, biological information of a subject can be measured with higher accuracy. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a schematic diagram showing an example of the system configuration of a biological information monitoring system 1 according to a first embodiment. [Figure 2] 1 is a schematic diagram showing an example of an embodiment of a biological information monitoring system 1. FIG. [Figure 3] FIG. 10 is a schematic diagram showing the feature amounts of peaks. [Figure 4] FIG. 10 is a schematic diagram showing an example of the results of classification of peak feature amounts into three classes by k-means clustering. [Figure 5] 5 is a schematic diagram showing the results of identifying heartbeat peak data corresponding to the classification results shown in FIG. 4. FIG. [Figure 6] 10 is a flowchart illustrating the flow of a heartbeat interval estimation process executed by the biological information monitoring system 1. [Figure 7] Schematic diagrams showing examples of sleeping positions of subjects, in which (A) is supine position, (B) is prone position, (C) is lateral position (right half of body up), (D) is lateral position (left half of body up), (E) flexed lateral position (right half of body up), (F) flexed lateral position (left half of body up), and (G) is semi-sitting position. [Figure 8] FIG. 1 is a schematic diagram showing an example of BCG data in which a subject is in a lateral position (right side up), in which peak features are classified into three classes by k-means clustering. [Figure 9] 9 is a schematic diagram showing the results of identifying heartbeat peak data corresponding to the classification results shown in FIG. 8. FIG. [Figure 10] FIG. 10 is a schematic diagram showing the results of tentative peak estimation in a signal identified as a heartbeat component. [Figure 11] FIG. 10 is a schematic diagram illustrating the concept of generating a template for identifying the peak of a heartbeat. [Figure 12] FIG. 10 is a schematic diagram showing a cross-correlation function between a generated template and a signal identified as a heartbeat component. [Figure 13] FIG. 1 is a schematic diagram illustrating the concept of identifying a peak in a cross-correlation function. [Figure 14] FIG. 14 is a schematic diagram showing peak intervals (heartbeat intervals) identified from the cross-correlation function shown in FIG. 13 and peak intervals (heartbeat intervals) of a gold standard (standard reference) in a signal waveform corresponding to the cross-correlation function. [Figure 15] 10 is a flowchart illustrating the flow of a heartbeat interval estimation process executed by the biological information monitoring system 1. [Figure 16] FIG. 10 is a diagram showing the results of a comparison between the accuracy rate of heart rate peaks estimated from various signals and the heart rate peaks estimated by the method of the present embodiment. [Figure 17] FIG. 10 is a schematic diagram showing an extracted analysis section (two periods of a waveform). [Figure 18]10 is a flowchart illustrating the flow of a heartbeat interval estimation process executed by the biological information monitoring system 1. DETAILED DESCRIPTION OF THE INVENTION

[0009] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. [First embodiment] In the biological information monitoring system of this embodiment, a sensor (e.g., a load sensor) is attached to the leg of a bed used by a subject while sleeping, and an output signal from the sensor is acquired. The acquired output signal from the sensor indicates data from the subject's ballistocardiogram (BCG: Ballist CardioGram). The acquired BCG data is then subjected to signal processing to acquire heart rate variability (RRI: RR Interval) and determine whether or not the subject has a heart disease. This makes it possible to acquire biometric information (heart rate variability) robustly against the subject's movements, posture, etc., based on signals detected at the legs of the bed, without having to install any equipment on the subject's body. Therefore, according to the biological information monitoring system of this embodiment, the biological information of the subject can be measured with higher accuracy.

[0010] [System Configuration] Fig. 1 is a schematic diagram showing an example of the system configuration of a biological information monitoring system 1 according to this embodiment, and Fig. 2 is a schematic diagram showing an example of an embodiment of the biological information monitoring system 1. The biological information monitoring system 1 is configured by an information processing device such as a PC (Personal Computer) or a tablet terminal. As shown in FIG. 1, the biological information monitoring system 1 includes a CPU (Central Processing Unit) 11, a ROM (Read Only Memory) 12, a RAM (Random Access Memory) 13, an input unit 14, a display unit 15, a memory unit 16, a communication unit 17, sensor units 18a to 18d, and an imaging unit 19. As shown in FIG. 2, the biological information monitoring system 1 also has sensors 18a to 18d attached to the legs of the bed used by the subject, and acquires signals representing the detection results of the load on each leg of the bed.

[0011] The CPU 11 executes various programs stored in the storage unit 16 to control the entire biological information monitoring system 1. For example, the CPU 11 processes output signals from the sensors 18a to 18d installed on the legs of the bed and executes a program for processing to estimate heartbeat intervals (hereinafter referred to as "heartbeat interval estimation processing").

[0012] By executing the program for the heartbeat interval estimation process, the CPU 11 is formed with the following functional components: a data acquisition unit 11a, a preprocessing unit 11b, and a signal analysis unit 11c.

[0013] The data acquiring unit 11a acquires (receives) output signals from the sensor units 18a to 18d, and stores the acquired output signal data in association with time in the storage unit 16. That is, the data acquiring unit 11a stores BCG (ballistocardiogram) data represented by the output signals from the sensor units 18a to 18d in chronological order in the storage unit 16. The pre-processing unit 11b performs pre-processing on the waveforms of the output signals of the sensors 18a to 18d in order to perform analysis in the signal analysis unit 11c. In this embodiment, the pre-processing unit 11b performs filtering on the waveforms of the output signals of the sensors 18a to 18d using a band-pass filter (specifically, a band-pass filter that passes signal waveforms of 1 to 8.5 Hz). Furthermore, the pre-processing unit 11b normalizes the signals after filtering. The normalized signals have a uniform center of amplitude, making the waveform easier to analyze. Furthermore, performing processing using the band-pass filter has the effect of removing noise from the output signals of the sensors 18a to 18d.

[0014] The signal analysis unit 11c identifies peaks in the waveforms of the output signals of the sensor units 18a to 18d. At this time, the signal analysis unit 11c identifies all peaks (here, only positive peaks) in the waveforms of the output signals of the sensor units 18a to 18d processed by the preprocessing unit 11b. Furthermore, the signal analysis unit 11c extracts the feature amount of each identified peak.

[0015] FIG. 3 is a schematic diagram showing the feature quantities of the peaks. As shown in FIG. 3, the signal analysis unit 11c extracts, as features, the magnitude of the amplitude from the negative maximum value immediately before the peak to the peak (feature 1), the magnitude of the negative maximum value immediately before the peak (feature 2), the amplitude of the peak (feature 3), the time from the negative maximum value immediately before the peak to the peak (feature 4), and the time from the peak to the negative maximum value immediately after the peak (feature 5). Furthermore, the signal analysis unit 11c classifies the feature amounts by k-means clustering. At this time, the signal analysis unit 11c classifies the feature amounts into three classes and identifies data of heartbeat peaks and data other than heartbeat peaks.

[0016] Fig. 4 is a schematic diagram showing an example of the results of classifying peak feature quantities into three classes by k-means clustering. Fig. 5 is a schematic diagram showing the results of identifying heartbeat peak data corresponding to the classification results shown in Fig. 4. In Fig. 5, the points identified as peaks are indicated by circles. FIG. 4 shows the results of k-means clustering performed on BCG data of a subject lying on his back on a bed (supine position). As shown in Figure 4, as a result of k-means clustering, the data group in the upper left (data in the area surrounded by a solid line) represents one classification, and the data belonging to this classification corresponds to the heart rate peak data in Figure 5. That is, by extracting the feature values ​​of waveform peaks from BCG data and classifying them using k-means clustering, it is possible to identify the heartbeat peaks that represent the heartbeat intervals.

[0017] Returning to FIG. 1, the ROM 12 has various system programs for controlling the biological information monitoring system 1 written therein in advance. The RAM 13 is configured by a semiconductor memory such as a DRAM (Dynamic Random Access Memory), and stores data generated when the CPU 11 executes various processes. The input unit 14 is configured with input devices such as a keyboard, a mouse, or a touch sensor (touch panel), and accepts various pieces of information input to the biological information monitoring system 1 by a user.

[0018] The display unit 15 is configured by a display device such as an LCD (Liquid Crystal Display), and displays various processing results of the biological information monitoring system 1.

[0019] The storage unit 16 is configured with a nonvolatile storage device such as a hard disk or flash memory, and stores programs for the heartbeat interval estimation process, etc. The storage unit 16 also stores various processing results of the biological information monitoring system 1 (data on the heartbeat interval estimation result, etc.).

[0020] The communication unit 17 has a communication interface that performs signal processing based on a predetermined communication standard, such as a wired or wireless LAN (Local Area Network) or USB (Universal Serial Bus), and controls communication between the biological information monitoring system 1 and other devices.

[0021] The sensor units 18a to 18d are sensors installed on the legs of the bed used by the subject, and in this embodiment, are configured as load sensors that detect the force applied to the legs of the bed. However, sensors that detect things other than force, such as acceleration sensors, can also be used as long as they can detect the mechanical influence (physical quantity) caused by the subject's heartbeat, such as force or vibration generated on the legs of the bed. In this embodiment, when estimating the heartbeat interval from the output signals of the sensor units 18a to 18d, it is possible to perform the estimation by focusing on the output signal of any one of the sensor units 18a to 18d, or to extract the most reliable output signal from all of the output signals of the sensor units 18a to 18d and perform the estimation. The imaging unit 19 is configured by an imaging device equipped with a lens, an imaging element, etc., and captures a digital image of the subject. In this embodiment, the imaging unit 19 may detect the sleeping posture of the subject, thereby determining whether or not the situation is such that the output signals of the sensor units 18a to 18d are likely to be disturbed.

[0022] [Operation] Next, the operation of the biological information monitoring system 1 will be described. FIG. 6 is a flowchart illustrating the flow of the heartbeat interval estimation process executed by the biological information monitoring system 1. The heartbeat interval estimation process is started in response to an instruction to execute the heartbeat interval estimation process being input via the input unit 14.

[0023] When the heartbeat interval estimation process is started, in step S1, the data acquisition unit 11a acquires (receives) output signals from the sensors 18a to 18d. The acquired data (BCG data) of the output signals from the sensors 18a to 18d is stored in the storage unit 16 in association with time. In step S2, the preprocessing unit 11b performs preprocessing on the waveforms of the output signals of the sensor units 18a to 18d in order to perform analysis in the signal analysis unit 11c. Specifically, the preprocessing unit 11b performs filtering on the waveforms of the output signals of the sensor units 18a to 18d using a bandpass filter (a bandpass filter that passes signal waveforms of 1 to 8.5 [Hz]).

[0024] In step S3, the signal analysis unit 11c identifies peaks in the waveforms of the output signals of the sensor units 18a to 18d. In step S4, the signal analysis unit 11c extracts the feature amount of each identified peak. In step S5, the signal analysis unit 11c classifies the feature amounts by k-means clustering. At this time, the signal analysis unit 11c classifies the feature amounts into three classes and identifies heartbeat peak data and data other than heartbeat peaks. By identifying the heartbeat peak data, the heartbeat interval can be estimated. The processing results in step S5 (identified heartbeat peak data or heartbeat interval data, etc.) are stored in the storage unit 16. After step S5, the heartbeat interval estimation process ends. By performing such processing, it becomes possible to remove noise from the BCG data and to appropriately identify the peak of the heartbeat.

[0025] [Verification of effectiveness] Next, the effects of this embodiment will be verified. Figure 7 is a schematic diagram showing examples of sleeping positions of the subject, in which (A) is the supine position, (B) is the prone position, (C) is the lateral position (right half of the body up), (D) is the lateral position (left half of the body up), (E) is the flexed lateral position (right half of the body up), (F) is the flexed lateral position (left half of the body up), and (G) is the semi-sitting position. As shown in FIG. 7, the posture of the subject while sleeping can vary in various ways, and the state in which the subject is resting in a certain posture or in which the subject is moving the body also varies. In the heartbeat interval estimation process of this embodiment, for BCG data of a subject resting in a supine position, the heartbeat peaks could be estimated with a certain degree of reliability, as shown in FIG. On the other hand, it was found that when the waveform of the BCG data was relatively distorted, the accuracy of estimating the heartbeat interval decreased slightly.

[0026] Fig. 8 is a schematic diagram showing an example of a state in which peak feature amounts are classified into three classes by k-means clustering in BCG data in which the subject is in a lateral position (right side up). Fig. 9 is a schematic diagram showing the results of identifying heartbeat peak data corresponding to the classification results shown in Fig. 8. In Fig. 9, the points identified as peaks are indicated by circles. The BCG data shown in FIG. 8 has relatively large waveform disturbances, and the classification results in this case vary widely, as shown in the data group on the lower right (data in the area surrounded by a solid line). In this case, as shown in FIG. 9, the accuracy of the result of identifying the heartbeat peak data is reduced. In other words, when using the heartbeat interval estimation process of this embodiment, it is effective to select a section where the waveform of the BCG data is stable (data when the subject is at rest) and estimate the peak of the heartbeat. Therefore, the biological information monitoring system 1 of this embodiment can measure the biological information of the subject with higher accuracy.

[0027] [Second embodiment] Next, a biological information monitoring system 1 according to a second embodiment of the present invention will be described. The system configuration of a biological information monitoring system 1 according to this embodiment is almost the same as the system configuration shown in FIG. 1 of the first embodiment. However, in the biological information monitoring system 1 of this embodiment, the configuration of the signal analysis unit 11c is different from that of the first embodiment, and therefore, the following description will mainly focus on the signal analysis unit 11c.

[0028] The signal analysis unit 11c performs signal separation using independent component analysis (ICA) on the waveforms of the output signals from the sensors 18a to 18d processed by the preprocessing unit 11b. As a result, heartbeat components and noise components are separated. The signal analysis unit 11c then identifies the separated signal with the smallest kurtosis as the heartbeat component and uses it to estimate the heartbeat interval. Furthermore, the signal analysis unit 11c detects peaks that are spaced apart by a set time (for example, 0.7 [s]) or more in the signal identified as the heartbeat component, and sets these as provisional peak estimation results.

[0029] 10 is a schematic diagram showing the results of tentative peak estimation in a signal identified as a heartbeat component. In FIG. 10, the points estimated as tentative peaks are indicated by circles. Then, the signal analysis unit 11c generates a template for identifying the heartbeat peak based on the estimated tentative peak in the signal identified as the heartbeat component. Specifically, the signal analysis unit 11c extracts signals from a set time (for example, 0.4 seconds) before and after the estimated tentative peak, and performs averaging on the extracted signals. By performing averaging, it is possible to remove noise components and generate a typical waveform.

[0030] FIG. 11 is a schematic diagram showing the concept of generating a template for identifying the peak of a heartbeat. As shown in Fig. 11, when a template for identifying a heartbeat peak is generated, signals are extracted from a signal identified as a heartbeat component, with a set time (e.g., 0.4 s) before and after the estimated tentative peak at the center. Then, a single signal is generated by averaging the extracted signals. This signal exhibits a typical waveform of the extracted signals.

[0031] Next, the signal analysis unit 11c uses the generated template to perform template matching on the signal identified as the heartbeat component, and identifies the peak of the heartbeat. Specifically, the signal analysis unit 11c calculates a cross-correlation function between the generated template and the signal identified as the heartbeat component.

[0032] FIG. 12 is a schematic diagram showing a cross-correlation function between the generated template and a signal identified as a heartbeat component. The cross-correlation function shown in FIG. 12 exhibits a larger value as the degree of match between the generated template and the signal identified as the heartbeat component increases. Furthermore, the signal analysis unit 11c regards the peaks in the cross-correlation function as peaks of heartbeats, and calculates (estimates) the heartbeat interval.

[0033] Fig. 13 is a schematic diagram showing the concept of identifying peaks in a cross-correlation function. Fig. 14 is a schematic diagram showing peak intervals (heartbeat intervals) identified from the cross-correlation function shown in Fig. 13 and peak intervals (heartbeat intervals) of a gold standard (standard reference) in a signal waveform corresponding to the cross-correlation function. In Fig. 13, the points identified as peaks are indicated by circles. As shown in Figure 14, it can be seen that the peak intervals (heartbeat intervals) identified from the cross-correlation function shown in Figure 13 coincide to a certain extent with the peak intervals (heartbeat intervals) of the gold standard (standard reference) in the signal waveform corresponding to the cross-correlation function.

[0034] [Operation] Next, the operation of the biological information monitoring system 1 will be described. FIG. 15 is a flowchart illustrating the flow of the heartbeat interval estimation process executed by the biological information monitoring system 1. The heartbeat interval estimation process is started in response to an instruction to execute the heartbeat interval estimation process being input via the input unit 14.

[0035] When the heartbeat interval estimation process is started, in step S11, the data acquisition unit 11a acquires (receives) output signals from the sensors 18a to 18d. The acquired data (BCG data) of the output signals from the sensors 18a to 18d is stored in the storage unit 16 in association with time. In step S12, the preprocessing unit 11b performs preprocessing on the waveforms of the output signals of the sensor units 18a to 18d in order to perform analysis in the signal analysis unit 11c. Specifically, the preprocessing unit 11b performs filtering on the waveforms of the output signals of the sensor units 18a to 18d using a bandpass filter (a bandpass filter that passes signal waveforms of 1 to 8.5 [Hz]).

[0036] In step S13, the signal analysis unit 11c performs signal separation by independent component analysis (ICA) on the waveforms of the output signals from the sensors 18a to 18d processed by the preprocessing unit 11b. At this time, among the separated signals, the one with the smallest kurtosis is identified as the heartbeat component and used to estimate the heartbeat interval. In step S14, the signal analysis unit 11c detects peaks that are spaced apart by a set time (for example, 0.7 [s]) or more in the signal identified as the heartbeat component, and sets these as provisional peak estimation results. In step S15, the signal analysis unit 11c generates a template for identifying the heartbeat peak based on the estimated tentative peak in the signal identified as the heartbeat component.

[0037] In step S16, the signal analysis unit 11c performs template matching on the signal identified as the heartbeat component using the generated template. At this time, the signal analysis unit 11c calculates a cross-correlation function between the generated template and the signal identified as the heartbeat component. In step S17, the signal analysis unit 11c identifies the peak of the heartbeat by determining the peak in the cross-correlation function as the peak of the heartbeat. In step S18, the signal analysis unit 11c calculates (estimates) the heartbeat interval from the identified heartbeat peak. The processing results in step S18 (data on the identified heartbeat peak or data on the heartbeat interval, etc.) are stored in the storage unit 16. After step S18, the heartbeat interval estimation process ends.

[0038] This allows a template for identifying the heartbeat peak to be generated from the heartbeat component signal obtained by signal separation using independent component analysis (ICA), and the heartbeat peak to be estimated based on the cross-correlation function between the generated template and the signal identified as the heartbeat component.

[0039] [Verification of effectiveness] Next, the effects of this embodiment will be verified. Fig. 16 is a diagram showing the results of comparing the accuracy rate of heart rate peaks estimated from various signals with the heart rate peaks estimated by the method of this embodiment. Fig. 16 shows the results of comparing data from 39 subjects. The accuracy rate of heart rate peaks indicates the percentage of estimated heart rate peaks that match the correct heart rate peaks, where the correct heart rate peaks are those identified by directly measuring the pulse wave on the subject's body. As shown in Figure 16, when comparing signals, the results estimated by independent component analysis (ICA) (accuracy rate of approximately 80%) show a higher accuracy rate than the results estimated from other signals.

[0040] Furthermore, in comparing sleeping positions, except for the results estimated by independent component analysis (ICA), the accuracy rate was lower for the lateral position (right half of the body up) and the lateral position (left half of the body up) compared to the supine position, but the results estimated by independent component analysis (ICA) showed the same accuracy rate for all sleeping positions. In other words, it can be seen that the results estimated by independent component analysis (ICA) corrected for waveform disturbances and were able to estimate the heart rate peak with higher accuracy. Furthermore, in tests conducted with different sleeping positions and body orientations in bed, the estimation method of this embodiment achieved an accuracy rate of approximately 80%, demonstrating that the heart rate peak can be estimated with greater accuracy. Therefore, the biological information monitoring system 1 of this embodiment can measure the biological information of the subject with higher accuracy.

[0041] [Third embodiment] Next, a biological information monitoring system 1 according to a third embodiment of the present invention will be described. The system configuration of a biological information monitoring system 1 according to this embodiment is almost the same as the system configuration shown in FIG. 1 of the first embodiment. However, in the biological information monitoring system 1 of this embodiment, the configurations of the data acquisition unit 11a, pre-processing unit 11b, and signal analysis unit 11c are different from those of the first embodiment, so the following will mainly describe the data acquisition unit 11a, pre-processing unit 11b, and signal analysis unit 11c.

[0042] The data acquiring unit 11a acquires (receives) the output signals of the sensors 18a to 18d and inverts the phase of one of the output signals of the sensors installed on the diagonally opposite legs of the bed. Here, the output signals of the sensors 18c and 18d are inverted in phase. The data acquiring unit 11a then stores the output signals of the sensors 18a and 18b and the phase-inverted output signals of the sensors 18c and 18d in the storage unit 16 in association with time. The output signals of the sensors installed on the legs of the bed tend to be phase-inverted relative to the output signals of the sensors installed at diagonally opposite positions on the bed. Therefore, in this embodiment, the output signals of the sensors 18c and 18d are phase-inverted and stored in a state where their correlation with the output signals of the sensors 18a and 18b is enhanced.

[0043] The pre-processing unit 11b performs pre-processing on the waveforms of the output signals from the sensors 18a to 18d to prepare for analysis in the signal analysis unit 11c. In this embodiment, the pre-processing unit 11b performs filtering on the waveforms of the output signals from the sensors 18a to 18d using a band-pass filter (specifically, a band-pass filter that passes signal waveforms of 3 to 9 Hz). Furthermore, the pre-processing unit 11b performs standardization on the filtered signals (here, standardization using a Z-score). By performing standardization, a threshold can be set and data in sections that include fluctuations due to body movement can be removed.

[0044] The signal analysis unit 11c extracts an analysis section from the waveforms of the output signals of the sensors 18a to 18d. At this time, the signal analysis unit 11c extracts an analysis section w i Extract [v] (i is an integer) and extract the analysis interval by shifting the center of the interval by Δt = 0.2 [s]. Here, v represents the number of samples (time element), and wi [v] represents the i-th interval signal in the signal w[v] expressed as a function of sample v.

[0045] FIG. 17 is a schematic diagram showing the extracted analysis section (two periods of the waveform). In Figure 17, the horizontal axis represents the number of samples v, and the standardized waveform w i The vertical axis shows the state where two periods, each consisting of 300 samples, are extracted. Furthermore, the signal analysis unit 11c estimates the heartbeat interval from the signal in the extracted analysis section. Specifically, the signal analysis unit 11c estimates the appearance interval of similar waveform patterns (pitch tracking).

[0046] In pitch tracking, the degree of correlation is obtained assuming a predetermined period N. Specifically, in pitch tracking, a correlation index is calculated to estimate the appearance interval of similar waveform patterns. Note that a minimum period and a maximum period are set in the calculation range of the correlation index. For example, the minimum period can be set to 0.6 [s] (120 samples), and the maximum period can be set to 1.5 [s] (300 samples).

[0047] As correlation indices for estimating the appearance intervals of similar waveform patterns, for example, the following three indices can be used. (1) Autocorrelation function of the signal The signal analysis unit 11c calculates the autocorrelation function of the signal in the extracted analysis interval as a correlation index. Corr [N] can be expressed by the following formula (1).

number

[0048] (2) Average amplitude difference The signal analysis unit 11c calculates the average of the amplitude differences in the signal in the extracted analysis section as a correlation index. The average of the amplitude differences S assuming a predetermined period N AMDF[N] can be expressed by the following formula (2).

number

[0049] (3) Maximum sum of amplitudes The signal analysis unit 11c calculates the maximum value of the sum of amplitudes of the extracted signal in the analysis interval as a correlation index. In this case, first, the autoentropy w of the signal obtained by first differentiation is calculated. se [n] is calculated. The autoentropy w of the first derivative signal se [n] can be expressed by the following equation (3).

number

[0050] Next, the signal analysis unit 11c calculates the autoentropy w of the first-order differentiated signal. se Peak extraction and smoothing processing is performed on [n]. Peak extraction emphasizes the timing of the heartbeat and obtains the maximum amplitude in the entropized signal.

[0051] Furthermore, the signal analysis unit 11c calculates the maximum value w of the sum of amplitudes assuming a predetermined period N. se Calculate the maximum value S of the sum of amplitudes assuming a given period N. MAP [N] can be expressed by the following equation (4).

number

[0052] The signal analysis unit 11c then integrates the three calculated correlation indices. Specifically, the signal analysis unit 11c calculates the probability that the assumed predetermined period N is the heartbeat interval, based on the three calculated correlation indices. That is, the signal analysis unit 11c calculates the autocorrelation function S Corr [N], mean amplitude difference S AMDF[N] and the maximum value of the sum of the amplitudes S MAP The posterior probability density function is converted using [N] as an element. Here, the most likely N according to Bayes' theorem is calculated using the following equation (5).

number

[0053] Next, the signal analysis unit 11c performs independent component analysis (ICA) to obtain the autocorrelation function S Corr [N], mean amplitude difference S AMDF [N] and the maximum value of the sum of the amplitudes S MAP The conditional probability p(N|S Corr ), p(N|S AMDF ) and p(N|S MAP ) multidimensional data N i As a result, multidimensional data N i The signal analysis unit 11c separates the intervals with low accuracy from the intervals with high accuracy. Then, the signal analysis unit 11c estimates the heartbeat intervals again in the intervals with low accuracy.

[0054] The signal analysis unit 11c then integrates the previous and new estimates of the heartbeat interval and adopts the period N that provides the greatest accuracy as the heartbeat interval estimate, based on the J wave position on the electrocardiogram determined by the maximum amplitude sum.

[0055] [Operation] Next, the operation of the biological information monitoring system 1 will be described. FIG. 18 is a flowchart illustrating the flow of the heartbeat interval estimation process executed by the biological information monitoring system 1. The heartbeat interval estimation process is started in response to an instruction to execute the heartbeat interval estimation process being input via the input unit 14.

[0056] When the heartbeat interval estimation process is started, in step S21, the data acquisition unit 11a acquires (receives) output signals from the sensors 18a to 18d and inverts the phase of one of the output signals from the sensors installed on the legs diagonally positioned on the bed. Here, the output signals from the sensors 18c and 18d are inverted in phase. Data on the output signals from the sensors 18a and 18b and the phase-inverted output signals from the sensors 18c and 18d are stored in the storage unit 16 in association with time.

[0057] In step S22, the preprocessing unit 11b performs preprocessing on the waveforms of the output signals from the sensor units 18a to 18d to prepare for analysis in the signal analysis unit 11c. Specifically, the preprocessing unit 11b performs filtering on the waveforms of the output signals from the sensor units 18a to 18d using a bandpass filter (a bandpass filter that passes signal waveforms of 3 to 9 Hz). Furthermore, the preprocessing unit 11b performs standardization processing (standardization processing using Z-score) on the signals after filtering. A threshold is set on the signals after standardization processing to remove data in sections that include fluctuations due to body movement. In step S23, the signal analysis unit 11c extracts an analysis section in the waveforms of the output signals of the sensors 18a to 18d.

[0058] In step S24, the signal analysis unit 11c estimates the heartbeat interval from the signal in the extracted analysis section. Specifically, the signal analysis unit 11c estimates the appearance interval of similar waveform patterns (pitch tracking). In pitch tracking, a correlation index (autocorrelation function S Corr [N], mean amplitude difference S AMDF [N] and the maximum value of the sum of the amplitudes S MAP [N]) is calculated, and using these as elements, the most likely N is calculated according to Bayes' theorem (see equation (5)).

[0059] In step S25, the signal analysis unit 11c corrects the estimated result of the heartbeat interval for a section where the accuracy of the estimated result is low (where the probability that N represents the heartbeat interval is lower than a predetermined threshold). That is, the signal analysis unit 11c performs an independent component analysis (ICA) to calculate the autocorrelation function S Corr [N], mean amplitude difference S AMDF [N] and the maximum value of the sum of the amplitudes S MAP The conditional probability p(N|S Corr ), p(N|S AMDF ) and p(N|S MAP ) multidimensional data N i As a result, multidimensional data N i The signal analysis unit 11c separates low-accuracy sections from high-accuracy sections. The signal analysis unit 11c then re-estimates the heartbeat interval in the low-accuracy sections. The signal analysis unit 11c then integrates the previous and re-estimated heartbeat intervals and adopts the period N that provides the greatest accuracy as the heartbeat interval estimation result. After step S25, the heartbeat interval estimation process ends.

[0060] As a result, in the waveforms of the output signals of the sensor units 18a to 18d, one of the output signals of the sensors installed on the diagonally opposite legs of the bed is phase-inverted, and the heartbeat interval can be estimated by estimating the interval at which similar waveform patterns appear. Therefore, the biological information monitoring system 1 of this embodiment can measure the biological information of the subject with higher accuracy.

[0061] As described above, the biological information monitoring system 1 according to this embodiment includes the data acquiring unit 11a and the signal analyzing unit 11c. The data acquiring unit 11a is installed on a member supporting the subject, and acquires signals indicating physical quantities from a plurality of sensors that detect dynamic physical quantities (for example, loads). The signal analysis unit 11c acquires biological information relating to the heartbeat of the subject supported by the member, based on the peaks of the signals indicating the physical quantities acquired by the data acquisition unit 11a. This makes it possible to acquire biometric information (heart rate variability) robustly against the subject's movements, posture, etc., based on signals detected at the legs of the bed, without having to install any equipment on the subject's body. Therefore, according to the biological information monitoring system of this embodiment, the biological information of the subject can be measured with higher accuracy.

[0062] The signal analysis unit 11c extracts features related to the waveform of the peak of the signal indicating the physical quantity, classifies the features, and identifies the heartbeat peak from the peak of the signal indicating the physical quantity, and obtains biological information related to the heartbeat based on the identified heartbeat peak. This makes it possible to easily identify the peak of the heartbeat from the peak of the signal indicating the physical quantity by performing classification such as clustering.

[0063] The signal analysis unit 11c separates the signals indicating the physical quantities by performing component analysis processing, identifies the heartbeat component from the separated signals, and acquires biometric information related to the heartbeat based on the matching result between the identified heartbeat component and a template generated based on the peak of the identified heartbeat component. This makes it possible to obtain more accurate biological information related to the heartbeat using a template obtained by separating the heartbeat component contained in the signal indicating the physical quantity.

[0064] The signal analysis unit 11c acquires biological information relating to the heartbeat based on the peak of the correlation function between the template and the identified heartbeat component. This makes it possible to obtain biological information related to heart rate using clear indicators.

[0065] The data acquiring unit 11a inverts the phase of at least one of the signals indicating physical quantities acquired from a plurality of sensors. The signal analysis unit 11c extracts a section to be analyzed from the signal indicating the physical quantity, which is the processing result of the data acquisition unit 11a, and estimates the appearance interval of similar waveforms, thereby acquiring biological information related to the heartbeat. This allows the correlation between signals indicating physical quantities obtained from multiple sensors to be increased, making it possible to estimate the interval between occurrences of similar waveforms, thereby making it possible to obtain more accurate biological information related to heartbeats.

[0066] The signal analysis unit 11c calculates multiple correlation indices to estimate the appearance interval of similar waveforms, and acquires biological information related to the heartbeat by calculating the probability that the estimated appearance interval represents the heartbeat interval based on the calculated multiple correlation indices. This makes it possible to determine the correlation between the template and the identified heartbeat component using a variety of indices, thereby making it possible to acquire more accurate biological information related to the heartbeat.

[0067] The signal analysis unit 11c re-estimates the appearance interval of a similar waveform for a section in which the probability that the estimated appearance interval represents a heartbeat interval is lower than a set threshold. This makes it possible to further improve the accuracy of estimating the heartbeat interval in a section where the accuracy of estimating the heartbeat interval is low.

[0068] The present invention can be modified and improved as appropriate within the scope of the effects of the present invention, and is not limited to the above-described embodiment. For example, in the above embodiment, the heartbeat interval is acquired as biological information related to the subject's heartbeat, but this is not limiting. As an example, the biological information monitoring system 1 according to the present invention can acquire various biological information related to the heartbeat, such as the heart rate, heartbeat speed, and heartbeat regularity.

[0069] In the above embodiment, the sensors 18a to 18d are described as being installed on the legs of a bed, but this is not limiting. That is, as long as the mechanical influence (physical quantity) of the subject's heartbeat can be detected, the present invention can be implemented by installing sensors on members that directly or indirectly support the subject (supporting parts of various equipment such as chairs, sofas, and examination tables, the floor surface on which the subject's weight acts, in or under the floor, etc.). Furthermore, in the above-described embodiment, load sensors, acceleration sensors, etc. have been given as examples of the sensor units 18a to 18d, but various sensors such as displacement sensors and speed sensors that acquire information on the position of the members supporting the subject, and various sensors that acquire information on the force of the members supporting the subject, such as the air pressure of an airbag that is pressed directly or indirectly by the subject, can also be used.

[0070] Furthermore, the various methods used in the above-described embodiments are merely examples, and other methods may be used as long as they can achieve the same objective. For example, in the first embodiment, feature values ​​are classified using k-means clustering, but other classification methods may be used as long as they can classify data. Similarly, in the second embodiment, signal separation is performed using independent component analysis (ICA), but other analysis methods may be used as long as they can separate multidimensional signals. Similarly, in the third embodiment, the autocorrelation function of the signal, the average amplitude difference, and the maximum amplitude sum are used as correlation indices for estimating the interval between appearances of similar waveform patterns, but other indices may be used as long as they represent the correlation of the signal. Furthermore, the examples described in the above-described embodiments can be combined as appropriate to implement the present invention.

[0071] The above-described series of processes can be executed by hardware or software. In other words, the functional configurations in the above-described embodiments are merely examples and are not particularly limited. That is, it is sufficient that any of the computers constituting the biological information monitoring system 1 has a function capable of executing the above-described series of processes as a whole, and the functional blocks used to realize this function are not particularly limited to the examples shown. Furthermore, one functional block may be configured as a single piece of hardware, a single piece of software, or a combination thereof.

[0072] Furthermore, the recording medium containing the program for executing the above-mentioned series of processes may be configured not only as a removable medium distributed separately from the device main body in order to provide the program to the user, but also as a recording medium provided to the user in a state where it is pre-installed in the device main body.

[0073] The above embodiment shows an example of application of the present invention and does not limit the technical scope of the present invention. In other words, the present invention can be modified in various ways, such as by omission or substitution, without departing from the gist of the present invention, and various embodiments other than the above embodiment can be adopted. The various embodiments and modifications that the present invention can adopt are included in the scope of the invention described in the claims and their equivalents. [Explanation of symbols]

[0074] 1 biological information monitoring system, 11 CPU, 11a data acquisition unit, 11b preprocessing unit, 11c signal analysis unit, 12 ROM, 13 RAM, 14 input unit, 15 display unit, 16 memory unit, 17 communication unit, 18a to 18d sensor units, 19 imaging unit

Claims

1. a physical quantity acquiring means for acquiring a signal indicating a mechanical physical quantity from at least one of a plurality of sensors that are installed on each of the members that support the subject relative to the floor or on each of the portions of the floor on which a load acts from the members, and that detects the mechanical physical quantity; a signal analysis means for acquiring biological information relating to the heartbeat of the subject supported by the member based on peaks of the signal indicating the physical quantity acquired by the physical quantity acquisition means; A biological information monitoring system comprising:

2. 2. The biological information monitoring system according to claim 1, wherein the signal analysis means extracts features related to the waveform of peaks of the signal indicating the physical quantity, classifies the signals based on the features, identifies heartbeat peaks from the peaks of the signal indicating the physical quantity, and acquires biological information related to the heartbeat based on the identified heartbeat peaks.

3. 2. The biological information monitoring system according to claim 1, wherein the signal analysis means separates the signals indicating the physical quantities by performing component analysis processing, identifies a heartbeat component from the separated signals, and acquires biological information related to the heartbeat based on a matching result between a template generated based on the peak of the identified heartbeat component and the identified heartbeat component.

4. 4. The biological information monitoring system according to claim 3, wherein the signal analysis means acquires biological information relating to the heartbeat based on a peak of a correlation function between the template and the identified heartbeat component.

5. the physical quantity acquisition means inverts the phase of at least one of the signals indicating the physical quantities acquired from the plurality of sensors; The biological information monitoring system according to claim 1, characterized in that the signal analysis means extracts a section to be analyzed from the signal indicating the physical quantity that is the processing result of the physical quantity acquisition means, and acquires biological information related to the heartbeat by estimating an appearance interval of similar waveforms.

6. The biological information monitoring system according to claim 5, characterized in that the signal analysis means calculates a plurality of correlation indices for estimating the appearance interval of the similar waveforms, and calculates the probability that the estimated appearance interval represents a heartbeat interval based on the calculated plurality of correlation indices, thereby acquiring biological information related to the heartbeat.

7. The biological information monitoring system according to claim 6, characterized in that the signal analysis means re-estimates the appearance interval of the similar waveform for sections in which the probability that the estimated appearance interval represents a heartbeat interval is lower than a set threshold.

8. The information processing device a physical quantity acquiring step of acquiring a signal indicating a mechanical physical quantity from at least one of a plurality of sensors that are installed on each of the members that support the subject against the floor or on each of the portions of the floor on which a load acts from the members, and that detect the mechanical physical quantity; a signal analysis step of acquiring biological information related to the heartbeat of the subject supported by the member based on peaks of the signal indicating the physical quantity acquired in the physical quantity acquisition step; A biological information monitoring method comprising:

9. On the computer, a physical quantity acquisition function that acquires a signal indicating a mechanical physical quantity from at least one of a plurality of sensors that are installed on each of the members that support the subject relative to the floor or on each of the portions of the floor on which a load acts from the members, and detects the mechanical physical quantity; a signal analysis function for acquiring biological information related to the heartbeat of the subject supported by the member based on peaks of the signal indicating the physical quantity acquired by the physical quantity acquisition function; A program characterized by realizing the above.

Images