Biological state estimation device, computer program, and recording medium

Abstract

A biological state estimation device according to the present invention includes: an attractor configuration unit for obtaining an attractor by reconstructing a three-dimensional delay coordinate system by using a Lorenz plot for time-series data of an instantaneous heart rate; a reference plane computation unit for finding, as a reference plane, a plane obtained by subjecting the attractor to plane approximation by using a least-squares method; a correlation data computation unit for calculating, for each point of the instantaneous heart rate plotted in the three-dimensional delay coordinate system, an average value of distances from the reference plane and a standard deviation of the distances, and finding correlation data between the average value of the distances and the standard deviation of the distances for a healthy subject group and a cardiovascular disease subject group; and an estimation unit for finding, for a subject to be estimated, correlation data for estimation regarding the average value of the distances and the standard deviation of the distances by the correlation data calculation unit, comparing the correlation data for estimation with the correlation data, and estimating effects of administering medication to the subject to be estimated.

Description

Biological condition estimation device, computer program, and recording medium

[0001] The present invention relates to a technique for estimating the effects of medication on a person with cardiovascular disease.

[0002] The applicant has disclosed in Patent Document 1 a blood pressure estimation device that analyzes biosignal data obtained from a biosignal detection sensor using a three-dimensional knitted fabric, captures vibration information within the body caused by blood flow fluctuations from the ventricular filling period to the isovolumic contraction period, obtains an index showing the fluctuations by applying the Lorenz plot method, and estimates blood pressure values ​​from the index (Patent Document 1).

[0003] Specifically, this technology captures biosignals (acoustic vibration data (Acoustic Pulse Wave: APW)) from the body surface of the back using a biosignal detection sensor, filters the time-series waveform in a predetermined frequency band, and obtains a filtered waveform that reveals the cardiac cycle. This filtered waveform is compared with electrocardiogram data (ECG) obtained from an electrocardiograph measured at the same time, and the waveform components from the ventricular filling period to the isovolumic contraction period are identified. After that, the Lorenz plot method is applied, and using the slope of a large number of point clouds plotted at a predetermined measurement time as a reference, a new time series is constructed for the angle difference with the slope of point clouds at shorter measurement times. The time-series waveform is frequency-analyzed, and the frequency analysis results are displayed on a double logarithmic axis to find a regression line for the range from VLF to LF in the analyzed waveform. The slope of the regression line is defined as the fractal slope, and blood pressure is estimated from the fractal slope.

[0004] In Patent Document 2, a predetermined time width is applied to the time series waveform of the dorsal surface pulse wave near 1 Hz extracted from acoustic vibration data, and a sliding calculation is performed to obtain a time series waveform of the frequency gradient, and the biological condition is estimated from the tendency of the change, for example, whether the amplitude tends to increase or decrease. It also discloses that the biological signal is frequency analyzed, and the power spectrum of each frequency corresponding to the function adjustment signal, fatigue acceptance signal, and activity adjustment signal belonging to a predetermined ULF band (extremely low frequency band) to VLF band (very low frequency band) is obtained, and the human condition is determined from the time series change of each power spectrum.

[0005] Patent Documents 3 and 4 disclose means for determining the homeostatic function level, and Patent Document 5 discloses a technique for easily extracting the waveform of the apex of the heart by determining the boundary frequency between the vibration caused by the apex of the heart and the vibration caused by the heart sound from acoustic vibration data.

[0006] JP 2019-122502 A JP 2011-167362 A JP 2014-117425 A JP 2014-223271 A JP 2022-71786 A

[0007] The present applicant has developed a technology that uses APW and ECG to capture, as described above, human conditions related to bioregulation functions, such as blood pressure, sleep onset prediction phenomenon, impending sleep phenomenon, state of drowsiness, homeostatic function level, apical pulsation, etc. However, the conditions of bioregulation functions are not limited to these and are diverse, and it is desirable to be able to capture as many conditions as possible using APW and ECG.

[0008] In recent years, the advent of a super-aging society, the resulting shortage of medical professionals, and advances in information technology have combined to promote the widespread use of telemedicine and online medical consultations. APW measurements can be easily performed by simply placing a sensor on the back of the patient. For example, by incorporating a sensor like that shown in Figure 2 into a seat like the one shown in Figure 3 (described below), measurements can be taken while the patient is sitting or lying down, making it convenient for the elderly. Acoustic vibration data can be collected simply by the elderly sitting or lying down on the seat, and receiving this data can facilitate telemedicine and other procedures. It also facilitates data collection in home nursing. Since the target population for such telemedicine and home nursing is the elderly, many suffer from cardiovascular diseases such as hypertension and heart disease. Therefore, many patients are prescribed vasopressors and cardiac inotropes, which often have severe side effects. Therefore, it is desirable to be able to objectively evaluate the efficacy and severity of side effects of prescribed medications as part of in vivo information. In particular, since telemedicine reduces face-to-face encounters with patients, collecting these evaluations as information would enable more appropriate medication prescriptions. It is also useful when people are unable to effectively explain their condition due to dementia or other reasons.

[0009] The present invention has been made in view of the above, and an object of the present invention is to provide a technique that can estimate medication effects from time-series data of instantaneous heart rate.

[0010] In order to solve the above problem, the inventors noticed that there was a significant difference between a group of healthy subjects and a group of subjects with cardiovascular disease in the relationship between the average value and standard deviation of the distance between the reference plane of an attractor reconstructed into a three-dimensional delay coordinate system using time series data of instantaneous heart rate and the plotted points of each instantaneous heart rate, and thus completed the present invention.

[0011] That is, the present invention provides a biological state estimation device having: an attractor construction unit that determines an attractor by reconstructing a three-dimensional delay coordinate system using a Lorenz plot for time-series data of instantaneous heart rate; a reference plane calculation unit that determines, as a reference plane, a plane obtained by planar approximation of the attractor using the least squares method; a correlation data calculation unit that calculates, for each point of the instantaneous heart rate plotted on the three-dimensional delay coordinate system, an average value of the distance from the reference plane and a standard deviation of the distance, and determines correlation data between the average value of the distance and the standard deviation of the distance for a group of healthy subjects and a group of subjects with cardiovascular disease; and an estimation unit that determines correlation data for estimation regarding the average value of the distance and the standard deviation of the distance for a subject to be estimated by using the correlation data calculation unit, compares the correlation data for estimation with the correlation data, and estimates the effect of medication on the subject to be estimated.

[0012] The estimation unit is preferably configured to estimate a "high medication effect" when the estimation subject is a subject with cardiovascular disease and the estimation correlation data is within the approximate range of the correlation data of the group of healthy subjects. The estimation unit is preferably configured to estimate a state of pharmacological action caused by medication based on the position of the estimation correlation data when the estimation subject is a subject with cardiovascular disease and the estimation correlation data is within the approximate range of the correlation data of the group of subjects with cardiovascular disease. The estimation unit is preferably configured to estimate a "low medication effect" when the estimation subject is a subject with cardiovascular disease and the estimation correlation data is not within the approximate range of the correlation data of the group of subjects with cardiovascular disease.

[0013] As the time series data of the instantaneous heart rate, it is preferable to use electrocardiogram data or acoustic vibration data obtained from the body surface.

[0014] It is preferable that the time series data of the instantaneous heart rate is made up of two data, electrocardiogram data and acoustic vibration data obtained from the body surface, and the attractor construction unit finds an attractor for each of them, the reference plane calculation unit finds a reference plane for each of the attractors, the correlation data calculation unit creates correlation data for each of them, and the estimation unit estimates the medication effect on the subject based on the correlation data.

[0015] The present invention also provides a computer program that causes a computer to function as a biological state estimation device related to medication effects, which causes the computer to execute the following steps: a procedure of reconstructing a three-dimensional delay coordinate system using a Lorenz plot for time-series data of instantaneous heart rate to obtain an attractor; a procedure of obtaining a plane as a reference plane by planar approximation of the attractor using the least squares method; a procedure of calculating the average value of distance and the standard deviation of distance from the reference plane for each point of the instantaneous heart rate plotted in the three-dimensional delay coordinate system, and obtaining correlation data between the average value of distance and the standard deviation of distance for a group of healthy subjects and a group of subjects with cardiovascular disease; and a procedure of obtaining correlation data for estimation regarding the average value of distance and the standard deviation of distance for a subject to be estimated, comparing the correlation data for estimation with the correlation data, and estimating the medication effects on the subject to be estimated.

[0016] The present invention also provides a computer-readable recording medium on which the computer program is recorded.

[0017] According to the present invention, correlation data for estimation relating to the average value and standard deviation of the distance between the reference plane of the attractor of a subject with cardiovascular disease and the plotted points of each instantaneous heart rate are compared with correlation data of a group of healthy subjects and correlation data of a group of subjects with cardiovascular disease, which have been previously compiled into a database, and the effectiveness of medication can be estimated based on whether the correlation data for estimation approximates to either of the correlation data or their positional relationship. Since the instantaneous heart rate can be measured using acoustic vibration data from the body surface, which can be easily obtained while the subject is sitting or lying down, or electrocardiogram data, of which compact models are widely available, it can be easily measured at home, which is useful for remote medical consultations and home nursing, and can lead to more appropriate medication prescriptions.

[0018] FIG. 1 is a block diagram illustrating the configuration of a biological state estimation device according to one embodiment of the present invention. FIGS. 2(a) to 2(c) are diagrams illustrating the configuration of an APW sensor. FIG. 3 is a diagram illustrating an example of a seat with an APW sensor built into the backrest. FIGS. 4(a) to 4(c) are diagrams illustrating the relationship between APW waveforms and ECG waveforms. FIG. 5 is a graph illustrating the correlation between heart rates obtained from ECGs and those obtained from APWs. FIGS. 6(a) to 6(c) are diagrams illustrating attractors calculated using instantaneous heart rates obtained from ECGs. (a) shows data from a male subject B in his 20s, (b) from a male subject C in his 70s, and (c) from a female subject D in her 80s. FIGS. 7(a) to 7(c) are diagrams illustrating attractors calculated using instantaneous heart rates obtained from APWs. (a) shows data from a male subject B in his 20s, (b) from a male subject C in his 70s, and (c) from a female subject D in her 80s. Figures 8(a) and 8(b) are correlation diagrams of the mean distance (Mean) and standard deviation (SD) of the distance between the reference plane of the attractor and each point obtained using ECG, where (a) is the correlation data for the subjects in the healthy group, and (b) is the correlation data for the subjects in the outpatient group. Figures 9(a) and 9(b) are correlation diagrams of the mean distance (Mean) and standard deviation (SD) of the distance between the reference plane of the attractor and each point obtained using APW, where (a) is the correlation data for the subjects in the healthy group, and (b) is the correlation data for the subjects in the outpatient group. Figure 10(a) is a correlation diagram of the mean distance (Mean) and standard deviation (SD) of the distance for 171 subjects (260 cases), and Figure 10(b) is an enlarged view of that diagram.

[0019] (Biological state estimation device) Fig. 1 shows a schematic configuration of a biological state estimation device 1 according to one embodiment. The biological state estimation device 1 of this embodiment is configured by a computer (the type of computer is not limited, and includes a personal computer, a microcomputer, a mobile information terminal such as a smartphone, etc.), receives instantaneous heart rate data, which is biological signal data, and executes predetermined processing.

[0020] Specifically, the biological state estimation device 1 has an attractor construction unit 11, a reference plane calculation unit 12, a correlation data calculation unit 13, and an estimation unit 14. The biological state estimation device 1 is composed of a computer, and a computer program for executing procedures functioning as the attractor construction unit 11, the reference plane calculation unit 12, the correlation data calculation unit 13, and the estimation unit 14 is stored in a storage unit (memory). Note that the storage unit (memory) includes not only a recording medium such as a hard disk built into the biological state estimation device 1 composed of a computer, but also various removable recording media, recording media of other computers connected via communication means, and the like. The biological state estimation device 1 can also be realized using an electronic circuit having a memory in which the above-mentioned computer program is embedded.

[0021] The computer program can also be provided by being stored on a recording medium. The recording medium storing the computer program may be a non-transitory recording medium. Non-transitory recording media are not particularly limited, and examples thereof include flexible disks, hard disks, CD-ROMs, MOs (magneto-optical disks), DVD-ROMs, and memory cards. The computer program can also be transmitted to a computer via a communication line and installed.

[0022] The attractor construction unit 11 reconstructs a three-dimensional delay coordinate system using a Lorentz plot of time-series data of instantaneous heart rate (HR = 60 / Δt) to determine an attractor related to heart rate fluctuations. The time-series data of instantaneous heart rate can be, for example, electrocardiogram data obtained from an electrocardiograph. Alternatively, acoustic vibration data (APW) representing biosignals from the body surface of the back, collected by placing the APW sensor 100 shown in Figures 2(a) to 2(c) against a person's back, can also be used. Figures 4(a) to 4(c) show, from top to bottom, the original APW waveform (Figure 4(a)), the APW vibration wave waveform (Figure 4(b)), and electrocardiogram data (ECG) obtained from an electrocardiograph (Figure 4(c)). The APW vibration waveform in Figure 4(b) is a waveform near 1 Hz calculated by full-wave rectification and detection of the waveform (resonant carrier wave) obtained by applying a 10-30 Hz band-pass filter to the original APW waveform. Comparing Figures 4(b) and 4(c), the peak interval of the oscillation wave of the APW in Figure 4(b) shows a period that is close to the R-R interval (RRI) shown in the ECG in Figure 4(c). Therefore, as shown in Figure 5, there is a high correlation between the APW and the ECG.

[0023] The Lorenz plot uses three points from the time series data of instantaneous heart rate (HR) and plots them while moving the range. ｎ＋１ , H.R. ｎ＋２ , H.R. ｎ＋３ ..., HR on the y-axis ｎ＋２ , H.R. ｎ＋３ , H.R. ｎ＋４ ..., HR on the z-axis ｎ , H.R. ｎ＋１ , H.R. ｎ＋２ ... and plot their intersections in turn. This reconstructs the time series data of instantaneous heart rate into a three-dimensional delayed coordinate system, and an attractor is drawn.

[0024] The reference plane calculation unit 12 approximates the attractor drawn by the attractor construction unit 11 to a plane by the least squares method, and obtains a plane that shows the characteristics of the attractor. The obtained plane is used as the reference plane.

[0025] 6(a)-(c) and 7(a)-(c) are diagrams showing examples of attractors and reference planes of heart rate variability calculated by the attractor construction unit 11 and the reference plane calculation unit 12. Figures 6(a)-(c) use time series data of instantaneous heart rate obtained from ECG, while Figures 7(a)-(c) use time series data of instantaneous heart rate obtained from APW. Figures 6(a) and 7(a) show data from male subject B in his 20s who belongs to the healthy group of an experimental example described below. Figures 6(b) and 7(b) show data from male subject C in his 70s who belongs to the outpatient group of the same experimental example. Figures 6(c) and 7(c) show data from female subject D in her 80s who also belongs to the outpatient group. Male subject C suffers from a lifestyle-related disease (diabetes), heart disease (supraventricular tachycardia, atrial fibrillation), and chronic hepatitis, while female subject D suffers from high blood pressure, heart disease (atrial fibrillation), and cerebral infarction.

[0026] The attractors of male subject B in the healthy group (Fig. 6(a) and Fig. 7(a)) are all drawn within a fixed range of 70 to 90 / min. The attractors of male subject C in the outpatient group (Fig. 6(b) and Fig. 7(b)) are all concentrated around 50 / min, but show large fluctuations at several points. For female subject D in the outpatient group, the attractor using ECG data (Fig. 6(c)) fluctuates between 45 and 116 / min, and the attractor using APW data (Fig. 7(c)) fluctuates between 31 and 112 / min, both of which show large fluctuations.

[0027] From these findings, the attractor of heart rate variability using ECG or APW data shows fluctuations corresponding to the subject's health condition, characteristics of the disease, etc. Because both male subject C and female subject D in the outpatient group suffered from heart disease, it is thought that abnormalities in circulatory dynamics such as arrhythmia appeared as peculiar fluctuations in the attractor, and it is thought that circulatory dynamics can be estimated from the shape of the attractor of heart rate variability.

[0028] The correlation data calculation unit 13 calculates correlation data for determining characteristics of the shape of the heart rate variability attractor. The correlation data calculation unit 13 uses the distance between the reference plane calculated by the reference plane calculation unit 12 and each point of the instantaneous heart rate constituting the attractor plotted in the three-dimensional delay coordinate system as an element characterizing the shape of the attractor. Specifically, the correlation data calculation unit 13 calculates the average value of the distances of each point from the reference plane and the standard deviation of the distances, and calculates correlation data between them. Furthermore, correlation data is generated for a group of healthy subjects (the "healthy group" in the experimental example described below) and a group of subjects with cardiovascular disease (the "outpatient group" in the experimental example described below), and the correlation data is stored in a memory unit of the biological state estimation device 1, which is composed of a computer, to construct a database (DB) of the correlation data.

[0029] The estimation unit 14 obtains data on the instantaneous heart rate of the subject for whom the medication effect is to be estimated using an ECG or APW, and uses this data to calculate an attractor and its reference plane using the attractor construction unit 11 and the reference plane calculation unit 12. Furthermore, the correlation data calculation unit 13 calculates the reference plane and the average value and standard deviation of the distances between each point. This is used as correlation data for estimation, and this correlation data for estimation is compared with the correlation data stored in the database to estimate the medication effect on the subject. The comparison is performed by adding correlation data from a group of subjects with cardiovascular disease (outpatient group) to correlation data from a group of healthy subjects (healthy group), and details will be explained in the experimental example below.

[0030] (Experimental Examples) (1) Sensing System - APW The APW sensor 100 is similar to the sensor used to measure APW in Patent Document 5, and as shown in Figures 2(a) to 2(c), has a laminated structure of an air pack 110 and a gel pack 120. The air pack 110 is composed of a three-dimensional knitted fabric (3D net) 111 and an elastomer film 112 that hermetically houses the three-dimensional knitted fabric (3D net) 111. The gel pack 120 has a microphone 122 fixedly disposed within a case 121, and gel 123 filled around the microphone 122. Using two resonance mechanisms, stochastic resonance and string vibration, the APW, including the apex beat and heart sounds in the range of 0.5 to 80 Hz, is efficiently extracted.

[0031] The APW sensor 100 is placed on the backrest of the seat shown in FIG. 3 so that the microphone 122 corresponds to the fifth rib area, 10 cm to the left of the midline of the chest.

[0032] ECG An electrocardiograph manufactured by Nihon Kohden Corporation, product name: Bedside Monitor BSM-2301, was used.

[0033] (2) Experimental Protocol To evaluate cardiac and vascular hemodynamics, subjects were individuals without underlying cardiovascular diseases (healthy group) and individuals visiting a hospital for cardiovascular diseases such as lifestyle-related diseases and heart disease (outpatient group). The subjects were a healthy group of 50 subjects aged 20 to 70 (48 men, 2 women, mean age 37.8 ± 12.4 years) and an outpatient group of 50 subjects aged 40 to 90 (33 men, 17 women, mean age 74.4 ± 10.7 years). Measurement items included APW, electrocardiogram, and brachial blood pressure. Subjects sat in the seat shown in Figure 3 and, while resting, APW was measured for 6 minutes, and lead II electrocardiogram data (ECG) was recorded. Brachial blood pressure was measured before and after APW and ECG measurements.

[0034] (Experimental Results) The measured APW and ECG were analyzed by the biological state estimation device 1. Using the time series data of the instantaneous heart rate obtained from the APW and the time series data of the instantaneous heart rate obtained from the ECG, the attractor construction unit 11 created an attractor of heart rate fluctuations, the reference plane calculation unit 12 found a reference plane of the attractor, and the correlation data calculation unit 13 created correlation data relating to the average value of the distance between the reference plane and each point and the standard deviation of the distance, as described above.

[0035] 8(a)-(b) and 9(a)-(b) are correlation diagrams of the mean distance (Mean) and standard deviation (SD) of the distance between each point and the reference plane of the heart rate variability attractor for 50 healthy subjects and 50 outpatient subjects. Figs. 8(a)-(b) were obtained using ECG data, and Figs. 9(a)-(b) were obtained using APW data. Figs. 8(a) and 9(a) are correlation data for the healthy subject, and Figs. 8(b) and 9(b) are data for the outpatient subject. The variance of each point group plotted as correlation data between the mean distance (Mean) and standard deviation (SD) of the distance for the healthy subject can be approximated by a linear approximation equation, and the coefficient of determination R ２ = 0.9226, and for correlation data using APW, R ２ = 0.9134, showing a high correlation. The correlation data using ECG were plotted along a line at an angle of 54 degrees to the horizontal axis (Fig. 8(a)), while the correlation data using APW were plotted along a line at an angle of 45 degrees to the horizontal axis (Fig. 9(a)).

[0036] The variance of each point group plotted as correlation data for the outpatient group can be approximated by a quadratic approximation formula, and the coefficient of determination R ２ = 0.909, and for correlation data using APW, R ２ = 0.9352, showing a high correlation. In the correlation data using ECG in Figure 8(b), when the mean (Mean) was 2 or less, there were more subjects plotted below the 45-degree line than in the healthy group. Also, many plots were observed where the mean (Mean) and standard deviation (SD) were both 4 or more. In the correlation data using APW in Figure 9(b), when the mean (Mean) and standard deviation (SD) were both 10 or less, there were many subjects plotted below the 45-degree line. On the other hand, when the mean (Mean) and standard deviation (SD) were both 10 or more, there were many subjects plotted above the 45-degree line, showing a different appearance from the plots in the healthy group.

[0037] As mentioned above, whether APW or ECG is used, there is a difference in that the healthy group shows a linear approximation and the outpatient group shows a nonlinear approximation. Therefore, the correlation data between the mean (Mean) of the distance and the standard deviation (SD) of the distance in the attractor of heart rate variability using APW and ECG data can distinguish between differences in the mechanisms of circulatory dynamics.

[0038] Looking more closely, in Figure 9(b), most of the subjects with heart disease in the outpatient group ("HDP" in the figure) have mean (Mean) and standard deviation (SD) values ​​both greater than 10, or are plotted below the 45-degree line. This is thought to be due to the occurrence of arrhythmias such as bradycardia, tachycardia (e.g., atrial fibrillation), and premature ventricular contractions, resulting in unstable heart rate variability, which is manifested as the spatial spread (Mean) and spatial variability (Standard Deviation (SD)) of the heart rate variability attractor. On the other hand, several subjects were also observed plotting near the 45-degree line. The area near the 45-degree line is similar to the tendency of the plot positions of the subjects in the healthy group in Figure 9(a), suggesting that these subjects are able to maintain hemodynamics similar to that of the healthy group through medication. The plots for subject C shown in Figures 6(b) and 7(b) are below the 45-degree line in Figures 8(b) and 9(b), while the data for subject D shown in Figures 6(c) and 7(c) are plotted in Figures 8(b) and 9(b) where the mean (Mean) and standard deviation (SD) are both 10 or more. Subject C is characterized by a decrease in the mean (Mean) relative to the quadratic approximation line, while subject D is characterized by an increase in the mean (Mean). Furthermore, the mean (Mean) and standard deviation (SD) for both subjects C and D are plotted in areas where the quadratic approximation line is highly nonlinear. Plotting in areas of strong nonlinearity is thought to indicate that the pharmacological effects of the beta-blocker, such as reducing heart rate, are at work, and that this pharmacological effect maintains balance in the physiological states of subjects C and D.

[0039] As described above, the shape and spatial structure of the attractor of heart rate fluctuations differ depending on the pathological condition and medication effects, so by classifying the characteristics of the attractor shape using the mean value (Mean) of the distance from the reference plane and the standard deviation (SD) of the distance, it is possible to classify the changes in circulatory dynamics due to the pathological condition and medication effects. Here, the following classification criteria 1 to 3 were set.

[0040] Classification criterion 1: Mean or SD is 10 or less, and the difference between Mean and SD is 1.5 or less. Classification criterion 2: Mean or SD is 10 or less, and the difference between Mean and SD is 1.5 or more. Classification criterion 3: Mean and SD are both 10 or more. Table 1 shows the classification results based on the above classification criteria for correlation data using APW.

[0041]

[0042] The majority of the healthy group had classification criterion 1, while the majority of the outpatient group had classification criterion 3. Fisher's exact test showed a P value of 7.1 × 10 －７ In the outpatient group, classification criterion 1 was more prevalent after classification criterion 3, and it is thought that the effect of medication was manifested as an increase in classification criterion 1.

[0043] Based on the above experimental results, the estimation unit 14 estimates the medication effect using the estimation correlation data of the subject as follows. Note that the attractor in this embodiment relates to heart rate variability created using time-series data of instantaneous heart rate, and the correlation data obtained from this attractor is analyzed by comparing the correlation data of subjects with cardiovascular disease with the correlation data of a healthy group, as described above. Therefore, in this embodiment, the subjects for whom medication effect can be estimated are subjects with cardiovascular disease such as high blood pressure and heart disease.

[0044] The estimation unit 14 first obtains time-series data of instantaneous heart rate from the APW or ECG of the subject to be estimated as having a cardiovascular disease. Then, as described above, the attractor construction unit 11, the reference plane calculation unit 12, and the correlation data calculation unit 13 calculate correlation data for estimation relating to the reference plane and the average and standard deviation of the distances between each point. The calculated correlation data for estimation is plotted against the correlation data of the outpatient group stored in the database.

[0045] The correlation data for the outpatient group can be estimated from the above experiment as follows: Estimation criterion 1: The correlation data for estimation is plotted within the approximate range of the approximation line for the healthy group (54-degree line when based on ECG, 45-degree line when based on APW). If the data is plotted within the approximate range, it is estimated that medication has maintained hemodynamics close to that of the healthy group, and that the medication is highly effective.

[0046] Estimation criterion 2: When the correlation data for estimation is within the approximate range of the correlation data of the outpatient group, the plot position is plotted near the area with strong nonlinearity. If it is plotted near the area with strong nonlinearity, it is estimated from the results of subjects C and D above that a state in which balance is maintained by pharmacological action is maintained.

[0047] Estimation Criterion 3: The correlation data for estimation is not within the approximate range of the correlation data for the outpatient group. In this case, it is estimated that the medication effect is low. For example, as shown in Figures 10(a) and 10(b), this is data that falls far outside the quadratic approximation line for the outpatient group (data marked P1 and P2 in the figure). When such an estimation result is output, a change in medication can be considered. Note that Figure 10(a) is a correlation diagram of the mean (Mean) and standard deviation (SD) of the distance between the attractor and each point for a total of 171 subjects (260 cases), including 95 subjects in the healthy group and 76 subjects in the outpatient group, obtained using data obtained with a larger number of subjects than in the above experimental example. Figure 10(b) is an enlarged view of the range in Figure 10(a) where both the mean (Mean) and standard deviation (SD) are 5 or less.

[0048] In the above estimation criteria 1 to 3, whether or not something is within the "approximation range" can be determined, for example, by setting a threshold value for the distance to the approximation line, and the strength of the nonlinearity can be specified, for example, by setting a threshold value for the curvature, and whether or not something is close to a nonlinear part of a predetermined strength can be determined, for example, by setting a threshold value for the distance between the two.

[0049] The estimation unit 14 can also be configured to obtain correlation data for the subject from both the ECG and APW data, and compare the two correlation data with the correlation data obtained from the ECG ( FIG. 8( b) ) and the correlation data obtained from the APW ( FIG. 9( b) ), respectively, to estimate the medication effect. While the ECG corresponds to information on cardiac activity itself, the APW collects sound propagating through the body surface and therefore includes acoustic information originating from sources other than the heart. Comparing FIG. 8( b) and FIG. 9( b) , the correlation data for each subject does not necessarily match, and the correlation data based on the APW is more dispersed than the correlation data based on the ECG. This is due to non-cardiac information contained in the APW. Therefore, for example, when the correlation data for estimation is plotted against the correlation data based on the ECG, it may be within the approximate range of the healthy group, but when the correlation data for estimation is plotted against the correlation data based on the APW, it may not be within the approximate range of the healthy group. This indicates the possibility that the pharmacological effect is not sufficient, and the accuracy of the estimation can be improved by configuring the system to estimate that the medication is highly effective only when both sets of data are within the approximate range of the healthy group.

[0050] Therefore, it is preferable that the estimation unit 14 is configured to obtain correlation data for estimation using both ECG and APW data, and make an estimation by comparing it with correlation data for both ECG and APW stored in the database.

[0051] As described above, in this embodiment, focusing on the fact that an attractor formed using time-series data of instantaneous heart rate from an ECG or APW differs depending on the circulatory dynamics of a subject, a configuration is adopted that captures the characteristics of the shape of the attractor. Therefore, by comparing the correlation data for estimation of a subject to be estimated with correlation data from a group of healthy subjects (healthy group) and a group of subjects with cardiovascular disease (outpatient group), it is possible to estimate the medication effect of the subject to be estimated.

[0052] REFERENCE SIGNS LIST 1 biological state estimation device 11 attractor construction unit 12 reference plane calculation unit 13 correlation data calculation unit 14 estimation unit 100 APW sensor

Claims

1. A biological state estimation device having an attractor construction unit that uses a Lorenz plot to reconstruct a three-dimensional delay coordinate system from time-series data of instantaneous heart rate to determine an attractor; a reference plane calculation unit that calculates a plane obtained by planar approximation of the attractor using the least squares method as a reference plane; a correlation data calculation unit that calculates the average value of the distance and the standard deviation of the distance from the reference plane for each point of the instantaneous heart rate plotted on the three-dimensional delay coordinate system, and determines correlation data between the average value of the distance and the standard deviation of the distance for a group of healthy subjects and a group of subjects with cardiovascular disease; and an estimation unit that uses the correlation data calculation unit to determine correlation data for estimation regarding the average value of the distance and the standard deviation of the distance for a subject to be estimated, compares the correlation data for estimation with the correlation data, and estimates the effect of medication on the subject to be estimated.

2. The biological state estimation device according to claim 1, wherein the estimation unit estimates that "the medication effect is high" when the subject to be estimated has a cardiovascular disease and the correlation data for estimation is within an approximate range of the correlation data of the group of healthy subjects.

3. The biological state estimation device according to claim 1, wherein, when the subject to be estimated is a subject with cardiovascular disease and the correlation data for estimation is within an approximate range of the correlation data of the group of subjects with cardiovascular disease, the estimation unit estimates the state of pharmacological action caused by medication based on the position of the correlation data for estimation.

4. The biological state estimation device according to claim 1, wherein the estimation unit estimates that "medication effect is low" when the subject to be estimated has a cardiovascular disease and the correlation data for estimation is not included in the approximate range of the correlation data of the group of subjects with cardiovascular disease.

5. The biological state estimating device according to claim 1, wherein electrocardiogram data or acoustic vibration data obtained from the body surface is used as the time series data of the instantaneous heart rate.

6. A biological state estimation device as described in claim 1, wherein the time series data of instantaneous heart rate are made up of electrocardiogram data and acoustic vibration data obtained from the body surface, the attractor construction unit finds an attractor for each, the reference plane calculation unit finds a reference plane for each attractor, the correlation data calculation unit creates correlation data for each, and the estimation unit estimates the effect of medication on the subject based on the correlation data.

7. A computer program that causes a computer to function as a biological state estimation device for medication effects, comprising the steps of: reconstructing a three-dimensional delay coordinate system using a Lorenz plot for time-series data of instantaneous heart rate to obtain an attractor; obtaining a plane as a reference plane by planar approximation of the attractor using the least squares method; calculating the average value of the distance and the standard deviation of the distance from the reference plane for each point of the instantaneous heart rate plotted in the three-dimensional delay coordinate system, and obtaining correlation data between the average value of the distance and the standard deviation of the distance for a group of healthy subjects and a group of subjects with cardiovascular disease; and obtaining correlation data for estimation of the average value of the distance and the standard deviation of the distance for a subject to be estimated, comparing the correlation data for estimation with the correlation data, and estimating the medication effects on the subject to be estimated.

8. A computer-readable recording medium on which the computer program according to claim 7 is recorded.

Images