Pulse analysis device, pulse analysis method, and program

Abstract

To provide a technology to appropriately remove noise from time-series data on a heartbeat or a pulse.SOLUTION: A pulsation analysis device includes: a beat interval time-series data acquisition unit for acquiring beat interval time-series data for time of a beat interval, which is a time interval of a heartbeat or a pulse; a variation deriving unit for deriving a change speed in the beat interval or an acceleration in the beat interval from the beat interval time-series data on the basis of the beat interval time-series data acquired by the beat interval time-series data acquisition unit; a maximum frequency acquisition unit for classifying the change speed or the acceleration derived by the variation deriving unit into a plurality of sections and acquiring a maximum frequency out of these sections; and a removal unit for removing noise data from the interval time-series data according to the maximum frequency acquired by the maximum frequency acquisition unit.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a pulsation analysis device, a pulsation analysis method, and a program. [Background technology]

[0002] Time-series data of pulse intervals or instantaneous pulse rate obtained from pulse wave signals such as optical pulse wave sensors (hereinafter referred to as pulse interval time-series data) are used to monitor the body's state in daily life. Pulse interval time-series data contains various physiological and physical noises caused by the effects of ambient light, respiration, and body movement on pulse wave signal measurement. In particular, pulse interval time-series data recorded during activity contains a huge amount of pulse-like noise associated with body movement, which greatly reduces the reliability of evaluating the body's state using pulse interval time-series data recorded during activity.

[0003] Specifically, much of the noise in the time-series pulse interval data recorded during activity consists of abnormally long pulse interval data due to detection errors of one to several pulse waves, and abnormally short pulse wave interval data due to the misidentification of noise contained in the pulse wave as a pulse wave. Both the abnormally long pulse interval data and the abnormally short pulse wave interval data mentioned above generate pulse-like noise. These noises need to be distinguished from pulse wave signals.

[0004] There are various methods for removing noise contained in pulse waves. The most widely used method is to extract the pulse wave signal by bandpass filtering, which allows only the frequency band relevant to the pulse wave to pass through, based on frequency analysis. Specifically, for example, a bandpass filter that passes through the frequency band of 0.7 to 2.0 Hz is used to remove signal components other than the pulse component (for example, Patent Document 1). However, with the above method, the pulse-like noise contained in large quantities in the pulse wave signal affects a wide frequency band, so the true pulse wave signal cannot be properly extracted. Furthermore, the inventors of the present invention have found that fluctuations in the RR interval and pulse interval are modulated in different frequency bands by autonomic nervous system function, respiration, and other factors (Non-Patent Document 1), and have experienced that conventional noise removal using bandpass filters is not effective in removing noise.

[0005] Another method involves creating a histogram with the pulse interval on the horizontal axis and the frequency of occurrence of that pulse wave interval on the vertical axis. The distribution of the true pulse interval data is then estimated from the shape of the histogram's distribution, and data that deviates from this distribution is identified as noise (for example, Patent Document 2).

[0006] The method described in Patent Document 2, etc., attempts to distinguish between pulse interval data and noise by estimating the distribution of pulse interval data. However, as described in Non-Patent Document 1, pulse intervals vary greatly between individuals and depending on the individual's state, so the distribution of true pulse interval data itself also shows large fluctuations. As a result, overlap with the distribution of noise data often occurs, and it is difficult to distinguish between true pulse interval data and noise based on the difference in their distributions.

[0007] Furthermore, while the methods described in Patent Document 2 and others consider noise to be removed when the number of samples is below a certain level, in reality, the number of samples at the noisy locations is not necessarily small, so there is a risk of deleting non-noise data. Other similar methods include removing pulse intervals below or above a certain interval as noise, but determining what interval to consider as noise often relies on empirical rules. Also, as described in Non-Patent Document 1, there are individual differences and intra-individual fluctuations in pulse intervals, and methods using a fixed threshold cannot perform appropriate noise removal. [Prior art documents] [Patent Documents]

[0008] [Patent Document 1] Japanese Patent Publication No. 2016-190022 [Patent Document 2] Japanese Patent Publication No. 2009-297184 [Non-patent literature]

[0009] [Non-Patent Document 1] E. Yuda et al., Journal of Physiological Anthropology(2020)39:21 [Overview of the Initiative] [Problems that the invention aims to solve]

[0010] In view of the above circumstances, the present invention aims to provide a technique that can suitably remove noise from time-series data of heart rate or pulse intervals. [Means for solving the problem]

[0011] A beat analysis device, a beat analysis method, or a program according to an aspect of the present invention includes: a beat interval time series data acquisition unit that acquires beat interval time series data, which is the time interval of a heartbeat or a pulse, with respect to time; a variation derivation unit that derives, based on the beat interval time series data acquired by the beat interval time series data acquisition unit, the change rate of the beat interval or the acceleration of the beat interval from the beat interval time series data; a maximum frequency acquisition unit that classifies the change rate or the acceleration derived by the variation derivation unit into a plurality of sections and acquires the maximum frequency in each section; and a removal unit that removes the noise data from the interval time series data according to the maximum frequency acquired by the maximum frequency acquisition unit.

Effect of the Invention

[0012] According to the present invention, it is possible to provide a technique capable of suitably removing noise from beat interval time series data indicating the time interval of a heartbeat or a pulse.

Brief Description of the Drawings

[0013] [Figure 1] It is a functional block diagram showing the functional configuration of the beat analysis device 100. [Figure 2] It is a diagram showing an example of pulse wave data. [Figure 3] It is a diagram showing an enlarged view of the pulse wave data. [Figure 4] It is a diagram showing an example (baseline fluctuation) of the pulse wave data. [Figure 5] It is a diagram showing a histogram of d(i). [Figure 6] It is a diagram showing time series data measured as the R-R interval. [Figure 7] It is a diagram showing time series data of the pulse wave interval obtained from the pulse wave signal measured by the optical heart rate sensor provided in the wearable device. [Figure 8] It is a diagram showing time series data with noise data removed. [Figure 9] It is a flowchart showing the flow of the process of the beat analysis device. [Modes for carrying out the invention]

[0014] The following describes specific examples of the present invention with reference to the drawings.

[0015] [Pulsation analysis device of the present invention] Figure 1 is a functional block diagram showing the functional configuration of the pulsation analysis device 100. The pulsation analysis device 100 includes a CPU (Central Processing Unit), memory, and auxiliary storage devices connected by a bus, and functions as a device comprising a communication unit 110, a pulse interval time-series data storage unit 141, and a control unit 120 by executing a pulsation analysis program. Note that all or part of the functions of the control unit 120 may be implemented using hardware such as an ASIC (Application Specific Integrated Circuit), PLD (Programmable Logic Device), or FPGA (Field Programmable Gate Array). The pulsation analysis program may be recorded on a computer-readable recording medium. Computer-readable recording media include, for example, portable media such as flexible disks, magneto-optical disks, ROMs, and CD-ROMs, and storage devices such as hard disks built into computer systems. The pulsation analysis program may be transmitted via a telecommunications line.

[0016] The beat interval time series data storage unit 141 is configured using a storage device such as a magnetic hard disk drive or a semiconductor storage device. The beat interval time series data storage unit 141 stores beat interval time series data that indicates the time interval between heartbeats or pulses.

[0017] The communication unit 110 is an interface for communicating with other devices. The communication unit 110 is, for example, a NIC (Network Interface Card) or a USB (Universal Serial Interface) interface.

[0018] In Figure 1, the control unit 120 controls the operation of each part of the pulse analysis device 100. By executing the pulse analysis program, the control unit 120 functions as a pulse interval time series data acquisition unit 121, a variation derivation unit 122, a maximum frequency acquisition unit 123, and a removal unit 124.

[0019] [Beat interval time series data acquisition unit] The beat interval time series data acquisition unit 121 receives signals related to heart rate or pulse from other devices, such as a wearable device, via the communication unit 110, and acquires time series data of heart rate or pulse interval based on the received signals. The time series data of heart rate or pulse interval x(i) (hereinafter sometimes referred to as "beat interval") acquired by the beat interval time series data acquisition unit 121 is stored in the beat interval time series data storage unit 141. In this embodiment, the beat interval time series data is denoted as (t(i), x(i)) (where t(i) is the temporal position of x(i) and i is an integer).

[0020] The following explanation describes the time-series data of heart rate or pulse intervals (t(i), x(i)). Figure 2 shows an example of pulse wave data. Figure 2 shows a noise-free pulse wave over one minute. In Figure 2, the horizontal axis represents time (0 to 60 seconds), and the vertical axis represents pulse wave intensity (amplitude intensity). Figure 3 is an enlarged view of the pulse wave data from Figure 2 (with the horizontal axis representing 0 to 20 seconds). In Figure 3, the time interval x(i) between peaks, indicated by double-headed arrows, represents the pulse interval, and the time of the start of x(i) is t(i). This is measured for all pulses, and the time series data with time on the horizontal axis and pulse (pulse, heartbeat) interval on the vertical axis becomes the pulse interval time series data (t(i), x(i)).

[0021] Figures 2 and 3 illustrate the case without noise, but actual data contains a lot of noise. Let's explain the noise in detail. Figure 4 shows an example of pulse wave data including baseline fluctuation noise. In Figure 4, the horizontal axis represents time (0 to 320 seconds), and the vertical axis represents pulse wave intensity.

[0022] In Figure 4, the pulse wave rises around 210 seconds, then falls around 280 seconds, before rising again. Thus, it can be seen that noise, including baseline fluctuations, causes periods of reduced pulse wave intensity and shifts in the time interval between pulses. Other causes of this noise include low amplitude, the intrusion of ambient light into the optical sensor, and the separation of the optical sensor from the skin, each of which causes various types of noise in the pulse interval data.

[0023] [Variation Derivation Section] The variation derivation unit 122 derives the rate of change of the beat interval, or the acceleration of the beat interval, from the beat interval time series data (t(i), x(i)) of the beat interval x(i).

[0024] The method for deriving the rate of change is one that uses the difference between adjacent intervals. The rate of change d(i) from the immediately preceding pulse interval x(i-1) to each individual beat interval x(i) is expressed by the following (Equation 1). d(i) = x(i) - x(i-1) (Equation 1) In this method, d(i) = 0 when x(i) = x(i-1), that is, when the distance between adjacent elements does not change.

[0025] The method for deriving acceleration is to use the acceleration d of the beat intervals before and after each beat interval x(i). 2 This method uses (i). Acceleration, d 2 (i) is expressed by the following (Equation 2). d 2 (i)=x(i+1)-2x(i)+x(i-1) (Formula 2) In this method, if there is no change in the difference, in other words, if the second of three consecutive intervals is the same as the first and third intervals, then d(i)=0. The acceleration can also be derived by differentiating (Equation 1), which was derived using the method for deriving the rate of change, with respect to time.

[0026] In this embodiment, either the rate of change in beat interval or the acceleration of the beat interval may be used.

[0027] The reason for using such a rate of change in pulse interval or acceleration of pulse interval in this invention will be explained. A characteristic of noise in pulse wave signals is that the pulse interval itself fluctuates greatly due to physiological factors or measurement accuracy, and it is generally difficult to distinguish it from abnormal intervals caused by noise due to differences in their distribution.

[0028] On the other hand, the transfer function of autonomic nerve-mediated heart rate regulation is due to frequency characteristics and therefore appears as fluctuations across multiple pulses, and there is a physiological limit to the value of these fluctuations. The rate of change of the beat interval d(i) or the acceleration of the beat interval obtained from d 2 When (i) is calculated, although there are individual differences due to age, constitution, and disease state, the fluctuations will fall within a predetermined range because there is a physiological limit to the value.

[0029] In contrast, noise caused by pulse wave detection errors or false detections results in sudden, spike-like or pulse-like fluctuations, exhibiting non-physiological fluctuation velocities and accelerations. Therefore, by determining the fluctuation velocity and acceleration, it becomes possible to distinguish between physiological fluctuations and fluctuations caused by noise.

[0030] This invention utilizes this characteristic of beat intervals to identify speeds that deviate from the physiological beat interval change rate or acceleration distribution estimated for each individual as noise, thereby enabling accurate noise identification tailored to each individual.

[0031] [Maximum frequency acquisition unit] Next, we will discuss noise identification methods tailored to individual users. The rate of change of beat interval d(i) or the acceleration of beat interval d, which was determined as described above. 2 (i) A histogram of the values ​​obtained by the maximum frequency acquisition unit 123 is created.

[0032] The maximum frequency acquisition unit 123 obtains the rate of change of the beat interval d(i) or the acceleration of the beat interval d 2 The values ​​obtained in (i) are classified into non-overlapping intervals b (hereinafter referred to as "bin width b"). An example of a bin width is 8 to 24 ms, but it is not limited to this.

[0033] The maximum frequency acquisition unit 123 obtains the rate of change of the beat interval d(i) or the acceleration of the beat interval d 2 (i) is converted into a histogram. Probabilistically, the highest frequency occurs when there is no velocity or acceleration change between adjacent beats, so the frequency is highest near 0.

[0034] The rate of change in the interval between beats d(i) or the acceleration of the interval between beats d 2 Let N be the total number of data points in (i), and F be the height of the highest peak in the resulting histogram (highest frequency) (hereinafter referred to as "highest frequency"). Note that it is acceptable for the highest frequency F to occur in multiple intervals.

[0035] Creating a histogram in this way results in a histogram that shows a distribution spreading out to the left and right, centered around 0, as shown in Figure 5. The area of ​​this histogram is equal to the bin width b × the number of data points N, and is expressed by the following equation (Equation 3). N b (Equation 3) The portion of this histogram distribution containing the highest frequency F in the center represents changes in the heart rate due to physiological fluctuations (the desired peak (pulse)), while some frequencies in the histograms extending to the left and right periphery can be considered to be due to spike-like or pulse-like noise caused by pulse wave detection errors or false detections.

[0036] Furthermore, the peak frequency F in this histogram is the most characteristic of the changes in beat due to individual physiological variations, and there are large individual differences. For example, young people tend to have a low peak frequency F, and the frequency distribution of the histogram tends to be wide. On the other hand, elderly people and diabetic patients tend to have a high peak frequency F, and the frequency distribution of the histogram tends to be narrow.

[0037] In other words, from the histogram, the frequency information around the highest frequency F represents physiological information, and the further away from the center, the more the value exceeds the physiological limit, i.e., it represents information associated with noise. On the other hand, the height of the highest frequency F represents the trend of an individual's pulse wave, and it can be said that it indicates the tendency of the frequency to be concentrated in each individual.

[0038] [Removal unit] Therefore, the removal unit 124 determines the estimation range of the distribution of the individual's physiological variations from the histogram created by the maximum frequency acquisition unit 123. The removal unit 124 determines the range of the distribution of physiological variations (conversely, the range for removing noise) according to the height of the highest frequency F indicating the individual's physiological characteristics.

[0039] As an example, it is regarded as a triangle with the height around the highest frequency F as the apex A, and the length of the base between B - C of that range is set as 2TI. At this time, the area of the triangle ABC is represented by the following (Equation 4). TI·F (Equation 4) At this time, assuming that the originally obtained data is supposed to fall within the range of the triangle with a predetermined coefficient α, since the area formed by the triangle (Equation 4) and the area of the histogram (Equation 3) are equal, the length of the base TI is represented by the following (Equation 5). αTI·=N·b / F (Equation 5)

[0040] Fig. 5 shows the superposition of the triangle ABC when α = 1 on the histogram. The horizontal axis is at the height F of 0, and it is a triangle with an area of N·b. 2TI is the length of the base BC represented by 2Nb / F. The coordinate of B is (Nb / F, 0), and the coordinate of C is (-Nb / F, 0).

[0041] That is, in the case of the velocity variation d(i) of the pulse, the acceleration d of the pulse interval in (Equation 6) 2 (i) In the case of, x(i) corresponding to the range in (Equation 7) can be estimated to be due to physiological variations. -αTI < d(i) < αTI (Equation 6) -αTI < d 2 (i) < αTI (Equation 7) Conversely, in the case of the velocity variation d(i) of the pulse, the acceleration d of the pulse interval in (Equation 8) 2 (i) In the case of, x(i) corresponding to the range in (Equation 9) can be estimated to be due to pulse-like noise. d(i) < -αTI, αTI < d(i) (Equation 8) d2 (i) <- αTI, αTI <d 2 (i) (Equation 9) Here, α is a coefficient, and in the inventor's research, the optimal value is around 1.0.

[0042] In this invention, the estimated range of the distribution of physiological fluctuations is defined as the pulse velocity fluctuation d(i) or the pulse interval acceleration d 2 (i) Standard deviation is not used. This is a pulsed noise d(i) or d 2 (i) is because it often shows positive or negative values ​​with very large absolute values. The standard deviation is strongly affected by positive or negative values ​​with very large absolute values, which can lead to poor accuracy. For example, if there is a period of several minutes during which the pulse cannot be taken, the standard deviation will be very large. On the other hand, the TI of this invention, which is considered as a triangle with the height around the highest frequency F as its apex, is such a large noise d(i) or d 2 (i) is unaffected by the value of (i) and allows for accurate separation of peaks associated with physiological fluctuations (heart rate or pulse) from noise.

[0043] In the example above, TI was calculated as a triangle with vertices at the heights around the highest frequency F, but this is not the only way to determine TI. For example, in Figure 5, the triangle is an isosceles triangle with the highest frequency F as the center and base ±TI. However, depending on the highest frequency F and the state of the frequency distribution, the shape of the triangle may be changed according to the proportion of the frequency distribution. In other words, the highest frequency F is the vertex, but it does not necessarily have to be the center of the base. The shape of the triangle can be changed according to the state of the frequency distribution.

[0044] Furthermore, the histogram does not need to be triangular. Alternatively, we can use a different function g(F) (>0) with the highest frequency F as its vertex, which can encompass the histogram data. In that case, we determine the threshold (±g(F)) using a function g such that its derivative g'(F) < 0 (F>0). Furthermore, if the highest frequency F occurs in multiple intervals, depending on the state of the frequency distribution, for example, the interval where the average number of data points from the preceding and succeeding intervals falls may be designated as the interval with the highest frequency F. The data after noise reduction is stored in a memory unit (not shown) or provided to other devices via the communication unit 110.

[0045] [Examples of the present invention] An example of data after noise has been removed using the noise data removal method of the present invention will be described. Figure 6 shows time-series data obtained by detecting the R wave of an electrocardiogram and measuring the interval between each beat as the RR interval. In Figure 6, the horizontal axis represents minutes (min), and the vertical axis represents the RR wave interval (milliseconds). The time-series data shown in Figure 6 is time-series data of the RR wave interval that represents the standard for the true value of the pulse interval without noise. Physiologically, the average values ​​of the RR wave interval and the pulse interval per minute are the same, but high-frequency fluctuations are not the same. Therefore, the time-series data in Figure 6 is not a perfect match and should only be used as a guideline.

[0046] Figure 7 shows time-series data of pulse wave intervals obtained from pulse wave signals measured by an optical heart rate sensor installed on a wearable device (Silmee W22® (manufactured by TDK Corporation)) at the same time period as measured in Figure 6. The time-series data shown in Figure 7 is an example of pulse interval time-series data. Figure 7 contains multiple pulses and noise. Figure 8 shows time-series data obtained by removing noise data from the time-series data shown in Figure 7 using the noise reduction method of the present invention. In each figure, the horizontal axis represents minutes (min), and the vertical axis represents the pulse interval (milliseconds).

[0047] The interval time series data shown in Figure 7, like the time series data shown in Figure 6, appears to have a mode of 800 milliseconds. However, it can be seen that there are a very large number of data points reaching 1000 milliseconds, which is not seen in Figure 6. In other words, it is thought that there is a very large amount of noise data.

[0048] On the other hand, the time-series data shown in Figure 8 has data reaching 1000 milliseconds removed, and although there are some data with slightly larger amplitudes compared to the RR wave interval shown in Figure 6, the data is generally similar to the RR wave interval. In other words, it is time-series data with noise suitably removed. Note that, according to the noise data removal method of this embodiment, pulse wave intervals due to premature contractions and blocks are also removed as noise data, but fluctuations caused by respiration are not removed as noise data because the d(i) due to those fluctuations is included in the interval [-TI,TI].

[0049] [Processing steps of the present invention] Next, the processing flow of the pulsation analysis device 100 described above will be explained using a flowchart. Figure 9 is a flowchart of the processing flow of the pulsation analysis device 100. Note that the following explanation will use the fluctuating velocity d(i), but the acceleration d 2 (i) can be performed in the same manner. In Figure 9, the beat interval time series data acquisition unit 121 acquires beat interval time series data (t(i), x(i)) (step S101). The variation derivation unit 122 derives d(i) based on the beat interval time series data (t(i), x(i)) (step S102). In the process shown in Figure 9, the total number of d(i) is N (d(1) to d(N)).

[0050] The maximum frequency acquisition unit 123 acquires the bin width b (step S103), creates a histogram, and acquires the maximum frequency F (step S104). The bin width b is set by default and may be stored in the memory device or may be set by the operator of the pulse analysis device 100. The maximum frequency acquisition unit 123 selects the most rationally enclosing shape (function) and coefficient α with the maximum frequency F as the vertex of the created histogram and determines TI. If a triangle is selected as the shape and α is set to 1, TI = Nb / F is calculated. The removal unit 124 acquires TI = Nb / F (step S105).

[0051] The removal unit 124 initializes the loop counter k to 1 (step S106). The removal unit 124 determines whether d(k) < -TI or TI < d(k) (step S107). If -TI ≤ d(k) ≤ TI (step S107: NO), the removal unit 124 proceeds to step S109. If d(k) < -TI or TI < d(k) (step S107: YES), the removal unit 124 removes x(k) as noise data (step S108). The removal unit 124 increments the loop counter k by 1 (step S109) and determines whether k > N (step S110). If k > N (step S110: YES), the removal unit 124 ends the process. If k ≤ N (step S110: NO), the removal unit 124 returns to step S107.

[0052] [Others] In the embodiments described above, the method for measuring the inter-beat interval may be any method. Further, this embodiment is applicable not only to the inter-beat interval and the pulse interval, but also to the following time series data ((1) to (6)). (1) Time series data of the inter-beat interval or the pulse interval for each beat (2) Time series data of the instantaneous heart rate or the instantaneous pulse rate for each beat (3) Time series data of the instantaneous or average inter-beat interval at regular intervals (4) Time series data of the instantaneous or average pulse interval at regular intervals (5) Time series data of the instantaneous or average heart rate at regular intervals (6) Time series data of the instantaneous or average pulse rate at regular intervals Regarding the time series data shown in (1) to (6) above as (t(i), x(i)) in this embodiment, this embodiment can be applied by deriving d(i) or d 2 (i) by the first method or the second method.

[0053] Also, in the present invention, a histogram is created, but it is not necessary to create a histogram. Instead, simply obtaining the frequencies classified into a plurality of sections, obtaining the maximum frequency, and determining the range for removing noise by calculation processing may be sufficient.

[0054] The functions of the pulse analysis device 100 according to the above embodiment may be implemented using a wearable device. In this case, the wearable device would include a sensor for detecting pulse waves and the functions of the pulse analysis device 100. Alternatively, the functions of the pulse analysis device 100 may be implemented using a smartphone or server that acquires time-series data at intervals from the wearable device.

[0055] As described above, this embodiment provides a technique that can suitably remove noise from time-series data representing the time intervals between heartbeats or pulses.

[0056] The functions of the pulsation analyzer 100 in the above-described embodiment may be implemented using a computer. In that case, the functions may be implemented by recording a program for implementing these functions on a computer-readable recording medium, loading the program recorded on this recording medium into a computer system, and executing it. Here, "computer system" includes hardware such as an OS and peripheral devices. Furthermore, "computer-readable recording medium" refers to portable media such as flexible disks, magneto-optical disks, ROMs, CD-ROMs, and storage devices such as hard disks built into a computer system. Moreover, "computer-readable recording medium" may also include those that dynamically hold programs for a short period of time, such as communication lines used when transmitting programs via networks such as the Internet or communication lines such as telephone lines, and those that hold programs for a certain period of time, such as volatile memory inside a computer system that acts as a server or client in such cases. Furthermore, the above-mentioned program may only implement a part of the functions described above, and may also be able to implement the above-mentioned functions in combination with programs already recorded in the computer system.

[0057] While several embodiments of the present invention have been described, these embodiments are presented as examples only and are not intended to limit the scope of the invention. These embodiments can be carried out in a variety of other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their variations are included in the scope and spirit of the invention, as well as in the claims and their equivalents. [Explanation of symbols]

[0058] 100 Pulse analysis device, 120 Control unit, 121 Pulse interval time series data acquisition unit, 122 Fluctuation derivation unit, 123 Maximum frequency acquisition unit, 124 Removal unit, 141 Pulse interval time series data storage unit

Claims

1. A beat interval time series data acquisition unit acquires beat interval time series data for the duration of beat intervals, which are the time intervals between heartbeats or pulses. Based on the beat interval time series data acquired by the beat interval time series data acquisition unit, a variation derivation unit derives the rate of change of beat interval or the acceleration of beat interval from the beat interval time series data, A maximum frequency acquisition unit classifies the change rate or acceleration derived by the fluctuation derivation unit into multiple intervals and acquires the maximum frequency among each interval, A removal unit removes noise data from the beat interval time series data according to the maximum frequency acquired by the maximum frequency acquisition unit, Equipped with, The removal unit is a pulsation analyzer that, when the height of the highest peak in a histogram obtained by classifying the rate of change or the acceleration into multiple intervals is defined as the maximum frequency, and the height and base of an isosceles triangle are defined as the maximum frequency and the sides of the histogram along the axis representing the rate of change or the acceleration, respectively, the base of the isosceles triangle is determined based on the value obtained by dividing the area by the height when the area is expressed as the product of the bin width and the number of data points, and values ​​not enclosed by both ends of the base are removed as noise data.

2. The pulse analysis device according to claim 1, wherein the fluctuation derivation unit derives the rate of change from the difference between adjacent pulse intervals and the acceleration from the pulse intervals before and after the pulse interval.

3. The pulsation analysis device according to claim 1, wherein the maximum frequency acquisition unit creates a histogram classifying the change rate or acceleration into a plurality of intervals.

4. The pulsation analysis device according to claim 1, wherein the removal unit determines the noise to be removed using the total number of data in the beat interval time series data, the bin width of the interval, and the maximum frequency.

5. A step to acquire beat interval time series data that shows the time interval of heart rate or pulse, Based on the beat interval time series data acquired in the beat interval time series data acquisition step, a derivation step is performed to derive a variation derivation that derives the rate of change of beat interval or the acceleration of beat interval from the beat interval time series data, A maximum frequency acquisition step is performed to classify the rate of change of beat interval, or the acceleration of beat interval, derived in the above derivation step, into multiple intervals and obtain the maximum frequency among each interval. A removal step is performed to remove noise data from the beat interval time series data according to the maximum frequency obtained in the maximum frequency acquisition step, Equipped with, A pulsation analysis method in which, in the removal step, when the height of the highest peak in a histogram obtained by classifying the rate of change or the acceleration into multiple intervals is defined as the maximum frequency, and the height and base of an isosceles triangle are defined as the maximum frequency and the side along the axis representing the rate of change or the acceleration in the histogram, respectively, the base of the isosceles triangle is determined based on the value obtained by dividing the area by the height when the area is expressed as the product of the bin width and the number of data points, and values ​​not enclosed by both ends of the base are removed as noise data.

6. Computers, A beat interval time series data acquisition unit acquires beat interval time series data for the duration of beat intervals, which are the time intervals between heartbeats or pulses. Based on the beat interval time series data acquired by the beat interval time series data acquisition unit, a variation derivation unit derives the rate of change of beat interval or the acceleration of beat interval from the beat interval time series data, A maximum frequency acquisition unit classifies the change rate or acceleration derived by the fluctuation derivation unit into multiple intervals and acquires the maximum frequency among each interval, The unit that removes noise data from the beat interval time series data is configured to function as a removal unit, according to the maximum frequency obtained by the maximum frequency acquisition unit. The removal unit is a program that, when the height of the highest peak in a histogram obtained by classifying the rate of change or the acceleration into multiple intervals is defined as the maximum frequency, and the height and base of an isosceles triangle are defined as the maximum frequency and the sides of the histogram along the axis representing the rate of change or the acceleration, respectively, determines the base of the isosceles triangle based on the value obtained by dividing the area by the height when the area is expressed as the product of the bin width and the number of data points, and removes values ​​not enclosed by both ends of the base as noise data.

Images