JP6360017B2 - Heart rate detection method and heart rate detection device - Google Patents

Description

  The present invention relates to a heartbeat detection method and a heartbeat detection apparatus for extracting biological information such as a heartbeat interval (RR interval) from an electrocardiogram waveform, and more particularly, collectively processing a sampling data string of an electrocardiogram waveform accumulated in a storage medium. The present invention relates to a heartbeat detection method and a heartbeat detection device for detecting a heartbeat.

  Heart rate or its fluctuation is biological information obtained from ECG (Electrocardiogram) waveform, and it is used as an index of exercise intensity in sports-related fields, and for evaluation of autonomic nervous function in daily life and at rest. Is done.

  The ECG waveform is obtained by observing the electrical activity of the heart, and is measured by attaching electrodes to the body surface. There are various types of ECG waveform induction methods, i.e., electrode arrangements, and a stable waveform with a large amplitude can be obtained by placing the heart between the left and right chests, for example. When data processing such as heartbeat detection is performed on an ECG waveform, the ECG waveform is handled as a discrete data string (sampling data string) sampled at a constant time interval.

  FIG. 16 shows an example of an ECG waveform. In FIG. 16, the vertical axis represents potential and the horizontal axis represents time. The ECG waveform is composed of a continuous heartbeat waveform, and one heartbeat waveform is composed of components such as a P wave, a Q wave, an R wave, an S wave, and a T wave that reflect the activities of the atrium and the ventricle. Among them, the R wave is accompanied by ventricular contraction (ventricular muscle depolarization), and the amplitude is large. Therefore, the heartbeat is often detected using the R wave as a guide. In particular, it is easy to detect a heartbeat by taking a time difference of a sampling data string of an ECG waveform to make a sharp change from an R wave to an S wave stand out in a peak shape. The interval between heartbeats for each beat is called an RR interval, and is treated as a primary index of heartbeat variability.

  The following documents are known as conventional heartbeat detection methods. Patent Document 1 discloses a configuration for removing perturbation of the baseline of the ECG waveform. Patent Document 2 discloses a configuration for recognizing an R wave with a threshold value based on the amplitude of peaks and valleys of the waveform.

  Non-Patent Document 1 describes a method for obtaining an RR interval and the like based on a change in a value obtained by time difference of an ECG waveform. Specifically, when the sampling point of the ECG waveform is i (i is an integer equal to or greater than 1), the difference between the value of the (i + 1) th sampling data and the value of the (i-1) th sampling data is expressed as time. The difference value is obtained, the time when the value exceeds a certain threshold is recorded, and the interval between the times when the two consecutive thresholds are exceeded is defined as the RR interval.

JP 2002-78695 A JP 2003-561 A

“ECG Implementation on the TMS320C5515 DSP Medical Development Kit (MDK) with the ADS1298 ECG-FE”, Internet <http://www.ti.com/lit/an/sprabj1/sprabj1.pdf>

However, the above heart rate detection method has the following problems.
When noise is superimposed on the ECG waveform or its time difference, the method based on exceeding the threshold value may erroneously detect the heartbeat.
In addition, in the method in which the time exceeding the threshold value is regarded as the heartbeat time, each sampling data in the sampling data string is sequentially determined, so that a large amount of ECG waveform data accumulated in the storage medium is time-consuming for analysis. It will take.

  FIG. 17 is a graph showing a part of ECG waveform sampling data, where the horizontal axis represents time and the vertical axis represents cardiac potential [μV]. In this example, the level of the electrocardiogram is small due to the characteristics of the electrodes and the appliances used, and noise due to body movement is superimposed. In FIG. 17, it can be seen that an R wave appears at the location indicated by the arrow. The entire ECG waveform data in this example is sampled at intervals of 5 ms.

  FIG. 18 shows the time difference “(i + 1) th sampling data value− (i−1) th sampling data value” of the ECG waveform of FIG. 17, and further noise reduction by the effect of the low-pass filter. The target is a moving average process. Specifically, a total of 7 data is averaged for a section of ± 15 ms at each sampling point. Here, the horizontal axis represents time, and the vertical axis represents the difference value [μV] of the cardiac potential.

  The plot represented by “♦” in FIG. 18 indicates the heartbeat time detected based on the threshold value for the downward peak of the curve in FIG. 18 according to the method of Non-Patent Document 1, and RR based on it. An interval (right side of vertical axis, unit: [ms]) is represented. In the circled broken line, the noise is mistakenly detected as a heartbeat, and the actual heartbeat (shown by an arrow) immediately after it is skipped considering the maximum heart rate inserted after the heartbeat is detected. Since it is within the width of 240 ms, it is not detected. On the other hand, in the area surrounded by the solid line circle, the amplitude of the peak derived from the heartbeat is small, and it can be seen that it is missed. Further, the entire ECG waveform data in this example is 30 minutes sampled at intervals of 5 ms, and it is necessary to repeat the condition determination loop processing 360,000 times in order to detect heartbeats.

  The present invention has been made to solve such a problem, and an object of the present invention is to provide a method for accurately and quickly performing heartbeat detection on a sampling data string of an ECG waveform. To do.

  In order to achieve such an object, the present invention provides i + 1 for each sampling point i from a sampling data string of the electrocardiogram waveform of a living body where i is a sampling point accumulated in the storage medium (i is an integer of 1 or more). A time difference value calculation step for obtaining a value obtained by subtracting the value of the (i−1) th sampling data from the value of the first sampling data as the time difference value y (i), and before and after the sampling point i for each sampling point i A minimum value acquisition step for acquiring a minimum value z (i) of a time difference value in a predetermined time region, and for each sampling point i, the time difference value y (i) of the sampling point i is before and after the sampling point i. An index value calculating step for obtaining a value obtained by subtracting a minimum value z (i) of time difference values in a predetermined time region as an index value w (i); From the standard value w (i), the index value at a point that falls below a predetermined threshold value and the trend of the index value change starts from decreasing to increasing is specified as a downward peak, and the specified downward peak And a heartbeat time determination step using the time as the heartbeat time.

  When the time difference “(i + 1) th sampling data value− (i−1) th sampling data value” of the sampling data string of the ECG waveform is taken, a downward peak derived from the heartbeat appears, but below the threshold value. If it tries to detect it, it is easy to be influenced by noise. However, by subtracting the minimum value of the time difference value in a predetermined time region before and after the time difference value at a certain sampling point i (for example, a section having a width of 100 ms from a position 25 ms away before and after the sampling point i). , The peaks are more emphasized and become a sharp unimodal shape. In order to specify this peak (downward peak), the present invention applies a condition that the value falls below a predetermined threshold and starts to decrease and increases. As a result, batch processing can be performed without sequentially repeating the determination. Also, the time required for heartbeat detection can be greatly shortened. In addition, the heartbeat time can be detected with high accuracy while suppressing the influence of noise.

  In the present invention, when the trend of change in index value changes from decrease to increase, the index value at the point where change starts from decrease to increase is identified as a downward peak, and the trend of change in index value increases from level to level after increasing. In this case, the index value at the point where the level changes from leveling to the increase or the index value at the point where the level changes from leveling down to the level may be specified as a downward peak. In addition, among index values w (i) for each sampling point i, a positive value may be replaced with “0” to identify a downward peak. In the present invention, the time difference value y (i) at the sampling point i is obtained by performing a moving average process on a value obtained by subtracting the value of the (i−1) th sampling data from the value of the (i + 1) th sampling data for each sampling point i. It may be obtained as a value.

  According to the present invention, from the sampling data sequence of the electrocardiogram waveform of the living body with the sampling point i stored in the storage medium as i, for each sampling point i, from the value of the i + 1th sampling data, A value obtained by subtracting the value is obtained as a time difference value y (i), and for each sampling point i, a minimum value z (i) of the time difference value in a predetermined time region before and after the sampling point i is obtained. For each i, the index value w (i) is obtained by subtracting the minimum value z (i) of the time difference values in a predetermined time region before and after the sampling point i from the time difference value z (i) of the sampling point i. From among the index values w (i) for each sampling point i, the index value at a point that falls below a predetermined threshold and starts to decrease and increases is identified as a downward peak. Since the specified downward peak time is used as the heartbeat time, the sampling data string of the ECG waveform accumulated in the storage medium is processed in a batch while suppressing the influence of noise and accurately and quickly. The time can be detected.

It is a figure explaining how to obtain | require the time difference value for every sampling point in the process sequence at the time of detecting the heartbeat time by applying one Embodiment of the heartbeat detection method which concerns on this invention. It is a figure explaining how to obtain | require the minimum value in the predetermined time area | region for every sampling point. It is a figure which shows the predetermined time area | region before and behind a sampling point. It is a figure explaining how to obtain | require the index value (time difference value-minimum value in a predetermined time area | region) for every sampling point. It is a figure which compares and shows the case where the index value for every sampling point becomes "decrease-> increase" and the case where it becomes "decrease-> level-> increase." It is a figure explaining how to obtain | require the column vector which shows the position where the tendency of the change of an index value changes from decrease to increase. It is a figure explaining how to obtain | require the column vector which shows the position where the tendency of the change of an index value changes from decrease to leveling. It is a figure explaining how to obtain | require the column vector which shows the position of the downward peak from the column vector which shows the position where the tendency of the change of an index value turns from increase to the increase, and the column vector which shows the position which changes from decrease to the same level. It is a figure explaining how to obtain | require the column vector which shows the value of a downward peak. It is a figure explaining how to obtain | require the column vector which shows a heart-rate position. It is a figure explaining how to obtain | require the column vector which shows heartbeat time. It is a figure which shows the graph of the index value for every sampling point. It is a figure which expands and shows the graph of the time difference value for every sampling point partially. It is the figure which plotted the RR space | interval extracted by the conventional method and the heart rate detection method based on this invention. It is a functional block diagram of the principal part of the heart-rate detection apparatus to which the heart-rate detection method concerning the present invention is applied. It is a figure which shows the example of an ECG waveform. It is a figure which shows a part of graph of the sampling data of an ECG waveform. It is a figure explaining the conventional problem.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIGS. 1-11 is a figure explaining the process sequence at the time of detecting heartbeat time by applying one Embodiment of the heartbeat detection method based on this invention. In this processing procedure, a sampling data string of an ECG waveform is handled like a column vector to detect a heartbeat time.

  In FIG. 1A, [x (1), x (2),..., X (n)] is a sampling data string (column vector) of an ECG waveform acquired at a constant sampling interval (Δt). . A column vector of this ECG waveform is shifted by one (-1 shift) in the minus direction of the row and one shifted (+1 shift) in the plus direction (see FIG. 1 (b)). When the latter is subtracted from the former for each row, a column vector [y (1), y (2),... Y (n) of the time difference value y (i) = x (i + 1) −x (i−1) of the ECG waveform. ]] Is obtained (FIG. 1 (c)). Here, i is an arbitrary integer from 1 to n.

  Note that since cyclic shift is used here, both end portions (here, y (1) and y (n)) are different from the actual time difference values, but only from the entire data string. It is a part and hardly affects the detection of the heartbeat time. The same applies to other parts using the cyclic shift in this processing procedure.

  Next, column vectors [y (1), y (2),..., Y (n)] of time difference values are respectively shifted by k, k + L,. (See FIG. 2 (a)). As shown in FIG. 2 (a), these column vectors become an n × (m + 1) matrix, and when looking at the components for each row, from the sampling point i toward the future, k * Δt further, If the time domain of m * L * Δt toward the future (corresponding to the “predetermined time domain before sampling point i” in the present invention) is an integer greater than or equal to 2, the sampling interval of the sampling data string It is a set of values obtained by picking up the time difference values of the ECG waveform at intervals of L * Δt that are coarser than those (see FIG. 3).

  Also, consider an m + 1 column vector obtained by shifting a column vector [y (1), y (2),..., Y (n)] of time difference values by k, k + L,. (See FIG. 2 (b)). As shown in FIG. 2 (b), these column vectors become an n × (m + 1) matrix. When the components for each row are viewed, the sampling point i is further away from k * Δt toward the past. If the time domain of m * L * Δt toward the past (corresponding to the “predetermined time domain after the sampling point i” in the present invention) is an integer of 2 or more, the sampling interval of the sampling data string It is a set of values obtained by picking up the time difference values of the ECG waveform at intervals of L * Δt that are coarser than those (see FIG. 3).

  Note that the values of k, m, and L are, for example, k * Δt = 25 ms, m * L * Δt = 100 ms, and L * Δt = 10 ms when applied to a human ECG waveform. Is appropriate.

  2A and 2B, the minimum value for each row in FIG. 2A is obtained, and the minimum value is obtained from the time region k * Δt to k * Δt + m * L * Δt before and after the sampling point i. The time difference value in the time domain of “)” is the minimum value z (i). The column vector of the minimum value z (i) is [z (1), z (2),..., Z (n)] (see FIG. 2C).

  The minimum value z (i) of the time difference value in a predetermined time region before and after the sampling point i corresponds to the floor level before and after the sampling point i, and the time difference in the predetermined time region before and after the sampling point i. By comparing the minimum value z (i) with the time difference value y (i) at the sampling point i, the clearance before and after the sampling point i, that is, the unimodality of the peak can be evaluated. In addition, since it is unlikely that a spike-like abnormal value narrower than the 10 ms width other than the original R wave is included in the data string, L * Δt can be coarsened to about 10 ms, thereby improving the accuracy of heartbeat detection. The calculation load can be reduced without dropping.

  Next, the column vector [z (1), z () of the minimum value of the time difference value in a predetermined time domain from the column vector [y (1), y (2), ..., y (n)] of the time difference value. 2),..., Z (n)] are subtracted (see FIG. 4A), and the column vector [w (1), w (2) of “time difference value−minimum value of time difference value in a predetermined time domain” ,..., W (n)] are obtained (see FIG. 4B). Each value w (i) of this column vector corresponds to an “index value” in the present invention. Hereinafter, w (i) is referred to as an index value. In the subsequent processing, only a negative value of w (i) is required. Therefore, of the index values of the column vector whose value is positive, the value is replaced with “0”.

  Next, the index value column vector [w (1), w (2),..., W (n)] falls below a predetermined threshold value and the change tendency of the index value decreases. The index value of the point that starts to increase is identified as a downward peak. In this case, in fact, in the case of simply “decrease → increase” (FIG. 5 (a)), the same value continues twice at the peak apex, and the trend of change in the index value goes flat from the decrease. There is a case of “decrease → flat level → increase” which turns to increase (FIG. 5B).

  Here, when trying to identify a downward peak based on the condition of (w (i + 1) −w (i)) * (w (i) −w (i−1)) <0, “decrease → level → increase “(W (i + 1) −w (i)) * (w (i) −w (i−1)) ≦ 0”, there are a plurality of “decrease → level → increase”. Counts times. Therefore, in the present embodiment, the following is performed.

  First, for a column vector [w (1), w (2),..., W (n)] of index values, (w (i + 1) −w (i)) * (w (i) −w (i−1) )) To obtain a column vector [a (1), a (2),..., A (n)] with the flag “1” set at a negative value (FIGS. 6A, 6B, 6). c)). Further, a column vector [b (1), b (2),..., B (n)] in which a flag “1” is set where (w (i) −w (i−1)) is “0”. Is obtained (see FIGS. 7A and 7B). When the column vector [a (1), a (2),..., A (n)] and the column vector [b (1), b (2),..., B (n)] are ORed, A column vector [c (1), c (2),..., C (n)] indicating the position of the peak is obtained (see FIGS. 8A and 8B).

  The column vectors [w (1), w (2),..., W (n)] and the column vectors [c (1), c (2),. To obtain a column vector [p (1), p (2),..., P (n)] indicating the downward peak value of the index value (see FIGS. 9A and 9B).

  Further, a flag “1” is set where the value of the column vector [p (1), p (2),..., P (n)] falls below a predetermined threshold, and the column vector [q ( 1), q (2),..., Q (n)] are obtained (see FIGS. 10A and 10B).

  The column vector [q (1), q (2),..., Q (n)] and the column vector [t (1), t (2),. Are multiplied for each row to obtain a column vector [r (1), r (2),..., R (n)] representing the heartbeat time (see FIGS. 11A and 11B).

  As a result, when the trend of change in the index value w (i) changes from decrease to increase, the index value at the point where the change starts from decrease to increase is identified as a downward peak, and the change trend of the index value w (i) In the case where the value changes from a decrease to a level increase and then increases, the index value at the point where the level changes from level to level increases is specified as a downward peak. In this example, when the tendency of change in the index value w (i) changes from decreasing to leveling and increasing, the index value at the point where leveling to leveling increases increases is specified as a downward peak. You may make it specify the index value of the point which turns to level as a downward peak.

  FIG. 12 shows an example in which the heartbeat detection method according to the present invention is applied to the ECG waveform data of FIG. The curve of the graph represents the index value w (i), that is, “time difference value−minimum value of time difference value in a predetermined time region” for each sampling point i, with the horizontal axis representing time and the vertical axis representing potential [μV. ]. Here, the predetermined time region is a section having a width of 100 ms from a position 25 ms away before and after the sampling point i.

  This will be described with reference to FIG. FIG. 13 is a partially enlarged view of the plot of the time difference value y (i) of the ECG waveform (corresponding to FIG. 17). “♦” in the figure corresponds to sampling points at intervals of 5 ms. In the figure, there are three locations indicated by combinations of vertical arrows, broken lines, and horizontal arrows. Consider a section of 100 ms each indicated by a horizontal arrow from a point 25 ms apart from the sampling point indicated by the vertical arrow on both front and rear sides indicated by a broken line.

  The part indicated by the arrow (1) is a place where there is no heartbeat, and the value at the sampling point and the minimum value of the time difference value in the section of the horizontal arrow are the same value with a small absolute value. Yes. Therefore, the index value w (i) is subtracted between similar values, and is substantially close to zero (see arrow (1) in FIG. 12).

  Further, at the location indicated by the arrow (2), the sampling point itself is not a heartbeat, but a peak derived from a heartbeat is applied in the section of the horizontal arrow. Therefore, the time difference value at the sampling point is a value with a small absolute value, but the minimum value of the time difference value in the section indicated by the horizontal arrow is a negative value with a large absolute value. Accordingly, the index value w (i) is a positive value having a large absolute value because a negative value having a large absolute value is subtracted from a value having a small absolute value (see arrow (2) in FIG. 12).

  On the other hand, at the point indicated by the arrow (3), that is, at the peak peak derived from the heartbeat, the index value w (i) is obtained by subtracting a small absolute value from a negative value having a large absolute value. (See arrow (3) in FIG. 12).

  That is, the above-described operation produces an effect that the values at the steep peaks are kept as they are with respect to the original waveform, and the values at the periphery thereof are recessed in the opposite direction. As shown in FIG. 12, by setting the index value w (i) to “time difference value−minimum value of time difference value in a predetermined time region”, only a sharp peak derived from the heartbeat is further emphasized. You can see that it is easier to catch.

  By emphasizing in this way, the downward peak can be detected under the condition that the increase / decrease is reversed below a certain threshold, and the accumulated data is searched while scanning in the time direction. There is no need. In FIG. 12, the plot represented by “♦” represents the heartbeat time detected by applying the heartbeat detection method according to the present invention and the RR interval (right side of vertical axis, unit: [ms]) based on the heartbeat time. ing.

  Here, the threshold for the curve indicating the index value w (i) (FIG. 12) is the minimum value in that section every 5 seconds (because a heartbeat is always included in 5 seconds, As the minimum value, a value obtained by multiplying 0.5 by a peak value derived from a heartbeat) was used. A fixed value may be used according to the state of ECG waveform measurement.

  In the case of the ECG waveform of FIG. 17, the amplitude (excluding the base line perturbation) is about −300 to +300 μV depending on the characteristics of the electrodes used, and the downward peak value of the time difference value is approximately − Since it is about 400 μV, for example, by setting it to −150 μV, those peaks can be detected. It can be seen that the heartbeat detection method according to the present invention can accurately detect the heartbeat even in the case of erroneous detection or missing in the conventional method of FIG.

  FIG. 14 is a plot of RR intervals extracted from the longer time ECG waveform data including FIG. 17 by the conventional method and the heart rate detection method according to the present invention (horizontal axis: time, vertical axis: [Ms]). It can be seen that the accuracy of heartbeat detection is improved in the heartbeat detection method according to the present invention as compared with the conventional method. In addition, since the heartbeat detection method according to the present invention can be performed by batch processing based on matrix calculation, a result can be obtained in a very short time even for ECG waveform data that takes a long time.

  FIG. 15 shows a functional block diagram of the main part of a heartbeat detecting device to which the heartbeat detecting method according to the present invention is applied. The heartbeat detection device 100 is realized by hardware including a processor and a storage device, and a program that realizes various functions in cooperation with the hardware, and includes a memory (storage medium) 1 and a time difference value calculation unit. 2, a minimum value acquisition unit 3, an index value calculation unit 4, and a heartbeat time determination unit 5.

  In this heartbeat detection device 100, the memory 1 stores a sampling data string of an ECG waveform with a sampling point i. Further, the time difference value calculation unit 2 subtracts the value of the (i−1) th sampling data from the value of the (i + 1) th sampling data for each sampling point i from the sampling data string of the ECG waveform stored in the memory 1. Is obtained as a time difference value y (i).

  The minimum value acquisition unit 3 accesses the time difference value calculation unit 2 and acquires, for each sampling point i, the minimum value z (i) of time difference values in a predetermined time region before and after the sampling point i. The index value calculation unit 4 accesses the time difference value calculation unit 2 and the minimum value acquisition unit 3, and for each sampling point i, a predetermined value before and after the sampling point i from the time difference value y (i) of the sampling point i. A value obtained by subtracting the minimum value z (i) of the time difference values in the time domain is calculated as the index value w (i) (time difference value−the minimum value of the time difference values in the predetermined time domain).

  The heartbeat time determination unit 5 accesses the index value calculation unit 4, and the index value at the point where the index value w (i) for each sampling point i falls below a predetermined threshold and turns from decrease to increase. Is specified as a downward peak, and the time of the specified downward peak is set as a heartbeat time.

  In the above-described embodiment, the value of the (i−1) th sampling data is subtracted from the value of the (i + 1) th sampling data for each sampling point i from the sampling data string of the ECG waveform stored in the memory 1. Although the value is obtained as a time difference value, a value obtained by moving average processing a value obtained by subtracting the value of the (i + 1) th sampling data from the value of the (i + 1) th sampling data for each sampling point i is a time difference value y ( i) may be used.

[Extension of the embodiment]
The present invention has been described above with reference to the embodiment. However, the present invention is not limited to the above embodiment. Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the technical idea of the present invention.

  The present invention can be applied to a technique for detecting a heartbeat of a living body.

  DESCRIPTION OF SYMBOLS 1 ... Memory (storage medium), 2 ... Time difference value calculation part, 3 ... Minimum value acquisition part, 4 ... Index value calculation part, 5 ... Heartbeat time determination part, 100 ... Heartbeat detection apparatus.

Claims (8)

From the sampling data string of the electrocardiogram waveform of the living body in which the sampling point accumulated in the storage medium is i (i is an integer equal to or greater than 1), for each sampling point i, the value of the (i + 1) th sampling data is the i-1th. A time difference value calculating step for obtaining a value obtained by subtracting the value of the sampling data as a time difference value y (i);
A minimum value acquisition step of acquiring a minimum value z (i) of the time difference value in a predetermined time region before and after each sampling point i;
For each sampling point i, a value obtained by subtracting the minimum value z (i) of the time difference value in the predetermined time region before and after the sampling point i from the time difference value y (i) at the sampling point i. An index value calculating step for obtaining the index value w (i);
The index value at a point where the index value w (i) for each sampling point i falls below a predetermined threshold and the tendency of the index value to change starts from decreasing to increasing is specified as a downward peak. And a heartbeat time determination step using the specified downward peak time as a heartbeat time. The heartbeat detection method according to claim 1,
The heartbeat time determining step includes:
When the trend of change in the index value starts from decrease to increase, the index value at the point where decrease starts to increase is identified as the downward peak,
When the trend of the change in the index value changes from decreasing to leveling and increasing, the index value at the point where the level changes from leveling to increase or the index value at the point where the level changes from decreasing to leveling is specified as the downward peak. Heart rate detection method. The heartbeat detecting method according to claim 1 or 2,
The heartbeat time determining step includes:
A heartbeat detection method characterized in that, among index values w (i) for each sampling point i, a positive one is replaced with “0” to identify the downward peak. In the heart rate detection method according to any one of claims 1 to 3,
The time difference value calculating step includes:
The time difference value y (i) is obtained as a value obtained by subtracting the value of the (i + 1) th sampling data from the value of the (i + 1) th sampling data for each sampling point i, as a value obtained by moving average processing. Heart rate detection method. From the sampling data string of the electrocardiogram waveform of the living body in which the sampling point accumulated in the storage medium is i (i is an integer equal to or greater than 1), for each sampling point i, the value of the (i + 1) th sampling data is the i-1th. A time difference value calculating means for obtaining a value obtained by subtracting the value of the sampling data as a time difference value y (i);
Minimum value acquisition means for acquiring a minimum value z (i) of the time difference value in a predetermined time region before and after each sampling point i;
For each sampling point i, a value obtained by subtracting the minimum value z (i) of the time difference value in the predetermined time region before and after the sampling point i from the time difference value y (i) at the sampling point i. Index value calculating means for obtaining the index value w (i);
The index value at a point where the index value w (i) for each sampling point i falls below a predetermined threshold and the tendency of the index value to change starts from decreasing to increasing is specified as a downward peak. A heartbeat detection device comprising: a heartbeat time determination unit that uses the specified downward peak time as a heartbeat time. The heartbeat detecting device according to claim 5, wherein
The heartbeat time determining means includes
When the trend of change in the index value starts from decrease to increase, the index value at the point where decrease starts to increase is identified as the downward peak,
When the trend of the change in the index value changes from decreasing to leveling and increasing, the index value at the point where the level changes from leveling to increase or the index value at the point where the level changes from decreasing to leveling is specified as the downward peak. Heart rate detection device. The heartbeat detecting device according to claim 5 or 6,
The heartbeat time determining means includes
Among the index values w (i) for each sampling point i, a positive value is replaced with “0”, and the downward peak is specified. In the heartbeat detecting device according to any one of claims 5 to 7,
The time difference value calculating means includes
The time difference value y (i) is obtained as a value obtained by subtracting the value of the (i + 1) th sampling data from the value of the (i + 1) th sampling data for each sampling point i, as a value obtained by moving average processing. Heart rate detection device.

Images