Biosignal processing device, biosignal measuring device, and biosignal processing method

Abstract

A plurality of heartbeat waveforms extracted from time-series data of an ECG signal are divided into a plurality of heartbeat waveform groups including at least a first heartbeat waveform group and a second heartbeat waveform group. A first representative P-wave section waveform and a second representative P-wave section waveform, which are waveforms of representative P-wave sections, are respectively acquired from the first heartbeat waveform group and the second heartbeat waveform group. When the similarity between the first representative P-wave section waveform and the second representative P-wave section waveform is higher than a threshold value, it is determined that P-waves are included in the time-series data of the ECG signal.

Description

Biological signal processing device, biological signal measuring device, and biological signal processing method

[0001] The present invention relates to a biological signal processing device, a biological signal measuring device, and a biological signal processing method.

[0002] Atrial fibrillation (AF) is a type of arrhythmia in which abnormal electrical signals generated from a location other than the sinus node cause the atria to contract rapidly, resulting in an irregular pulse (heartbeat intervals). Atrial fibrillation not only causes palpitations and shortness of breath, interfering with daily life, but can also lead to serious illnesses such as cerebral infarction and heart failure. Therefore, early detection and appropriate treatment of atrial fibrillation are desirable.

[0003] During an attack of atrial fibrillation, two characteristic features are known to appear on an electrocardiogram (ECG): "disturbances in heartbeat intervals" and "disappearance of P waves." Therefore, to automatically detect atrial fibrillation using signal processing, it is necessary to accurately evaluate the disturbances in heartbeat intervals and detect P waves (determine the presence or absence of P waves) from the time-series data of the ECG signal. Here, the disturbances in heartbeat intervals can be detected relatively easily and accurately by fluctuations in the RRI (RR interval; the time from the peak of an R wave to the peak of the next R wave) in the ECG signal. However, accurately determining the presence or absence of P waves is quite difficult. This is because the amplitude of P waves is much smaller than that of R waves and they are easily contaminated by noise (making them difficult to distinguish from noise). Furthermore, individual differences in the waveform of P waves (amplitude, shape, direction (polarity), etc.) also increase the difficulty of detecting P waves.

[0004] Patent Document 1 discloses an idea to accurately detect P waves by generating a P wave template based on an ECG signal, extracting only waveforms that have a high cross-correlation with the template from each heartbeat waveform of the ECG signal, and averaging these.

[0005] U.S. Pat. No. 5,840,038

[0006] Saclova, L., Nemcova, A., Smisek, R. et al., "Reliable P wave detection in pathological ECG signals", Sci Rep 12, 6589 (2022).

[0007] Conventional detection algorithms have had cases where they have erroneously detected "P waves present" when they were not, or conversely, erroneously determined "P waves absent" when P waves were present. Such erroneous detections or erroneous determinations affect subsequent atrial fibrillation determinations, significantly reducing the accuracy and reliability of atrial fibrillation detection.

[0008] The present invention has been made in consideration of the above-mentioned circumstances, and an object of the present invention is to provide a novel technique for reliably determining the presence of P waves from time-series data of EGG signals by signal processing.

[0009] The present disclosure includes a biological signal processing device having an acquisition unit that acquires time series data of an ECG signal measured from a user, and a P-wave determination unit that extracts a first P-wave section waveform from a first P-wave section that is set based on the time point of a first R wave in the time series data of the ECG signal, and extracts a second P-wave section waveform from a second P-wave section that is set based on the time point of a second R wave different from the first R wave, and determines that a P wave is included in the first P-wave section and the second P-wave section when the similarity between the first P-wave section waveform and the second P-wave section waveform is higher than a threshold value.

[0010] The present disclosure includes a biological signal processing device having an acquisition unit that acquires time series data of an ECG signal measured from a user, and a P-wave determination unit that extracts at least a first heartbeat waveform group and a second heartbeat waveform group from the time series data of the ECG signal, extracts a waveform of a P-wave section set based on the time point of an R-wave from each heartbeat waveform of the first heartbeat waveform group, and extracts a waveform of a P-wave section set based on the time point of an R-wave from each heartbeat waveform of the second heartbeat waveform group, generates a first representative P-wave section waveform based on the waveform group of a plurality of P-wave sections extracted from the first heartbeat waveform group, and generates a second representative P-wave section waveform based on the waveform group of a plurality of P-wave sections extracted from the second heartbeat waveform group, and determines that a P-wave is included in the first heartbeat waveform group and the second heartbeat waveform group when the similarity between the first representative P-wave section waveform and the second representative P-wave section waveform is higher than a threshold.

[0011] The present disclosure includes a biological signal processing device having an acquisition unit that acquires time series data of an ECG signal measured from a user; a candidate detection unit that detects P-wave candidates from the time series data of the ECG signal; and a P-wave determination unit that, when a P-wave candidate is detected by the candidate detection unit, divides a plurality of heartbeat waveforms extracted from the time series data of the ECG signal into a plurality of heartbeat waveform groups including at least a first heartbeat waveform group and a second heartbeat waveform group, acquires a first representative P-wave section waveform and a second representative P-wave section waveform, which are waveforms of representative P-wave sections, from the first heartbeat waveform group and the second heartbeat waveform group, respectively, and determines that a P-wave is included in the time series data of the ECG signal if the similarity between the first representative P-wave section waveform and the second representative P-wave section waveform is higher than a threshold.

[0012] The P wave determination unit may determine that it is impossible to determine whether or not a P wave is included in the time series data of the ECG signal when the similarity between the first representative P wave section waveform and the second representative P wave section waveform is equal to or less than the threshold value.

[0013] The P wave determination unit may assign, among the plurality of heartbeat waveforms extracted from the time-series data of the ECG signal, those extracted in the first half to the first heartbeat waveform group and those extracted in the second half to the second heartbeat waveform group. Alternatively, among the plurality of heartbeat waveforms extracted from the time-series data of the ECG signal, the P wave determination unit may alternately or randomly select those to be assigned to the first heartbeat waveform group and those to be assigned to the second heartbeat waveform group.

[0014] The P wave determination unit may divide the multiple heartbeat waveforms extracted from the time-series data of the ECG signal into the first heartbeat waveform group and the second heartbeat waveform group so that the number of heartbeat waveforms in the first heartbeat waveform group is equal to the number of heartbeat waveforms in the second heartbeat waveform group.

[0015] The P-wave determination unit may extract a waveform of a P-wave section set based on the time point of an R wave from each heartbeat waveform of the first heartbeat waveform group, and obtain the average, median, or mode of the waveforms of the multiple P-wave sections extracted from the first heartbeat waveform group as the first representative P-wave section waveform, and extract a waveform of a P-wave section set based on the time point of an R wave from each heartbeat waveform of the second heartbeat waveform group, and obtain the average, median, or mode of the waveforms of the multiple P-wave sections extracted from the second heartbeat waveform group as the second representative P-wave section waveform.

[0016] The P-wave determination unit may evaluate the similarity between the first representative P-wave section waveform and the second representative P-wave section waveform based on the cross-correlation between the first representative P-wave section waveform and the second representative P-wave section waveform.

[0017] The candidate detection unit may generate a representative heartbeat waveform that is a representative heartbeat waveform from a plurality of heartbeat waveforms extracted from the time-series data of the ECG signal, and detect P-wave candidates from the representative heartbeat waveform.

[0018] The candidate detection unit may extract a feature from a P-wave section that is set based on the time point of an R-wave in the representative cardiac waveform, and determine that there is a P-wave candidate when the feature satisfies a predetermined condition.

[0019] The candidate detection unit may select data from the time-series data of the ECG signal in a low-noise section in which the degree of noise is lower than a predetermined level, and generate the representative heartbeat waveform using a plurality of heartbeat waveforms extracted from the data in the low-noise section.

[0020] The P wave determination unit may determine that it is impossible to determine whether or not a P wave is included in the time series data of the ECG signal if the number of heartbeat waveforms extracted from the data in the low noise section is less than a predetermined number.

[0021] When no P wave candidate is detected by the candidate detection unit, the P wave determination unit may evaluate the degree of smoothness of a P wave section that is set based on the time point of an R wave in the representative heartbeat waveform, and if the degree of smoothness is equal to or greater than a predetermined value, determine that no P wave is included in the time series data of the ECG signal, and otherwise determine that it is impossible to determine whether or not a P wave is included in the time series data of the ECG signal.

[0022] The present disclosure includes a biosignal measuring device having an ECG sensor attached to a user's limb, and a biosignal processing device that processes an ECG signal measured by the ECG sensor.

[0023] The present disclosure includes a biosignal processing method executed by a biosignal processing device, the biosignal processing method comprising the steps of: acquiring time series data of an ECG signal measured from a user; extracting a first P-wave section waveform from a first P-wave section set based on the time point of a first R-wave in the time series data of the ECG signal; extracting a second P-wave section waveform from a second P-wave section set based on the time point of a second R-wave different from the first R-wave; and determining that a P-wave is included in the first P-wave section and the second P-wave section if the similarity between the first P-wave section waveform and the second P-wave section waveform is higher than a threshold value.

[0024] The present disclosure includes a biological signal processing method executed by a biological signal processing device, the biological signal processing method comprising the steps of: acquiring time series data of an ECG signal measured from a user; extracting at least a first heartbeat waveform group and a second heartbeat waveform group from the time series data of the ECG signal; extracting a waveform of a P-wave section set based on the time point of an R-wave from each heartbeat waveform of the first heartbeat waveform group; extracting a waveform of a P-wave section set based on the time point of an R-wave from each heartbeat waveform of the second heartbeat waveform group; generating a first representative P-wave section waveform based on the waveform group of a plurality of P-wave sections extracted from the first heartbeat waveform group; and generating a second representative P-wave section waveform based on the waveform group of a plurality of P-wave sections extracted from the second heartbeat waveform group; and determining that a P-wave is included in the first heartbeat waveform group and the second heartbeat waveform group when a similarity between the first representative P-wave section waveform and the second representative P-wave section waveform is higher than a threshold value.

[0025] The present disclosure includes a biosignal processing method executed by a biosignal processing device, the biosignal processing method comprising the steps of: acquiring time series data of an ECG signal measured from a user; detecting P-wave candidates from the time series data of the ECG signal; and, when a P-wave candidate is detected, dividing a plurality of heartbeat waveforms extracted from the time series data of the ECG signal into a plurality of heartbeat waveform groups including at least a first heartbeat waveform group and a second heartbeat waveform group; acquiring a first representative P-wave section waveform and a second representative P-wave section waveform, which are waveforms of representative P-wave sections, from the first heartbeat waveform group and the second heartbeat waveform group, respectively; and determining that a P-wave is included in the time series data of the ECG signal if the similarity between the first representative P-wave section waveform and the second representative P-wave section waveform is higher than a threshold value.

[0026] The present invention may be understood as a biosignal processing device having at least a portion of the above-described configuration, or as a biosignal measuring device or electrocardiogram measuring device including an ECG sensor and a biosignal processing device. Alternatively, the present invention may be understood as an atrial fibrillation detection device, an atrial fibrillation recording device, or an atrial fibrillation monitoring device. Furthermore, the present invention may also be understood as a biosignal processing method including at least a portion of the above-described processing, a device control method, a program for implementing such a method, or a recording medium on which such a program is non-temporarily recorded. The above-described configurations and processing may be combined with each other to the greatest extent possible to constitute the present invention.

[0027] According to the present invention, it is possible to reliably determine the presence of a P wave from time-series data of an EGG signal by signal processing.

[0028] FIG. 1 is a diagram showing a biosignal measuring device worn on the upper arm. FIG. 2 is a plan view of the biosignal measuring device. FIG. 3 is a perspective view of the biosignal measuring device. FIG. 4 is a block diagram showing the hardware configuration of the biosignal measuring device. FIG. 5 is a block diagram showing the functional configuration of the biosignal measuring device. FIG. 6 is a flowchart showing the flow of measurement processing. FIG. 7 is a diagram showing an example of an electrode pair and an ECG signal. FIG. 8 is a diagram explaining the waveform of an ECG signal and heart rate information. FIG. 9 is a block diagram showing the functional configuration of a P-wave detection unit. FIG. 10 is a flowchart showing the flow of P-wave detection processing. FIG. 11 is a diagram explaining P-wave candidate detection. FIG. 12 is a diagram explaining the algorithm for P-wave presence determination processing. FIG. 13 is a diagram explaining the algorithm for P-wave presence determination processing according to a modified example.

[0029] <Application Example> One application example of the present invention will be described with reference to FIG.

[0030] The biosignal measuring device 1 is a portable measuring device that is worn on a limb and is capable of 24-hour monitoring of ECG signals (electrocardiogram signals). The biosignal measuring device 1 includes an ECG sensor that measures ECG signals, and a biosignal processing device that analyzes time-series data of the ECG signals measured by the ECG sensor and automatically detects atrial fibrillation.

[0031] During measurement, the user wraps the band 10 around the measurement site and wears the biosignal measuring device 1. Examples of measurement sites to which the band 10 is worn include the upper limbs (upper arm, forearm, wrist, hand, fingers) and the lower limbs (thigh, lower leg, ankle, foot, fingers), and the like, and the location is selected appropriately depending on the type of biosignal to be measured, the measurement algorithm, etc.

[0032] As described above, to improve the accuracy and reliability of atrial fibrillation detection, it is necessary to determine the presence or absence of P waves with high accuracy. Therefore, the biosignal measuring device 1 first detects P wave candidates from the ECG signal using a simple algorithm, and then performs detailed P wave determination processing to make a final determination of the presence or absence of P waves. The P wave determination processing may include a "P wave presence determination" performed when a P wave candidate is detected and a "P wave absence determination" performed when a P wave candidate is not detected. In the "P wave presence determination," two waveforms of the P wave section are obtained from the time-series data of the ECG signal for the target period, and if the similarity between the two waveforms is higher than a threshold, a "P wave presence" is determined; otherwise, a "P wave presence or absence cannot be determined" is determined. In the "P wave absence determination," the smoothness of the P wave section is calculated from the time-series data of the ECG signal for the target period, and if the smoothness is high, a "P wave absence" is determined; otherwise, a "P wave presence or absence cannot be determined" is determined. In this specification, the "P wave interval" refers to a time range that includes the time range in which a P wave occurs before an R wave when the heart rate is normal, but does not include other nearby electrocardiographic waveforms (e.g., Q waves or T waves).

[0033] By this processing, if P waves are included with a high degree of certainty in the ECG signal for the target period, it is determined that "P waves are present," and if P waves are not included with a high degree of certainty in the ECG signal for the target period, it is determined that "P waves are not present." If it is difficult to determine with a high degree of certainty due to the influence of noise, etc., the determination result that "the presence or absence of P waves cannot be determined" is output.

[0034] For example, if ECG signals labeled as "P wave presence or absence cannot be determined" are excluded and only ECG signals labeled as "P wave present" or "P wave absent" are used to detect atrial fibrillation, it is expected that the accuracy and reliability of atrial fibrillation detection can be improved compared to conventional methods.

[0035] Hereinafter, as one embodiment of the present invention, a specific configuration example in which the present invention is applied to an upper arm electrocardiogram device will be described.

[0036] <Embodiments> (Device Configuration) An embodiment of the present invention will be described with reference to Figures 1 to 4. Figure 1 is a diagram showing a state in which a biosignal measuring device 1 is worn on the upper arm, Figure 2 is a plan view of the biosignal measuring device 1, Figure 3 is a perspective view of the biosignal measuring device 1, and Figure 4 is a block diagram showing the hardware configuration of the biosignal measuring device 1.

[0037] The biosignal measuring device 1 of this embodiment is an upper arm electrocardiogram device that is worn on the user's upper arm (preferably the left upper arm close to the heart) and is used to measure ECG (Electrocardiogram) signals (electrocardiogram signals) as biosignals.

[0038] The biosignal measuring device 1 mainly comprises a band 10 , a plurality of electrodes 11 fixed to the band 10 , and a control unit 12 fixed to the band 10 .

[0039] The band 10 is a member (fixing member) for fixing the electrodes 11 in a state where they are pressed against the living body. In this embodiment, a strip-shaped band 10 made of a flexible and pliable material (e.g., synthetic fiber, silicone, leather, etc.) is used. A fixing mechanism 13 is provided at the longitudinal end of the band 10. As shown in Figures 1 and 3, the band 10 can be looped and fastened with the fixing mechanism 13, allowing the biosignal measuring device 1 to be worn on the upper arm. The fixing mechanism 13 may be any type of fastener, such as a hook and loop fastener, a hook, a connector, a button, or a magnet.

[0040] Multiple electrodes 11 (also called an electrode array) are embedded and fixed in the band 10 with their contact surfaces with the living body exposed on the inside (living body side) of the band 10. The multiple electrodes 11 are arranged in a row at equal intervals along the longitudinal direction of the band 10. This allows the electrodes 11 to contact different positions around the arm when the band 10 is wrapped around the arm. The number of electrodes 11 can be designed as desired. For the purpose of measuring ECG signals, at least two electrodes 11 (one electrode pair) are sufficient, but three or more electrodes 11 may be provided to improve the reliability and robustness of the measurement. In this embodiment, a configuration with six electrodes 11 (three electrode pairs) is adopted.

[0041] The electrodes 11 are dry metal electrodes. Wet electrodes (such as gel electrodes) have problems such as the possibility of causing skin rash or itching when worn for a long period of time, and low durability and ease of maintenance, whereas the dry electrodes 11 do not have such problems. The biosignal measuring device 1 of this embodiment is intended to be worn for a long period of time and to monitor ECG signals for 24 hours, so dry electrodes 11 are preferable.

[0042] The control main unit 12 is a processing unit that controls the biosignal measuring device 1 and performs signal processing. As shown in FIG. 4 , the control main unit 12 has a structure in which a measurement circuit 120 for measuring ECG signals, a biosignal processing device 121 that performs various signal processing on the ECG signals, a battery 122, and other circuits are mounted inside a case made of, for example, resin or metal. The measurement circuit 120 is an analog circuit composed of a differential amplifier and the like, and the ECG sensor 14 is composed of the multiple electrodes 11 and the measurement circuit 120. The biosignal processing device 121 is composed of, for example, an AD converter, a filter, a processor, a memory, and the like. The battery 122 is a rechargeable secondary battery. The control main unit 12 may be provided with an alarm device that provides alarms using a physical switch, a display, sound, light, vibration, electrical stimulation, or the like.

[0043] (Control Unit) FIG. 5 is a block diagram showing an example of the functional configuration of the control unit 12. As shown in FIG.

[0044] The control main unit 12 includes components related to ECG signal measurement: a measurement electrode selection unit 30, an ECG measurement unit 20, an ECG signal processing unit 21, an ECG heart rate information calculation unit 22, an AF determination unit 23, a storage unit 24, and a communication unit 25. The measurement electrode selection unit 30 selects an electrode pair to be used for measuring the ECG signal. The ECG measurement unit 20 amplifies the potential difference between the selected electrode pair using a differential amplifier and outputs the amplified potential difference as an ECG signal. The ECG signal processing unit 21 performs AD conversion and filtering of the ECG signal, and includes an AD conversion unit 210 that performs AD conversion of the ECG signal, an electromagnetic noise removal unit 211 that removes electromagnetic noise from the ECG signal to increase the S / N ratio, and a baseline drift removal unit 212 that removes baseline drift (low-frequency fluctuation) from the ECG signal. The ECG heart rate information calculation unit 22 extracts various heart rate information from the ECG signal and includes an R wave detection unit 220, an RRI calculation unit 221, a heart rate variability calculation unit 222, and a P wave detection unit 223. The AF determination unit 23 detects the occurrence of AF (Atrial Fibrillation) based on the heart rate information and calculates AF-related indicators. The storage unit 24 is a non-volatile memory that stores measured or calculated data. The communication unit 25 wirelessly communicates data with external devices (e.g., the user's smartphone, other health devices, a home server, etc.). Of the functional units shown in FIG. 5, the ECG measurement unit 20 is a function provided by the measurement circuit 120 in FIG. 4, and the other functional units are functions provided by the biological signal processing device 121 in FIG. 4.

[0045] (ECG signal measurement processing) The ECG signal measurement processing by the biological signal measurement device 1 will be described with reference to Fig. 6 to Fig. 8. Fig. 6 is a flowchart showing the flow of the ECG signal measurement processing, Fig. 7 is a diagram showing examples of electrode pairs and ECG signals, and Fig. 8 is a diagram explaining the waveform of the ECG signal and heart rate information.

[0046] When the user wraps the band 10 around the upper arm, fastens the band 10 with the fixing mechanism 13, and then performs an operation to instruct the start of measurement, the processor of the control unit 12 starts the measurement process shown in FIG.

[0047] In step S500, three-channel ECG signals are measured using three electrode pairs by the ECG measurement unit 20. Specifically, the electrode pair (two electrodes 11) to be used for measurement is selected by the measurement electrode selection unit 30, and the potential difference between the selected electrode pair is amplified by a differential amplifier and captured as an ECG signal (analog voltage signal). By switching the selected electrode pair in turn, three-channel ECG signals are captured.

[0048] In this embodiment, as shown in FIG. 7 , three electrode pairs are set so that two electrodes 11 that exactly face each other when the band 10 is wrapped around the upper arm are paired. This is because the greater the distance between the two electrodes 11 in a pair, the greater the potential difference (i.e., an ECG signal with a higher S / N ratio) that can be measured. Note that the distance between the two electrodes 11 refers to the linear distance between the two electrodes 11 when measured with the band 10 wrapped around the upper arm, as shown in FIG. 3 . However, the method for setting the electrode pairs is not limited to this. For example, a multiplexer may be used to freely switch the combination of paired electrodes 11. In this case, it is possible to measure ECG signals of four or more channels from six electrodes 11.

[0049] In step S501, the ECG signal processing unit 21 performs AD conversion on the ECG signal and removes electromagnetic noise and baseline fluctuations by digital signal processing. For removing the electromagnetic noise and baseline fluctuations, known noise reduction techniques such as a band-pass filter, a notch filter, or a moving average can be used. The ECG signal (digital data) obtained by the ECG signal processing unit 21 is stored in the storage unit 24 together with information on the measurement time.

[0050] In step S502, the ECG heart rate information calculation unit 22 retrieves time series data of the ECG signal for a unit period from the storage unit 24 and analyzes the time series data of the ECG signal to calculate various pieces of heart rate information. The length of the unit period is arbitrary and can be set to, for example, a length of several tens of seconds to several tens of minutes.

[0051] As shown schematically in FIG. 8 , the waveform of one heartbeat in an ECG signal is primarily composed of P waves, QRS waves, and T waves. The QRS wave refers to a waveform consisting of an upward peak R wave and the downward Q and S waves that appear before and after it. In a normal heart, electrical impulses generated in the sinus node are transmitted from the atria to the ventricles, causing atrial excitation (contraction) and ventricular excitation (contraction) in sequence, thereby pumping blood. The P wave corresponds to atrial excitation, the QRS wave corresponds to ventricular excitation, and the T wave corresponds to the process of the ventricles recovering from excitation. An ECG signal contains multiple heartbeat waveforms along the time axis. For an average adult, the heart rate is approximately 60 to 80 bpm, so for example, one minute of ECG signal time series data contains approximately 60 to 80 heartbeat waveforms.

[0052] Examples of heartbeat information include R amplitude (height of the R wave from the baseline), QRS amplitude (defined, for example, as the sum of the height of the R wave from the baseline and the depth of the S wave from the baseline), P amplitude (height of the P wave from the baseline), T amplitude (height of the T wave from the baseline), P width (time from the start of the P wave to the end of the P wave), QRS width (time from the start of the Q wave to the end of the S wave), T width (time from the start of the T wave to the end of the T wave), PQ time (time from the start of the P wave to the start of the Q wave), QT time (time from the start of the Q wave to the end of the T wave), RRI (RR interval; time from the peak of the R wave to the peak of the next R wave), PPI (PP interval; time from the start of the P wave to the start of the next P wave), and heartbeat variability (time fluctuation of the RRI and PPI). It is not necessary to calculate all of this heartbeat information; only necessary heartbeat information may be calculated. Heartbeat information other than these may also be calculated. The calculated heart rate information is stored in the storage unit 24 together with information on the measurement time.

[0053] In step S503, the AF determination unit 23 detects the occurrence of AF (atrial fibrillation) based on the heart rate information and calculates an index related to AF. The AF detection result and the calculated index are stored in the storage unit 24 together with information on the measurement time.

[0054] The measurement process shown in FIG. 6 described above is repeatedly executed for each unit period, thereby performing 24-hour monitoring of the ECG signal and AF.

[0055] (Details of P-Wave Detector) The configuration of the P-wave detector 223 and the P-wave detection algorithm will be described in detail with reference to FIGS.

[0056] 9 is a functional block diagram of the P-wave detection unit 223. The P-wave detection unit 223 has, as its main functions, an acquisition unit 80, a candidate detection unit 81, and a P-wave determination unit 82. The acquisition unit 80 has a function of acquiring time-series data of the ECG signal from the storage unit 24, the candidate detection unit 81 has a function of detecting P-wave candidates from the time-series data of the ECG signal, and the P-wave determination unit 82 has a function of making a detailed determination of the presence or absence of a P wave based on the detection result of the candidate detection unit 81 and outputting a final determination result.

[0057] FIG. 10 is a flowchart showing the flow of the P-wave detection process of the P-wave detector 223.

[0058] In step S900, the acquisition unit 80 acquires time-series data of the ECG signal from the storage unit 24. As mentioned above, P waves are easily contaminated by noise, so to ensure the accuracy and reliability of P-wave detection, it is preferable to statistically process time-series data over a certain period of time. In this embodiment, 10 minutes of time-series data is used for P-wave detection. This period of time that is the target of P-wave detection is referred to as the "target period" in this specification. The length of the target period can be set arbitrarily, for example, to a length of several tens of seconds to several tens of minutes.

[0059] In step S901, the candidate detection unit 81 calculates the noise level of the time-series data of the ECG signal for the target period. For example, the RMS (Root Mean Square) of the amplitude may be used as the noise level. In step S902, the candidate detection unit 81 removes data from high-noise intervals (intervals where the noise level is equal to or greater than a predetermined level) from the time-series data of the ECG signal for the target period, and selects only data from low-noise intervals (intervals where the noise level is lower than a predetermined level). By using only data from low-noise intervals, i.e., intervals where the heartbeat waveform appears relatively clearly, for subsequent processing, the reliability of candidate detection can be improved.

[0060] In step S903, the candidate detection unit 81 detects R waves from the data in the low-noise section. In step S904, the candidate detection unit 81 checks whether a predetermined number of R waves (i.e., heartbeat waveforms) or more have been detected from the data in the low-noise section. The "predetermined number" may be set arbitrarily, for example, to a value of approximately 20 to 100 (25 in this embodiment). If the number of R waves (i.e., the number of heartbeat waveforms) extracted from the data in the low-noise section is less than the predetermined number (No in step S904), the P-wave determination unit 82 outputs a result indicating that it is impossible to determine whether or not a P wave is included in the ECG signal for the target period (step S911). By classifying ECG signals that are so noisy that R waves (heartbeat waveforms) are difficult to detect as indeterminable, it is possible to prevent erroneous determination results from being output in the subsequent P-wave determination process, thereby improving the reliability of the device as a whole.

[0061] If a predetermined number or more of R waves are detected from the data in the low-noise section (Yes in step S904), the process proceeds to step S905. In step S905, the candidate detection unit 81 extracts waveforms within a predetermined range before and after the peak position of each detected R wave as heartbeat waveforms, and generates a representative heartbeat waveform from the extracted multiple heartbeat waveforms. Note that, in this specification, a "representative waveform" refers to a waveform that represents multiple waveforms in terms of shape. It may be a waveform that has commonality in terms of waveform shape, a waveform that has commonality when focusing on specific features, an average shape of multiple waveforms, or a waveform obtained by excluding exceptional waveforms that are significantly different in shape. For example, it may be an arithmetic average of multiple waveforms, a waveform that corresponds to the median or mode when multiple waveforms are evaluated using a specific index, or a waveform generated from multiple waveforms using other statistical methods. Generating a "representative waveform" allows for estimation of a waveform with even less noise. As a result, the reliability of P-wave determination can be improved. Similarly, accurate heart rate information can be obtained, improving the accuracy and reliability of atrial fibrillation detection.

[0062] In step S906, the candidate detection unit 81 detects P-wave candidates from the representative heartbeat waveform. Fig. 11 shows an example of P-wave candidate detection. Specifically, the candidate detection unit 81 first sets a P-wave section in the representative heartbeat waveform based on the peak position of the R wave in the representative heartbeat waveform. In this embodiment, as an example, the definition proposed in Non-Patent Document 1 is adopted, and the start and end points of the P-wave section are set according to the following formula:

[0063] Start point = R - (1 - 0.71) x RRmedian - 40 [msec] End point = R - 0.07 x RRmedian - 20 [msec] R: Peak position of the R wave of the representative heart rate waveform RRmedian: Median RRI during the target period However, the above formula is not limitative and the start and end points of the P wave section may be determined by other methods.

[0064] Next, the candidate detection unit 81 extracts a feature quantity related to amplitude from the waveform within the P-wave section. For example, as shown in Fig. 11, the candidate detection unit 81 selects the maximum or minimum point within the P-wave section as a candidate point cP for the P-wave peak, and calculates the feature quantity of the candidate point cP using the following formula:

[0065] Here, amp1 is the amplitude (peak value) of the candidate point cP relative to the rising point (onset) of the peak of the candidate point cP, and amp2 is the amplitude (peak value) of the candidate point cP relative to the falling point (offset) of the peak of the candidate point cP (see FIG. 11). The rising point (onset) and the falling point (offset) can be detected by finding the inflection points of the waveform. Furthermore, QRSamp is the amplitude from the minimum point to the maximum point of the QRS wave.

[0066] The feature value is the average of the rising amplitude amp1 and the falling amplitude amp2. This allows for more accurate evaluation of the P wave amplitude. Furthermore, by dividing the P wave amplitude by the QRS amplitude, normalization is performed using the QRS amplitude as a reference, so that the magnitude of the P wave amplitude can be appropriately evaluated regardless of the strength of the ECG signal level.

[0067] If the feature quantity thus obtained satisfies a predetermined condition, the candidate detection unit 81 adopts the peak of the candidate point cP as a "P wave candidate" and outputs a result indicating "P wave candidate present (P wave candidate detected)." On the other hand, if the feature quantity does not satisfy the predetermined condition, the candidate detection unit 81 outputs a result indicating "P wave candidate absent (P wave candidate not detected)." Based on the knowledge that the amplitude of a P wave is generally approximately 1% to 5% of the QRS amplitude, in this embodiment, the determination condition is set to "≧0.01." That is, if the value of the feature quantity is 0.01 or more (in other words, if the amplitude of the P wave section is 1% or more of the QRS amplitude), it is determined that "P wave candidate present."

[0068] The detection algorithm of the candidate detector 81 merely detects convex or concave peaks in a simplified manner, which may result in false detection of noise as P-wave candidates, or conversely, miss a P-wave buried in noise and erroneously determine that a P-wave candidate is not present. Therefore, the P-wave determiner 82 performs a detailed determination to make a final judgment on the presence or absence of a P-wave. In this embodiment, the P-wave determiner 82 switches the detailed determination process depending on the detection result of the candidate detector 81. That is, if a P-wave candidate is detected (Yes in step S907), the P-wave determiner 82 performs a "P-wave presence determination" (steps S908 to S909), whereas if a P-wave candidate is not detected (No in step S907), the P-wave determiner 82 performs a "P-wave absence determination" (steps S912 to S913). Specific examples of the detailed determination process are described below.

[0069] (1) P-Wave Presence Determination The inventors have found, from the results of analyzing ECG signals obtained in subject experiments, that although there are individual differences in P-wave waveforms, in the ECG signals of the same subject, the PR interval (the interval between the P wave and the R wave) is maintained approximately constant and the P-wave waveforms are reproducible (similar). Based on this finding, the inventors have come up with the idea that when a P-wave section is set based on the time point (e.g., peak position) of the R wave and the waveforms of the P-wave section in two heartbeat waveforms are compared, if a P wave is present, the similarity between the two waveforms will be high, but if it is not a P wave but a noise undulation, the similarity between the two waveforms will be low. This idea is embodied in the P-wave presence determination algorithm described below.

[0070] 12 shows an example of the algorithm for the P wave presence determination process in step S908. The P wave determination unit 82 first extracts multiple heartbeat waveforms from the time-series data of the ECG signal for the target period. In this case, similar to steps S901 to S905, heartbeat waveforms may be extracted from data in the low-noise section. Alternatively, the heartbeat waveform data extracted in step S903 may be stored in the storage unit 24, and the P wave determination unit 82 may use that data to determine the presence of P waves. Next, the P wave determination unit 82 divides the extracted multiple heartbeat waveforms into a first heartbeat waveform group and a second heartbeat waveform group. Here, for example, if N heartbeat waveforms are extracted from the target period, the N / 2 heartbeat waveforms extracted in the first half are assigned to the first heartbeat waveform group, and the N / 2 heartbeat waveforms extracted in the second half are assigned to the second heartbeat waveform group. By allocating the same number of heartbeat waveforms to the first heartbeat waveform group and the second heartbeat waveform group, the conditions for the subsequent statistical processing can be made uniform.

[0071] Next, the P-wave determination unit 82 acquires a waveform of a representative P-wave section from the first heartbeat waveform group (referred to as a "first representative P-wave section waveform"). For example, the P-wave determination unit 82 may extract waveforms of P-wave sections set based on the time point of the R-wave from each heartbeat waveform in the first heartbeat waveform group and average these to obtain the first representative P-wave section waveform. In this case, as with the generation of the representative heartbeat waveform in step S905, the waveforms of multiple P-wave sections may be averaged, or a median or mode may be calculated from the waveforms of multiple P-wave sections. The P-wave determination unit 82 also acquires a waveform of a representative P-wave section from the second heartbeat waveform group (referred to as a "second representative P-wave section waveform"). The P-wave determination unit 82 then calculates a cross-correlation coefficient r between the first representative P-wave section waveform and the second representative P-wave section waveform. This cross-correlation coefficient r is an example of an index representing the similarity between the first representative P-wave section waveform and the second representative P-wave section waveform. In this embodiment, cross-correlation is used to evaluate the similarity between the first and second representative P-wave section waveforms, but other methods or indices may be used to evaluate the similarity. For example, the similarity may be evaluated by comparing feature quantities extracted from each representative P-wave section waveform (P amplitude, P width, relative positions of the onset, peak position, and offset, P wave area, etc.).

[0072] In step S909, the P-wave determination unit 82 compares the cross-correlation coefficient r with a predetermined threshold th1. If the cross-correlation coefficient r is equal to or greater than the threshold th1 (Yes in step S909), the P-wave determination unit 82 determines that the first and second representative P-wave section waveforms are similar (reproducible), and outputs a result indicating that P waves are included in the ECG signal for the target period (step S910). On the other hand, if the cross-correlation coefficient r is less than the threshold th1 (No in step S909), the P-wave determination unit 82 determines that the first and second representative P-wave section waveforms are dissimilar, and that the P-wave candidates are likely to be noise, and outputs a result indicating that it is impossible to determine whether P waves are included in the ECG signal for the target period (step S911). Note that the threshold th1 may be set arbitrarily, for example, to a value that allows appropriate separation of P waves from noise based on the results of subject experiments (in this embodiment, the threshold th1 is set to 0.7).

[0073] (2) P-Wave Absence Determination Even if the candidate detection unit 81 does not detect a P-wave candidate, it is possible that the P-wave was overlooked due to being buried in noise. Therefore, the P-wave determination unit 82 evaluates the degree of smoothness of the P-wave section of the heartbeat waveform to determine whether or not the detection result of the candidate detection unit 81 can be adopted. If the degree of smoothness is high, the noise level of the ECG signal can be considered low (the S / N ratio is high), and therefore the reliability of the detection result of the candidate detection unit 81 can be determined to be high. Conversely, if the degree of smoothness of the P-wave section is low, the noise level of the ECG signal can be considered high (the S / N ratio is low), and therefore the reliability of the detection result of the candidate detection unit 81 can be determined to be low.

[0074] In this embodiment, the degree of smoothness s of the P wave section is expressed as an index using the following formula: degree of smoothness s=(QRS amplitude) / (RMS value of the numerical derivative of the waveform within the P wave section)

[0075] The denominator of the index (the RMS value of the numerical derivative of the waveform within the P-wave section) represents the magnitude of the change (unevenness) in the waveform during the P-wave section. Therefore, the smaller the change (flatter) in the waveform during the P-wave section, the smaller the degree of smoothness. Furthermore, since the index is normalized based on the QRS amplitude, the degree of smoothness during the P-wave section can be appropriately evaluated regardless of the strength of the signal level of the ECG signal.

[0076] Specifically, the P-wave determination unit 82 calculates the smoothness degree s of the P-wave period using the representative heartbeat waveform generated in step S905 (step S912). The P-wave determination unit 82 then compares the smoothness degree s with a predetermined threshold th2. If the smoothness degree s is equal to or greater than the threshold th2 (Yes in step S913), the P-wave determination unit 82 determines that the reliability of the detection result from the candidate detection unit 81 is high and outputs a result indicating that P waves are not included in the ECG signal for the target period (step S914). On the other hand, if the smoothness degree s is less than the threshold th2 (No in step S913), the P-wave determination unit 82 determines that the reliability of the detection result from the candidate detection unit 81 is low and outputs a result indicating that it is impossible to determine whether P waves are included in the ECG signal for the target period (step S911). The threshold th2 may be set arbitrarily, for example, to a value that allows appropriate separation of noise based on the results of subject experiments (in this embodiment, the threshold th2 is set to approximately 800).

[0077] According to the P wave detection algorithm of the present embodiment described above, the presence or absence of P waves can be determined with high reliability from the time series data of the ECG signal, thereby enabling automatic and accurate detection of diseases such as atrial fibrillation.

[0078] <Modifications> In the above embodiment, as shown in Fig. 12, the target period was divided into two parts, and the first heartbeat waveform group was extracted from the first half and the second heartbeat waveform group was extracted from the second half. However, the method of dividing the first heartbeat waveform group and the second heartbeat waveform group is not limited to this method. For example, as shown in Fig. 13, multiple heartbeat waveforms extracted from time-series data of ECG signals may be divided alternately into the first heartbeat waveform group and the second heartbeat waveform group (i.e., odd-numbered heartbeat waveforms may be divided into the first heartbeat waveform group and even-numbered heartbeat waveforms may be divided into the second heartbeat waveform group). Alternatively, instead of dividing the heartbeat waveforms alternately and regularly, the heartbeat waveforms may be divided randomly into the first heartbeat waveform group and the second heartbeat waveform group.

[0079] When time division is performed as shown in FIG. 12 , the extraction period of the first heartbeat waveform group and the extraction period of the second heartbeat waveform group do not overlap, whereas when the time division is performed alternately or randomly, both the first heartbeat waveform group and the second heartbeat waveform group are extracted from the entire target period.

[0080] For example, if a change in the P-wave waveform occurs over time, such as a reversal of the direction (polarity) of the P-wave between the first and second halves of the target period, the cross-correlation coefficient between the first and second representative P-wave section waveforms will be small, and the method of Figure 12 may result in a determination that a P-wave is "undetectable" despite its presence. In contrast, if the waveforms are alternately or randomly assigned, both the first and second heartbeat waveform groups are extracted from the entire target period, so that changes in the P-wave waveform over time affect both the first and second representative P-wave section waveforms in the same way, thereby suppressing a decrease in the cross-correlation coefficient. In other words, the P-wave presence determination method according to this modification has the advantage of being robust to changes in the P-wave waveform over time.

[0081] <Others> The above-described embodiments merely exemplify exemplary configurations of the present invention. The present invention is not limited to the specific embodiments described above, and various modifications are possible within the scope of the technical concept. For example, the fixing member for fixing the electrode 11 in a pressed state against the living body may be a band-shaped (strip-shaped) member that wraps around the measurement portion of the living body while the electrode 11 is in contact with the body, as in the above-described embodiment, or a wrap-shaped or ring-shaped member that covers and encases the measurement portion of the living body. Furthermore, the fixing member may be stretchable or deformable to accommodate the size (diameter) of the measurement portion of the living body. Alternatively, if made of a non-stretchable material, it may have a structure that allows its length to be adjusted. The number of electrodes 11 may be one or more. In the case of multiple electrodes 11, the arrangement does not need to be in a single row, but may be arranged in a two-dimensional array. Furthermore, the electrodes 11 may be arranged at varying intervals rather than at equal intervals. The electrodes 11 may be integrated into the fixing member (band 10) or may be a separate structure from the fixing member. The control unit 12 does not need to be fixed to the band 10. For example, the control body 12 and the band 10 may be separate structures, and the control body 12 and the band 10 (electrodes 11) may be connected by a cable.

[0082] In the above embodiment, the presence of a P wave is determined by generating a representative P-wave section waveform based on a group of waveforms of multiple P-wave sections extracted from multiple heartbeat waveforms and evaluating the similarity between the two representative P-wave section waveforms. However, it is not necessary to use a representative P-wave section waveform in the P-wave presence determination; a P-wave section waveform extracted from a single heartbeat waveform may be used to evaluate the similarity. In this case, it is sufficient to evaluate whether the similarity between the waveform of a first P-wave section set based on the time point of a first R-wave in the time series data of the ECG signal and the waveform of a second P-wave section set based on the time point of a second R-wave different from the first R-wave is higher than a threshold. In the above embodiment, the multiple heartbeat waveforms extracted from the time series data of the ECG signal are divided into two groups, a first heartbeat waveform group and a second heartbeat waveform group. However, the multiple heartbeat waveforms may be divided into, for example, three or more groups and compared with each other. Alternatively, the plurality of heartbeat waveforms may be divided into three groups, such as a first heartbeat waveform group, a second heartbeat waveform group, and a heartbeat waveform group to be excluded from the similarity evaluation.

[0083] This specification includes the following disclosure.

[0084] [Supplementary Note 1] A biological signal processing device (121) comprising: an acquisition unit (80) that acquires time series data of an ECG signal measured from a user; and a P-wave determination unit (82) that extracts a first P-wave section waveform from a first P-wave section that is set based on the time point of a first R-wave in the time series data of the ECG signal, and extracts a second P-wave section waveform from a second P-wave section that is set based on the time point of a second R-wave different from the first R-wave, and determines that a P-wave is included in the first P-wave section and the second P-wave section when the similarity between the first P-wave section waveform and the second P-wave section waveform is higher than a threshold.

[0085] [Supplementary Note 2] A biological signal processing device (121) comprising: an acquisition unit (80) that acquires time series data of an ECG signal measured from a user; and a P wave determination unit (82) that extracts at least a first heartbeat waveform group and a second heartbeat waveform group from the time series data of the ECG signal, extracts a waveform of a P-wave section set based on a time point of an R-wave from each heartbeat waveform of the first heartbeat waveform group, and extracts a waveform of a P-wave section set based on a time point of an R-wave from each heartbeat waveform of the second heartbeat waveform group, generates a first representative P-wave section waveform based on the waveform group of a plurality of P-wave sections extracted from the first heartbeat waveform group, and generates a second representative P-wave section waveform based on the waveform group of a plurality of P-wave sections extracted from the second heartbeat waveform group, and determines that a P wave is included in the first heartbeat waveform group and the second heartbeat waveform group when the similarity between the first representative P-wave section waveform and the second representative P-wave section waveform is higher than a threshold.

[0086] [Supplementary Note 3] A biological signal processing device (121) comprising: an acquisition unit (80) that acquires time series data of an ECG signal measured from a user; a candidate detection unit (81) that detects P-wave candidates from the time series data of the ECG signal; and a P-wave determination unit (82) that, when a P-wave candidate is detected by the candidate detection unit (81), divides a plurality of heartbeat waveforms extracted from the time series data of the ECG signal into a plurality of heartbeat waveform groups including at least a first heartbeat waveform group and a second heartbeat waveform group, acquires a first representative P-wave section waveform and a second representative P-wave section waveform, which are waveforms of representative P-wave sections, from the first heartbeat waveform group and the second heartbeat waveform group, respectively, and determines that a P-wave is included in the time series data of the ECG signal if a similarity between the first representative P-wave section waveform and the second representative P-wave section waveform is higher than a threshold.

[0087] [Supplementary Note 4] The biological signal processing device according to Supplementary Note 3, wherein the P wave determination unit determines that it is impossible to determine whether a P wave is included in the time-series data of the ECG signal when the similarity between the first representative P wave section waveform and the second representative P wave section waveform is equal to or less than the threshold value.

[0088] [Supplementary Note 5] The biological signal processing device according to Supplementary Note 3 or Supplementary Note 4, wherein the P wave determination unit assigns, among the multiple heartbeat waveforms extracted from the time series data of the ECG signal, those extracted in the first half to the first heartbeat waveform group and those extracted in the second half to the second heartbeat waveform group.

[0089] [Supplementary Note 6] The biological signal processing device according to Supplementary Note 3 or Supplementary Note 4, wherein the P wave determination unit alternately or randomly selects, from among a plurality of heartbeat waveforms extracted from the time series data of the ECG signal, those to be assigned to the first heartbeat waveform group and those to be assigned to the second heartbeat waveform group.

[0090] [Supplementary Note 7] The biological signal processing device according to any one of Supplementary Note 3 to Supplementary Note 6, wherein the P wave determination unit divides the multiple heartbeat waveforms extracted from the time series data of the ECG signal into the first heartbeat waveform group and the second heartbeat waveform group so that the number of heartbeat waveforms in the first heartbeat waveform group is equal to the number of heartbeat waveforms in the second heartbeat waveform group.

[0091] [Supplementary Note 8] The biological signal processing device according to any one of Supplementary Note 3 to Supplementary Note 7, wherein the P-wave determination unit extracts a waveform of a P-wave section set based on a time point of an R wave from each heartbeat waveform of the first heartbeat waveform group, and obtains an average, median, or mode of the waveforms of the multiple P-wave sections extracted from the first heartbeat waveform group as the first representative P-wave section waveform, and extracts a waveform of a P-wave section set based on a time point of an R wave from each heartbeat waveform of the second heartbeat waveform group, and obtains an average, median, or mode of the waveforms of the multiple P-wave sections extracted from the second heartbeat waveform group as the second representative P-wave section waveform.

[0092] [Supplementary Note 9] The biological signal processing device according to any one of Supplementary Note 3 to Supplementary Note 8, wherein the P-wave determination unit evaluates a similarity between the first representative P-wave section waveform and the second representative P-wave section waveform based on a cross-correlation between the first representative P-wave section waveform and the second representative P-wave section waveform.

[0093] [Supplementary Note 10] The biological signal processing device according to any one of Supplementary Note 3 to Supplementary Note 9, wherein the candidate detection unit generates a representative heartbeat waveform that is a representative heartbeat waveform from a plurality of heartbeat waveforms extracted from the time-series data of the ECG signal, and detects P-wave candidates from the representative heartbeat waveform.

[0094] [Supplementary Note 11] The biological signal processing device according to Supplementary Note 10, wherein the candidate detection unit extracts a feature from a P-wave section that is set based on a time point of an R-wave in the representative cardiac waveform, and determines that there is a P-wave candidate when the feature satisfies a predetermined condition.

[0095] [Supplementary Note 12] The biological signal processing device according to Supplementary Note 10 or Supplementary Note 11, wherein the candidate detection unit selects data from a low-noise section in which a noise level is lower than a predetermined level from the time-series data of the ECG signal, and generates the representative heartbeat waveform using a plurality of heartbeat waveforms extracted from the data from the low-noise section.

[0096] [Supplementary Note 13] The biological signal processing device according to Supplementary Note 12, wherein the P wave determination unit determines that it is impossible to determine whether or not a P wave is included in the time-series data of the ECG signal when the number of heartbeat waveforms extracted from the data in the low-noise section is less than a predetermined number.

[0097] [Supplementary Note 14] The biological signal processing device according to any one of Supplementary Note 10 to Supplementary Note 13, wherein, when the candidate detection unit does not detect a P wave candidate, the P wave determination unit evaluates the degree of smoothness of a P wave section set based on the time point of an R wave in the representative heartbeat waveform, and determines that a P wave is not included in the time series data of the ECG signal if the degree of smoothness is equal to or greater than a predetermined value, and otherwise determines that it is impossible to determine whether a P wave is included in the time series data of the ECG signal.

[0098] [Supplementary Note 15] A biosignal measuring device (1) comprising: an ECG sensor (14) attached to a user's limb; and a biosignal processing device (12) according to any one of Supplementary Note 1 to Supplementary Note 14, which processes an ECG signal measured by the ECG sensor (14).

[0099] [Supplementary Note 16] A biosignal processing method executed by a biosignal processing device (121), comprising: a step of acquiring time series data of an ECG signal measured from a user; and a step of extracting a first P-wave section waveform from a first P-wave section set based on a time point of a first R-wave in the time series data of the ECG signal, and extracting a second P-wave section waveform from a second P-wave section set based on a time point of a second R-wave different from the first R-wave, and determining that a P-wave is included in the first P-wave section and the second P-wave section when the similarity between the first P-wave section waveform and the second P-wave section waveform is higher than a threshold value.

[0100] [Supplementary Note 17] A biological signal processing method executed by a biological signal processing device (121), comprising the steps of: acquiring time series data of an ECG signal measured from a user; and extracting at least a first heartbeat waveform group and a second heartbeat waveform group from the time series data of the ECG signal, extracting a waveform of a P-wave section set based on a time point of an R-wave from each heartbeat waveform of the first heartbeat waveform group, and extracting a waveform of a P-wave section set based on a time point of an R-wave from each heartbeat waveform of the second heartbeat waveform group, generating a first representative P-wave section waveform based on the waveform group of a plurality of P-wave sections extracted from the first heartbeat waveform group, and generating a second representative P-wave section waveform based on the waveform group of a plurality of P-wave sections extracted from the second heartbeat waveform group, and determining that a P-wave is included in the first heartbeat waveform group and the second heartbeat waveform group when a similarity between the first representative P-wave section waveform and the second representative P-wave section waveform is higher than a threshold value.

[0101] [Supplementary Note 18] A biological signal processing method executed by a biological signal processing device (12), comprising: a step (S900) of acquiring time series data of an ECG signal measured from a user; a step (S906) of detecting P-wave candidates from the time series data of the ECG signal; and, when a P-wave candidate is detected, the steps (S908 to S911) of dividing a plurality of heartbeat waveforms extracted from the time series data of the ECG signal into a plurality of heartbeat waveform groups including at least a first heartbeat waveform group and a second heartbeat waveform group, acquiring a first representative P-wave section waveform and a second representative P-wave section waveform, which are waveforms of representative P-wave sections, from the first heartbeat waveform group and the second heartbeat waveform group, respectively, and determining that a P-wave is included in the time series data of the ECG signal if a similarity between the first representative P-wave section waveform and the second representative P-wave section waveform is higher than a threshold.

[0102] 1: Biosignal measuring device 10: Band 11: Electrode 12: Control unit 14: ECG sensor 121: Biosignal processing device

Claims

1. A biological signal processing device comprising: an acquisition unit that acquires time series data of an ECG signal measured from a user; and a P-wave determination unit that extracts a first P-wave section waveform from a first P-wave section set based on a time point of a first R-wave in the time series data of the ECG signal, and extracts a second P-wave section waveform from a second P-wave section set based on a time point of a second R-wave different from the first R-wave, and determines that a P-wave is included in the first P-wave section and the second P-wave section when the similarity between the first P-wave section waveform and the second P-wave section waveform is higher than a threshold value.

2. A biological signal processing device comprising: an acquisition unit that acquires time series data of an ECG signal measured from a user; and a P-wave determination unit that extracts at least a first heartbeat waveform group and a second heartbeat waveform group from the time series data of the ECG signal, extracts a waveform of a P-wave section set based on the time point of an R-wave from each heartbeat waveform of the first heartbeat waveform group, and extracts a waveform of a P-wave section set based on the time point of an R-wave from each heartbeat waveform of the second heartbeat waveform group, generates a first representative P-wave section waveform based on the waveform group of a plurality of P-wave sections extracted from the first heartbeat waveform group, and generates a second representative P-wave section waveform based on the waveform group of a plurality of P-wave sections extracted from the second heartbeat waveform group, and determines that a P-wave is included in the first heartbeat waveform group and the second heartbeat waveform group when a similarity between the first representative P-wave section waveform and the second representative P-wave section waveform is higher than a threshold value.

3. A biological signal processing device comprising: an acquisition unit that acquires time series data of an ECG signal measured from a user; a candidate detection unit that detects candidates for P waves from the time series data of the ECG signal; and a P-wave determination unit that, when a candidate for P waves is detected by the candidate detection unit, divides a plurality of heartbeat waveforms extracted from the time series data of the ECG signal into a plurality of heartbeat waveform groups including at least a first heartbeat waveform group and a second heartbeat waveform group, acquires a first representative P-wave section waveform and a second representative P-wave section waveform, which are waveforms of representative P-wave sections, from each of the first heartbeat waveform group and the second heartbeat waveform group, and determines that a P wave is included in the time series data of the ECG signal if a similarity between the first representative P-wave section waveform and the second representative P-wave section waveform is higher than a threshold value.

4. The biological signal processing device according to claim 3, wherein the P wave determination unit determines that it is impossible to determine whether or not a P wave is included in the time series data of the ECG signal when the similarity between the first representative P wave section waveform and the second representative P wave section waveform is equal to or less than the threshold value.

5. The biological signal processing device according to claim 3, wherein the P wave determination unit assigns, of the multiple heart rate waveforms extracted from the time series data of the ECG signal, those extracted in the first half to the first heart rate waveform group, and those extracted in the latter half to the second heart rate waveform group.

6. The biological signal processing device according to claim 3, wherein the P wave determination unit alternately or randomly selects, from among a plurality of heart rate waveforms extracted from the time series data of the ECG signal, those to be assigned to the first heart rate waveform group and those to be assigned to the second heart rate waveform group.

7. The biological signal processing device according to claim 3, wherein the P wave determination unit divides a plurality of heart rate waveforms extracted from the time series data of the ECG signal into the first heart rate waveform group and the second heart rate waveform group so that the number of heart rate waveforms in the first heart rate waveform group is equal to the number of heart rate waveforms in the second heart rate waveform group.

8. The biological signal processing device of claim 3, wherein the P wave determination unit: extracts a waveform of a P wave section set based on the time point of an R wave from each heartbeat waveform of the first heartbeat waveform group, and obtains an average, median or mode of the waveforms of the multiple P wave sections extracted from the first heartbeat waveform group as the first representative P wave section waveform; extracts a waveform of a P wave section set based on the time point of an R wave from each heartbeat waveform of the second heartbeat waveform group, and obtains an average, median or mode of the waveforms of the multiple P wave sections extracted from the second heartbeat waveform group as the second representative P wave section waveform.

9. The biological signal processing device according to claim 3, wherein the P wave determination unit evaluates the similarity between the first representative P wave section waveform and the second representative P wave section waveform based on the cross-correlation between the first representative P wave section waveform and the second representative P wave section waveform.

10. The biological signal processing device according to claim 3, wherein the candidate detection unit generates a representative heartbeat waveform that is a representative heartbeat waveform from a plurality of heartbeat waveforms extracted from the time series data of the ECG signal, and detects candidates for P waves from the representative heartbeat waveform.

11. The biological signal processing device according to claim 10, wherein the candidate detection unit extracts features from a P-wave section that is set based on the time point of an R-wave in the representative cardiac waveform, and determines that there is a P-wave candidate when the features satisfy a predetermined condition.

12. The biological signal processing device according to claim 10, wherein the candidate detection unit selects data from a low-noise section in which the noise level is lower than a predetermined level from the time series data of the ECG signal, and generates the representative heartbeat waveform using multiple heartbeat waveforms extracted from the data from the low-noise section.

13. The biological signal processing device according to claim 12, wherein the P wave determination unit determines that it is impossible to determine whether or not a P wave is included in the time series data of the ECG signal if the number of heart rate waveforms extracted from the data in the low noise section does not reach a predetermined number.

14. The biological signal processing device of claim 10, wherein the P wave determination unit, when no P wave candidate is detected by the candidate detection unit, evaluates the degree of smoothness of a P wave section set based on the time point of an R wave in the representative heartbeat waveform, and determines that no P waves are included in the time series data of the ECG signal if the degree of smoothness is equal to or greater than a predetermined value, and otherwise determines that it is impossible to determine whether or not a P wave is included in the time series data of the ECG signal.

15. A biosignal measuring device comprising: an ECG sensor attached to a user's limb; and a biosignal processing device according to any one of claims 1 to 14, which processes an ECG signal measured by the ECG sensor.

16. A biosignal processing method executed by a biosignal processing device, comprising: a step of acquiring time series data of an ECG signal measured from a user; and a step of extracting a first P-wave section waveform from a first P-wave section set based on a time point of a first R-wave in the time series data of the ECG signal, and extracting a second P-wave section waveform from a second P-wave section set based on a time point of a second R-wave different from the first R-wave, and determining that a P-wave is included in the first P-wave section and the second P-wave section if the similarity between the first P-wave section waveform and the second P-wave section waveform is higher than a threshold value.

17. A biosignal processing method executed by a biosignal processing device, comprising the steps of: acquiring time series data of an ECG signal measured from a user; extracting at least a first heartbeat waveform group and a second heartbeat waveform group from the time series data of the ECG signal, extracting a waveform of a P-wave section set based on the time point of an R-wave from each heartbeat waveform of the first heartbeat waveform group, and extracting a waveform of a P-wave section set based on the time point of an R-wave from each heartbeat waveform of the second heartbeat waveform group, generating a first representative P-wave section waveform based on the waveform group of a plurality of P-wave sections extracted from the first heartbeat waveform group, and generating a second representative P-wave section waveform based on the waveform group of a plurality of P-wave sections extracted from the second heartbeat waveform group, and determining that a P-wave is included in the first heartbeat waveform group and the second heartbeat waveform group when a similarity between the first representative P-wave section waveform and the second representative P-wave section waveform is higher than a threshold value.

18. A biosignal processing method executed by a biosignal processing device, comprising the steps of: acquiring time series data of an ECG signal measured from a user; detecting candidates for P waves from the time series data of the ECG signal; and, when candidates for P waves are detected, dividing a plurality of heartbeat waveforms extracted from the time series data of the ECG signal into a plurality of heartbeat waveform groups including at least a first heartbeat waveform group and a second heartbeat waveform group, acquiring a first representative P-wave section waveform and a second representative P-wave section waveform, which are waveforms of representative P-wave sections, from each of the first heartbeat waveform group and the second heartbeat waveform group, and determining that a P wave is included in the time series data of the ECG signal if a similarity between the first representative P-wave section waveform and the second representative P-wave section waveform is higher than a threshold value.

Images