Heart rhythm detection method and apparatus, electronic device, and storage medium

Abstract

Embodiments of the present application provide a heart rhythm detection method and device, electronic equipment and storage medium, belonging to the technical field of biomedical signal processing. The method comprises: acquiring a pulse wave signal of a user and an acceleration signal of a wrist of the user; obtaining a signal quality evaluation result according to amplitude fluctuation information of the pulse wave signal; obtaining a motion state evaluation result according to a change rate of the acceleration signal; if the signal quality evaluation result indicates that it is qualified and the motion state evaluation result indicates that it is stationary, determining a systolic phase position corresponding to each pulse wave in the pulse wave signal; determining a systolic phase beat-to-beat interval corresponding to each two pulse waves according to a time interval of each two adjacent systolic phase positions; obtaining a heart rate feature according to the systolic phase beat-to-beat interval corresponding to each two pulse waves; and determining a heart rhythm detection result according to the heart rate feature. The embodiments of the present application can improve the accuracy of arrhythmia detection.

Description

Technical Field

[0001] This application relates to the field of biomedical signal processing technology, and in particular to a method, device, electronic device and storage medium for heart rhythm detection. Background Technology

[0002] With the widespread adoption of wearable devices, heart rhythm monitoring technology based on photoplethysmography (PPG) signals has become an important tool for daily health management. PPG signals are acquired through optical sensors on the skin's surface and reflect changes in blood volume caused by heartbeats. They offer advantages such as being non-invasive, continuous, and convenient, making them particularly suitable for devices like smartwatches and fitness trackers. Arrhythmia is a common cardiovascular disease, and early detection is crucial for preventing serious complications such as stroke and heart failure. Traditional arrhythmia detection relies on electrocardiograms (ECGs), but these are inconvenient to wear, costly, and difficult to implement for long-term daily monitoring. Therefore, automated arrhythmia detection based on PPG signals has become an important development direction in the field of health sensing.

[0003] However, PPG signals are susceptible to motion artifacts, ambient light interference, and changes in skin blood supply, resulting in significant fluctuations in signal quality. Furthermore, the peak characteristics of the pulse wave are less precise than those of the ECG R wave. Current techniques typically calculate the beat interval directly based on the peak position of the PPG waveform. Because the waveform rise is gradual and the peak point is easily affected by noise, the interval extraction error often exceeds tens of milliseconds, leading to low accuracy in arrhythmia detection. Summary of the Invention

[0004] The main objective of this application is to provide a heart rhythm detection method, device, electronic device, and storage medium that can improve the accuracy of arrhythmia detection.

[0005] To achieve the above objectives, a first aspect of this application provides a method for detecting heart rhythm, the method comprising: Acquire the user's pulse wave signal and the user's wrist acceleration signal; Based on the amplitude fluctuation information of the pulse wave signal, the signal quality assessment result is obtained; The motion state assessment result is obtained based on the rate of change of the acceleration signal; If the signal quality assessment result indicates that it is qualified and the motion state assessment result indicates that it is stationary, then the systolic position corresponding to each pulse wave in the pulse wave signal is determined, and the systolic position is the time point when the systolic phase rises the fastest in a pulse wave. The systolic interval is determined based on the time interval between two adjacent systolic positions; Heart rate characteristics are obtained based on the systolic interval between every two pulse waves. The heart rate characteristics are used to determine the heart rhythm detection results.

[0006] In some embodiments, obtaining the signal quality assessment result based on the amplitude fluctuation information of the pulse wave signal includes: The average amplitude of each pulse wave in the pulse wave signal is calculated to obtain the average amplitude. The standard deviation of each pulse wave amplitude relative to the mean amplitude is calculated to obtain the amplitude standard deviation; If the amplitude standard deviation is greater than the preset quality threshold, the signal quality assessment result indicates that it is unqualified, and the process jumps to the step of obtaining the user's pulse wave signal and the acceleration signal of the user's wrist. If the amplitude standard deviation is less than or equal to the quality threshold, the signal quality assessment result indicates that it is qualified.

[0007] In some embodiments, obtaining the motion state assessment result based on the rate of change of the acceleration signal includes: The first derivative of the acceleration signal is calculated to obtain the user's motion amplitude; If the amplitude of the movement is greater than a preset static threshold, the motion state evaluation result indicates motion; If the amplitude of the movement is less than or equal to the static threshold, the motion state assessment result indicates static.

[0008] In some embodiments, determining the systolic position corresponding to each pulse wave in the pulse wave signal includes: For each sampling point in the pulse wave signal, the average slope between the sampling point and multiple sampling points before the sampling point is calculated, and the maximum value of the average slope is taken as the maximum average slope value of the sampling point. The slope change sequence is obtained based on the maximum average slope value of all sampling points in the pulse wave signal, and the peak position in the slope change sequence is determined as the systolic position of the pulse wave.

[0009] In some embodiments, the heart rate characteristics include heart rate variability characteristics and deep breathing characteristics; The method of obtaining heart rate characteristics based on the systolic interval between every two pulse waves includes: Based on the systolic interval, the heart rate variability features are extracted, and the heart rate variability features include at least one of time-domain features, frequency-domain features, Poincaré scatter plot features, and sample entropy features; The deep breathing characteristics are obtained by calculating the absolute median difference and the absolute mean difference of the differences between adjacent systolic heart beats.

[0010] In some embodiments, the calculation of the absolute median difference between adjacent systolic intervals and the absolute mean difference between adjacent systolic intervals to obtain the deep breathing characteristics includes: The difference between each pair of adjacent intervals is calculated to obtain a difference sequence composed of the differences between multiple pairs of adjacent intervals; Obtain the median of the difference sequence, and take the median of the absolute values ​​of the differences between each difference and the median as the absolute median difference; The average value of the difference sequence is calculated, and the absolute value of the difference between each difference and the average value is calculated. The average of the absolute values ​​of the differences between each difference and the average value is taken as the absolute mean difference. The difference between the absolute median and the absolute mean is used as the deep breathing feature.

[0011] In some embodiments, determining the heart rhythm detection result based on the heart rate characteristics includes: The machine learning model, obtained through pre-training, classifies and identifies the heart rate variability features and the deep breathing features to obtain the heart rhythm detection results. The heart rhythm detection results include normal heart rhythm or arrhythmia. The machine learning model has the ability to classify and identify based on the heart rate variability features and the deep breathing features.

[0012] To achieve the above objectives, a second aspect of this application provides a heart rhythm detection device, the device comprising: The acquisition module is used to acquire the user's pulse wave signal and the acceleration signal of the user's wrist. The first evaluation module is used to obtain a signal quality evaluation result based on the amplitude fluctuation information of the pulse wave signal; The second evaluation module is used to obtain the motion state evaluation result based on the rate of change of the acceleration signal; The determination module is used to determine the systolic position of each pulse wave in the pulse wave signal if the signal quality assessment result indicates that it is qualified and the motion state assessment result indicates that it is stationary. The systolic position is the time point when the systolic phase rises the fastest in a pulse wave. The first calculation module is used to determine the systolic interval between each two pulse waves based on the time interval between each two adjacent systolic positions. The second calculation module is used to obtain heart rate characteristics based on the systolic interval between every two pulse waves. The detection module is used to determine the heart rhythm detection result based on the heart rate characteristics.

[0013] To achieve the above objectives, a third aspect of this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method described in the first aspect.

[0014] To achieve the above objectives, a fourth aspect of the present application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in the first aspect.

[0015] The heart rhythm detection method, device, electronic equipment, and storage medium proposed in this application acquire the user's pulse wave signal and wrist acceleration signal. They evaluate signal quality based on the fluctuation information of the pulse wave amplitude and determine the motion state based on the rate of change of the acceleration signal. Only when the signal quality is acceptable and the wrist is stationary, they further determine the point of fastest systolic rise in each pulse wave as the systolic position. Then, based on the time interval between adjacent systolic positions, they calculate a high-precision systolic beat interval and extract heart rate features to output the heart rhythm detection result. This application introduces amplitude fluctuation information to filter high-quality signals and uses the rate of change of acceleration to eliminate motion interference. By using the point of fastest systolic rise instead of the traditional peak for interval calculation, it avoids the positioning error caused by the smooth peak being easily interfered with, significantly improving the measurement accuracy of the beat interval, thereby enhancing the accuracy of heart rate feature extraction and the robustness of the heart rhythm detection results. Attached Figure Description

[0016] Figure 1 This is a schematic flowchart of the heart rhythm detection method provided in the embodiments of this application; Figure 2 This is a flowchart of the arrhythmia detection scheme provided in the embodiments of this application; Figure 3 This is a schematic diagram of a qualified pulse wave signal provided in an embodiment of this application; Figure 4 This is a schematic diagram of a defective pulse wave signal provided in an embodiment of this application; Figure 5 This is a schematic diagram of the original photoplethysmography (PPG) signal waveform provided in the embodiments of this application; Figure 6 This is a schematic diagram of the maximum average slope change sequence waveform provided in the embodiments of this application; Figure 7 This is a schematic diagram of the heart rhythm detection device provided in the embodiments of this application; Figure 8 This is a schematic diagram of the hardware structure of an embodiment of the electronic device provided in this application; Figure 9This is a schematic diagram of the hardware structure of another embodiment of the electronic device provided in this application. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0018] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0020] With the widespread adoption of wearable devices, heart rate monitoring technology based on photoplethysmography (PPG) signals has become an important tool for daily health management. PPG signals are acquired through optical sensors on the skin's surface and can reflect changes in blood volume caused by heartbeats. They offer advantages such as being non-invasive, continuous, and convenient, making them particularly suitable for devices like smartwatches and fitness trackers.

[0021] Irregular heartbeat is a common cardiovascular disease, and early detection is crucial for preventing serious complications such as stroke and heart failure. Traditional arrhythmia detection relies on electrocardiograms, especially Holter monitoring, but these methods are inconvenient to wear, costly, and difficult to implement for long-term daily monitoring. Therefore, automated arrhythmia detection based on PPG signals has become a research hotspot in the field of health sensing in recent years.

[0022] However, PPG signals are susceptible to motion artifacts, ambient light interference, and changes in skin blood supply, resulting in significant signal quality fluctuations. The pulse wave peak characteristics are less precise than the R wave of ECG, directly impacting the accuracy of heartbeat interval extraction and the reliability of arrhythmia identification. Physiological activities (such as deep breathing) can cause regular heart rate changes (respiratory sinus arrhythmia), whose patterns can be easily confused with pathological disorders such as atrial fibrillation, leading to false alarms. Existing methods typically have limitations in signal quality assessment, feature extraction, and classification models, making it difficult to simultaneously achieve detection accuracy, interpretability, and real-time performance.

[0023] Based on this, embodiments of this application provide a heart rhythm detection method, device, electronic device, and storage medium, aiming to provide a heart rhythm detection method based on discrete event flow. By introducing a pulse wave amplitude standard deviation threshold method for PPG signal quality detection, it enhances the signal filtering capability under motion and environmental interference, providing a high-quality data foundation for subsequent analysis. It uses the systolic interval instead of the traditional peak interval as the beat interval, significantly improving the interval measurement accuracy (error reduced to within 2 milliseconds), laying the foundation for accurate analysis of heart rate variability. Addressing the problem that existing methods easily misjudge regular heart rate changes caused by deep breathing as atrial fibrillation, it introduces the absolute median difference based on the difference between adjacent intervals. This feature, as a robust statistic, can effectively distinguish between physiological heart rate fluctuations with a certain regularity and periodicity caused by deep breathing and the completely random and disordered pathological heart rate fluctuations exhibited by atrial fibrillation, thereby significantly reducing the false alarm rate and improving the accuracy, robustness, and real-time performance of arrhythmia detection.

[0024] The heart rhythm detection method, device, electronic device, and storage medium provided in the embodiments of this application are specifically described through the following embodiments. First, the heart rhythm detection method in the embodiments of this application is described.

[0025] The heart rhythm detection method provided in this application relates to the field of biomedical signal processing technology. The heart rhythm detection method provided in this application can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a wearable device such as a smartwatch, smart bracelet, or smart ring, or a smartphone, tablet, or portable health monitor; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application implementing the heart rhythm detection method, embedded firmware, or a functional module in health management software, but is not limited to the above forms.

[0026] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0027] It should be noted that in all specific embodiments of this application, when processing data related to user identity or characteristics, such as user information, user behavior data, user historical data, and user location information, user permission or consent is obtained first. Furthermore, the collection, use, and processing of this data comply with relevant laws, regulations, and standards. In addition, when embodiments of this application require access to sensitive personal information of users, separate permission or consent from the user is obtained through pop-ups or redirection to confirmation pages. Only after obtaining the user's separate permission or consent is the necessary user-related data required for the proper functioning of these embodiments acquired.

[0028] Figure 1 This is an optional flowchart of the heart rhythm detection method provided in the embodiments of this application. Figure 1 The method may include, but is not limited to, steps S100 to S700.

[0029] Step S100: Acquire the user's pulse wave signal and the acceleration signal of the user's wrist.

[0030] In wearable device-based heart rate monitoring scenarios, pulse wave signals, reflecting changes in blood volume caused by heartbeats, are typically acquired using optical sensors, while acceleration signals from the user's wrist are acquired using accelerometer sensors. In this embodiment, both pulse wave and acceleration signals are simultaneously acquired using wearable devices such as smartwatches or smart bracelets. Since pulse wave signals are highly susceptible to motion artifacts, after acquiring the raw signals, both pulse wave and acceleration signals need to be filtered separately to eliminate basic interference such as ambient light and high-frequency noise, providing a clean signal source for subsequent analysis.

[0031] Step S200: Obtain the signal quality assessment result based on the amplitude fluctuation information of the pulse wave signal.

[0032] The amplitude fluctuation of a pulse wave signal directly reflects the degree of interference. A qualified pulse wave signal has a uniform amplitude and a complete waveform; while a signal affected by motion or environmental interference has a chaotic waveform and large amplitude variations. Therefore, amplitude fluctuation information is quantified by detecting the amplitude of each pulse wave, calculating the mean of all pulse wave amplitudes, and then calculating the standard deviation of the difference between each pulse wave amplitude and the mean. The larger the standard deviation, the more chaotic the pulse wave amplitude and the worse the signal quality. In this embodiment, the signal quality assessment result can be obtained by comparing the calculated standard deviation with a preset quality threshold. This quality threshold is determined based on the statistical characteristics of a large number of normal high-quality signals and poor-quality signals. If the standard deviation is greater than the quality threshold, the signal quality assessment result indicates that it is unqualified and unsuitable for heart rhythm detection; if it is less than or equal to the quality threshold, the signal quality assessment result indicates that it is qualified.

[0033] Step S300: Obtain the motion state evaluation result based on the rate of change of the acceleration signal.

[0034] Because pulse wave signals are extremely sensitive to motion artifacts, the user's wrist must remain still during measurement. Any movement will cause a change in the acceleration signal; therefore, the amplitude of the movement, i.e., the rate of change of the acceleration signal, can be quantified by calculating the first derivative of the acceleration signal. The first derivative of the acceleration signal in a swaying state will be significantly greater than that in a stationary state. In this embodiment, the motion state evaluation result can be obtained by comparing the first derivative of the acceleration signal with a preset stationary threshold. This stationary threshold is determined based on the statistical characteristics of the first derivative values ​​in stationary and swaying states. If the first derivative value is greater than the stationary threshold, the motion state evaluation result indicates non-stationary; if it is less than or equal to the stationary threshold, the motion state evaluation result indicates stationary.

[0035] Step S400: If the signal quality assessment result indicates that the signal is qualified and the motion state assessment result indicates that the signal is stationary, then determine the systolic position of each pulse wave in the pulse wave signal. The systolic position is the time point when the systolic phase rises the fastest in a pulse wave.

[0036] In this embodiment, reliable heartbeat feature extraction can only be performed when the signal quality is acceptable and the wrist is stationary. Traditional methods calculate the interval by finding the peak of the pulse wave, but because the peak or trough of the pulse wave is relatively flat, it is easily affected by replay waves and noise, making it impossible to accurately determine the true peak position, and may even produce an interval error of hundreds of milliseconds. This embodiment uses the maximum average slope method to determine the systolic position. For each current sampling point in the pulse wave signal, the average slope between the current sampling point and multiple sampling points within a preceding window is calculated, and the maximum value is taken as the maximum average slope value of the current sampling point. After traversing the calculation, the peak of the average slope accurately corresponds to the position where the pulse wave rises the fastest during systole. The time point corresponding to the peak of this slope is determined as the systolic position, thereby avoiding the positioning error caused by the flat peak.

[0037] Step S500: Determine the systolic interval between every two pulse waves based on the time interval between every two adjacent systolic positions.

[0038] In this embodiment, after accurately locating the systolic position of each pulse wave, the time difference between two adjacent systolic positions is calculated. This time difference is the systolic interval. Since the systolic position is the exact point in time when the waveform rises the fastest, it is not affected by the flatness of the peak or the sampling accuracy. The systolic interval determined in this way can significantly reduce the measurement error to less than ten milliseconds, laying the foundation for accurate analysis of heart rate variability.

[0039] Step S600: Obtain heart rate characteristics based on the systolic interval corresponding to every two pulse waves.

[0040] In this embodiment, after obtaining the precise systolic heart rate interval, conventional heart rate variability features can be extracted, including but not limited to: time-domain features (e.g., the statistical dispersion of the interval sequence), frequency-domain features (e.g., the power of different frequency components of heart rate fluctuations), Poincaré scatter plot features (e.g., the nonlinear distribution pattern of the interval sequence), and sample entropy features (e.g., the complexity of the interval sequence). Furthermore, to effectively distinguish between the regular heart rate fluctuations caused by physiological deep breathing and the disordered heart rate fluctuations of pathological atrial fibrillation, this embodiment introduces deep breathing features. Physiological heart rate fluctuations caused by deep breathing have a certain regularity and periodicity, while pathological heart rate fluctuations such as atrial fibrillation are completely random and disordered.

[0041] Specifically, deep breathing characteristics are obtained by calculating the median of the absolute values ​​of the differences between adjacent heartbeats (i.e., absolute median difference) and the average of the absolute values ​​of the differences between adjacent heartbeats (i.e., absolute mean difference). The absolute median difference and absolute mean difference of the regular fluctuations caused by deep breathing are significantly smaller than the corresponding values ​​of pathological random fluctuations. Combining these deep breathing characteristics with conventional heart rate variability characteristics yields complete heart rate characteristics, which can effectively reflect the regular differences in heart rate fluctuations.

[0042] Step S700: Determine the heart rhythm detection result based on the heart rate characteristics.

[0043] In this embodiment, the extracted heart rate features are input into a pre-trained machine learning model (such as a random forest model, bagged tree model, or support vector machine model), which automatically determines whether the current signal segment belongs to a normal heart rhythm or an irregular heart rhythm. The detection results are finally presented to the user through the display module of the wearable device.

[0044] The flowchart of the arrhythmia detection protocol is as follows: Figure 2 As shown, firstly, photoplethysmography (PPG) and accelerometer (ACC) signals are simultaneously acquired through the PPG module and accelerometer sensor, and then filtered respectively.

[0045] Pulse wave signal quality detection: such as Figure 3 As shown, a qualified pulse wave signal has a uniform amplitude and a complete waveform; as Figure 4 As shown, the waveforms of unqualified pulse wave signals are chaotic and the amplitude variations are large. By detecting the amplitude PAn (Pulse Amplitude) of each pulse wave, the mean of all pulse wave amplitudes PAn is calculated and denoted as mPA. The calculation formula is shown in formula (1): (1); Next, calculate the standard deviation (stdPA) of the difference between each pulse wave amplitude PAn and the mean pulse wave amplitude mPA. The calculation formula is shown in formula (2): (2); A larger standard deviation (stdPA) indicates a more chaotic pulse wave amplitude, meaning a poorer PPG signal quality. Setting an appropriate threshold is crucial. The threshold is determined by calculating the variance stdP1 of the PPG waveform with good signal quality, then calculating the variance stdP2 of the PPG waveform with poor signal quality, and finally setting the threshold THR. stdPA = stdP1 + (stdP2 - stdP1) * 0.35. If If the PPG signal quality is not up to standard, it is considered unsuitable for arrhythmia detection. The measurement will be terminated and the process will restart for the next round of measurement.

[0046] Stationary Detection: Because PPG signals are easily affected by motion artifacts, the wrist must remain completely still during the measurement process. An accelerometer sensor is used to simultaneously acquire the ACC signal for stationary detection. Any movement will cause a change in the ACC signal; the first derivative of the ACC signal is calculated. Quantifiable movement amplitude. Set an appropriate threshold. , It is determined by the first derivative of ACC. By calculating the first derivative of ACC at rest (stdACC1) and the first derivative of ACC during swaying (stdACC2), the first derivative of ACC during swaying is significantly larger than that at rest. A threshold is then set. = stdACC1 +(stdACC2- stdACC1)*0.27, if If the measurement is not in a static state, the measurement will be terminated and the next round of measurement will be restarted.

[0047] Systolic position detection: Because the peak or trough position of the pulse wave is relatively flat, it is easily affected by replay waves and noise interference, which can introduce significant errors. For example... Figure 5 As shown, each increment of X by 1 represents 10 milliseconds. It's impossible to determine whether X=128 or X=148 represents the true peak pulse wave value. Incorrectly identifying the peak value will result in a 200-millisecond error in the heartbeat interval. To address this issue, this embodiment innovatively proposes a scheme for detecting systolic phase using the maximum average slope method.

[0048] The maximum average slope method refers to the ratio of the current sampling point PPGt to the sampling points PPGt within a certain period of time. The average slope of n (the tnth sampling point within the window, where n is typically between FS / 2 and FS / 6, and FS is the sampling rate) is taken as the maximum average slope value asPPGt for that point. The maximum average slope value asPPGt is calculated for all PPGt values, as shown in formula (3): asPPGt=MAX(PPGt PPGt nn)(3; like Figure 6As shown in the figure, after plotting asPPGt, it is clear that the peak of the average slope precisely corresponds to the position where the systole rises the fastest. Then, the systole interval (SSI) is calculated. The traditional method of calculating the systole interval is to find the difference between the positions of adjacent peaks of PPG. This method is affected by the accuracy of the sampling rate and the waveform of the peaks, and the error is relatively large. The peak of the PPG waveform processed by the average slope is found to be the position where the systole rises the fastest. The difference between two adjacent peaks is the systole interval. The systole interval calculated in this way can be accurate to within 10 milliseconds.

[0049] Once the accurate systolic heart rate interval (SSI) is obtained, conventional heart rate variability features can be extracted, including but not limited to time-domain features: mean heart rate interval (mean), standard deviation of all normal heart rate intervals (SDNN), root mean square of the difference between adjacent heart rate intervals (RMSSD), standard deviation of the difference between adjacent heart rate intervals (SDSD), percentage of adjacent heart rate intervals with a difference greater than 50 milliseconds (PNN50), and frequency-domain features: ultra-low frequency power (ULF), ratio of low frequency power to high frequency power (LF / HF), Poincaré scatter plot, sample entropy, etc.

[0050] Deep breathing characteristics: This embodiment introduces a feature based on differences between adjacent intervals. The median absolute deviation (medianAD) and mean absolute deviation (meanAD) are calculated using formulas (4) to (7): (4); (5); =MEDIAN( (6); = MEAN( (7); The absolute median difference of physiological heart rate fluctuations caused by deep breathing, which have a certain regularity and periodicity. and absolute mean difference All of these characteristics are significantly smaller than the completely random and disordered pathological heart rate fluctuations exhibited by atrial fibrillation. The introduction of these features can significantly reduce the false alarm rate.

[0051] Inputting all the above features into a machine learning model will identify whether the heart rhythm is normal or irregular. The machine learning model can be a random forest, bagged tree, boosting tree, support vector machine, etc., and is not limited here.

[0052] This embodiment effectively filters out interference signals by conducting a dual assessment of signal quality and motion status beforehand, thus improving data reliability from the source. It uses the fastest rising point of systole instead of the traditional peak to calculate the beat interval, avoiding peak positioning errors and significantly improving the accuracy of interval measurement. It extracts the absolute median difference and absolute mean difference features of the differences between adjacent intervals, effectively distinguishing between regular physiological fluctuations caused by deep breathing and random pathological fluctuations caused by atrial fibrillation, significantly reducing the false alarm rate. Overall, it achieves higher accuracy, stronger robustness, and better interpretability in arrhythmia detection.

[0053] In some embodiments, step S200 may include, but is not limited to, steps S210 to S240: Step S210: Calculate the average amplitude of each pulse wave in the pulse wave signal to obtain the average amplitude. Step S220: Calculate the standard deviation of each pulse wave amplitude relative to the mean amplitude to obtain the amplitude standard deviation; Step S230: If the amplitude standard deviation is greater than the preset quality threshold, the signal quality evaluation result indicates that it is unqualified, and the process jumps to the step of obtaining the user's pulse wave signal and the acceleration signal of the user's wrist. Step S240: If the amplitude standard deviation is less than or equal to the quality threshold, the signal quality assessment result indicates that it is qualified.

[0054] A pulse wave signal consists of a series of continuous pulse waves, each corresponding to a change in blood volume caused by a single heartbeat. In this embodiment, the amplitude values ​​of each pulse wave are identified and extracted from the preprocessed pulse wave signal. Specifically, the difference between the peak amplitude and the trough amplitude of each pulse wave is determined as its amplitude value. The amplitude values ​​of all identified pulse waves are summed and divided by the total number of pulse waves to calculate the mean amplitude. This mean amplitude reflects the overall intensity level of the pulse wave signal within the current detection period, providing a benchmark reference for subsequent evaluation of the dispersion of each pulse wave amplitude.

[0055] Pulse wave signals are susceptible to factors such as motion artifacts, ambient light interference, and changes in skin blood supply, leading to fluctuations in the amplitude of each pulse wave. Greater amplitude fluctuations indicate poorer signal quality. In this embodiment, the difference between the amplitude of each pulse wave and the mean amplitude is calculated, and then the standard deviation of all differences is calculated to obtain the amplitude standard deviation. This amplitude standard deviation quantifies the degree of dispersion of each pulse wave amplitude relative to the overall average level. A larger amplitude standard deviation indicates more chaotic pulse wave amplitudes, i.e., poorer pulse wave signal quality; a smaller amplitude standard deviation indicates more uniform pulse wave amplitudes, i.e., better pulse wave signal quality.

[0056] In this embodiment, after the amplitude standard deviation is calculated, it is compared with a preset quality threshold to output an evaluation result. The quality threshold is a critical point for distinguishing available signals from unavailable signals, and can be determined based on statistics of a large amount of historical sample data. Specifically, it can be determined by calculating the amplitude variance of normal high-quality signals and the amplitude variance of low-quality signals, and performing weighted interpolation between the two. For example, a certain proportion of the difference between the low-quality variance and the high-quality variance added to the high-quality variance is taken as the threshold, so as to both allow high-quality signals to pass and block low-quality signals.

[0057] In this embodiment, when the amplitude standard deviation is greater than the quality threshold, it indicates that the amplitude difference between each beat in the current pulse wave signal is too large, the waveform is disordered, and the signal quality is insufficient to support subsequent systolic positioning and heartbeat interval analysis. In this case, the signal quality evaluation result is determined as unqualified, and the detection process does not proceed downward, but directly jumps back to step S100 (that is, the step of acquiring the user's pulse wave signal and the acceleration signal of the user's wrist) to restart a new round of signal acquisition and evaluation. Through this feedback mechanism, the system can automatically discard signal segments with unqualified quality until a signal meeting the requirements is acquired.

[0058] In this embodiment, when the amplitude standard deviation does not exceed the quality threshold, it indicates that the amplitude of each pulse wave in the current pulse wave signal is relatively uniform, the waveform is complete, the interference received is small, and it has the value for further analysis. At this time, the signal quality evaluation result is determined as qualified. Through this qualified judgment, it is ensured that only high-quality pulse wave signals can enter the stationary state detection and subsequent heartbeat interval extraction links, laying a data foundation for accurate heart rhythm detection.

[0059] This embodiment quantitatively evaluates the quality of pulse wave signals by calculating the mean and standard deviation of pulse wave amplitudes; automatically screens qualified signals and eliminates unqualified signals by setting a reasonable quality threshold; automatically jumps to restart acquisition when the signal quality is unqualified, forming a closed-loop quality control mechanism; through the pulse wave amplitude standard deviation threshold method, it effectively solves the problem that pulse wave signals are susceptible to interference and have large quality fluctuations in the environment of wearable devices, enhances the signal screening capability under motion and environmental interference, provides a high-quality data foundation for subsequent heart rhythm detection, and effectively improves the accuracy and robustness of heart rhythm detection.

[0060] In some embodiments, step S300 may include, but is not limited to, steps S310 to S330: Step S310: calculating a first-order derivative of the acceleration signal to obtain the user's motion amplitude; Step S320: if the motion amplitude is greater than a preset stationary threshold, the motion state evaluation result indicates motion; Step S330: If the amplitude of the movement is less than or equal to the static threshold, the motion state evaluation result indicates static.

[0061] Pulse wave signals are easily affected by motion artifacts, requiring the wrist to remain completely still during measurement. In this embodiment, an acceleration signal is synchronously acquired using a wrist accelerometer sensor to detect wrist movement. The acceleration signal reflects the acceleration changes of the wrist in three-dimensional space; any body movement will cause fluctuations in the acceleration signal. To quantify the magnitude of the movement, the first derivative of the acceleration signal is calculated. Specifically, the ratio of the change in acceleration signal at adjacent sampling times to the time interval is calculated to obtain the first derivative of the acceleration signal. This first derivative reflects the rate of change of the acceleration signal; a larger rate of change indicates a more intense wrist movement, while a smaller rate of change indicates a more static wrist. The calculated first derivative is used as the user's movement amplitude for subsequent comparison with a static threshold.

[0062] In this embodiment, after calculating the amplitude of the movement, it is compared with a preset rest threshold to output the evaluation result. The rest threshold is an empirical value determined by statistical analysis of the first derivative of a large number of acceleration signals in a resting state, representing the maximum range of amplitude fluctuations that may occur in a resting state. When the amplitude of the movement exceeds this rest threshold, it indicates that the user's wrist has undergone a significant movement. At this time, the acquired pulse wave signal is highly susceptible to interference from motion artifacts, such as muscle contraction during movement, relative displacement between the skin and the sensor, and changes in the inertial distribution of blood in the limb. These factors can all add significant noise components to the pulse wave signal, leading to peak distortion and interval extraction distortion. In this case, the motion state evaluation result indicates movement, the detection process does not continue to the next step, but jumps back to the signal acquisition step to start a new round of measurement.

[0063] In this embodiment, when the amplitude of the movement does not exceed the resting threshold, it indicates that the user's wrist is in an acceptable state of rest. Under this condition, the motion state assessment result indicates rest, and the currently acquired pulse wave signal is considered to be largely unaffected by motion artifacts and can be safely used for subsequent systolic phase localization and beat interval analysis.

[0064] This embodiment quantifies the motion amplitude by calculating the first derivative of the acceleration signal, and automatically determines the stationary or moving state based on the static threshold. In the moving state, the detection is terminated and the user is guided to cooperate again, forming a closed-loop control mechanism for motion interference. This effectively solves the problems of pulse wave signal distortion and high false alarm rate of heart rhythm detection caused by the user's wrist movement in the wearable device environment, and effectively improves the anti-interference ability and reliability of heart rhythm detection.

[0065] In some embodiments, step S400 may include, but is not limited to, steps S410 to S420: Step S410: For each sampling point in the pulse wave signal, calculate the average slope between the sampling point and multiple sampling points before the sampling point, and take the maximum value of the average slope as the maximum average slope value of the sampling point. Step S420: Obtain the slope change sequence based on the maximum average slope value of all sampling points in the pulse wave signal, and determine the peak position in the slope change sequence as the systolic position corresponding to the pulse wave.

[0066] The pulse wave signal is a discrete digital signal continuously acquired at a fixed sampling rate, with each sampling point corresponding to the amplitude at a specific moment. Traditional methods use the peak point of the pulse wave as the heartbeat location, but because the peak region has a flat waveform and is easily affected by dicrotic waves and noise, the positioning error is relatively large. If the peak position is incorrectly selected, it will lead to a large error in the heartbeat interval. In this embodiment, every sampling point in the signal (not just the peak or trough of each pulse wave) is processed. For each current sampling point in the pulse wave signal, multiple sampling points within a fixed-length window are selected as starting points, and the average slope from each starting point to the ending point is calculated. Specifically, the average slope between the current sampling point and a forward sampling point within the window is the ratio of the difference in amplitude between the two points to the interval of the number of sampling points. By traversing all forward sampling points within the window, multiple average slope values ​​are obtained. The length of this window is determined according to the sampling rate, typically half to one-sixth of the sampling rate value, to ensure that the window covers the main part of the rising edge of the pulse wave while avoiding cross-cycle interference.

[0067] After calculating multiple average slopes between the current sampling point and each forward sampling point within the window, these maximum average slope values ​​are arranged in chronological order according to the sampling points to obtain a new sequence, namely the slope change sequence. This sequence reflects the local steepness of the rise of the original pulse wave signal at various time points: when the original pulse wave signal is in a flat segment, the corresponding maximum average slope value is small; when the original pulse wave signal is in an rising segment, the corresponding maximum average slope value increases; especially when the original pulse wave signal is in the position of the fastest rise and steepest waveform during the contraction phase, the maximum average slope value of that sampling point will reach a local maximum, forming a peak in the slope change sequence. All peak positions in the slope change sequence are detected. Each time a peak is detected, the original time point corresponding to that peak is a contraction phase position in the original pulse wave signal, that is, the time point of the fastest rise during the contraction phase of the pulse wave. Because the maximum average slope method utilizes the average information of multiple sampling points in the rising segment and accurately locates the steepest point by taking the maximum value, this method has significantly higher noise resistance and positioning accuracy compared to directly finding the peaks or valleys of the original pulse wave signal.

[0068] This embodiment uses the maximum average slope method to extract the slope change sequence and locate the systolic phase, effectively overcoming the positioning error problem caused by the smooth rise of the waveform or interference in traditional peak detection. It significantly improves the accuracy and robustness of systolic phase location detection, providing a reliable positional reference for obtaining high-precision systolic intervals.

[0069] In some embodiments, the heart rate characteristics include heart rate variability characteristics and deep breathing characteristics; step S600 may include, but is not limited to, steps S610 to S620: Step S610: Based on the systolic interval, extract the heart rate variability features, which include at least one of time-domain features, frequency-domain features, Poincaré scatter plot features, and sample entropy features. Step S620: Calculate the absolute median difference of the difference between adjacent systolic heart beat intervals and the absolute mean difference of the difference between adjacent systolic heart beat intervals to obtain the deep breathing characteristics.

[0070] In this embodiment, after obtaining the precise systolic interval, conventional heart rate variability (HRV) features can be extracted. HRV refers to the variation in the difference between successive heartbeats. It contains information on the regulation of the cardiovascular system by neurohumoral factors, thereby assessing the condition and prevention of cardiovascular diseases. It may be a valuable indicator for predicting sudden cardiac death and arrhythmic events. HRV features include at least one of the following: time-domain features, frequency-domain features, Poincaré scatter plot features, and sample entropy features.

[0071] Among them, time-domain features are indicators obtained by statistical analysis of multiple systolic intervals, including the mean and standard deviation of all systolic intervals, the root mean square of the difference between two adjacent systolic intervals, the standard deviation of the difference between adjacent systolic intervals, and the proportion of adjacent systolic interval differences exceeding a preset time difference. Time-domain features directly quantify the dispersion and regularity of heart rate changes from a time dimension. The heartbeat intervals of arrhythmias are often of varying lengths and fluctuate dramatically, and time-domain features can intuitively capture this irregularity.

[0072] Frequency domain characteristics involve spectral analysis of the systolic interbeat sequence, decomposing heart rate fluctuations into different frequency components and calculating the power distribution of each frequency band. Specifically, this includes low-frequency power, high-frequency power, and the ratio of low-frequency power to high-frequency power. Frequency domain characteristics reflect the state of the autonomic nervous system's regulation of heart rhythm. Under normal circumstances, the activities of the sympathetic and parasympathetic nervous systems are in dynamic equilibrium, and the energy distribution of different frequency components follows a specific pattern. When arrhythmia occurs, this balance is disrupted.

[0073] The Poincaré scatter plot features a scatter plot drawn using the interval between two adjacent systolic beats as coordinates, analyzing the distribution pattern, major axis, minor axis, and the ratio of the major axis to the minor axis. A normal heart rhythm scatter plot exhibits a comet-like distribution, while pathological arrhythmias alter this distribution. The Poincaré scatter plot features can reveal the intrinsic laws governing heart rhythm changes from a geometric perspective.

[0074] Sample entropy features are used to measure the unpredictability and disorder of systolic interbeat sequences. The more regular the sequence, the smaller the sample entropy value; the more complex and disordered the sequence, the larger the sample entropy value. Sample entropy features reflect the complexity changes of heart rhythm from the perspective of dynamic systems. The heart rate variability of a healthy heart exhibits moderate complexity, indicating that the autonomic nervous system can flexibly adjust the heart rate to adapt to external changes, and sample entropy features can effectively capture this complexity information.

[0075] The above four types of features describe different aspects of heart rate variability from multiple dimensions such as linear statistics, linear spectrum, nonlinear geometry, and nonlinear complexity. They can be used individually or in combination to more comprehensively and accurately assess the state of cardiac autonomic nervous function.

[0076] Physiological activities such as deep breathing cause regular heart rate changes, and these patterns can be easily confused with pathological disorders such as atrial fibrillation, leading to false alarms. To effectively distinguish between regular physiological fluctuations and pathological, disordered fluctuations, this embodiment first performs a difference calculation on adjacent systolic intervals. Specifically, the difference between the previous and subsequent systolic intervals is obtained. This process is repeated for all systolic intervals to obtain a sequence of adjacent interval differences. This sequence reflects the instantaneous fluctuations in heart rate, providing fundamental data for subsequent calculations of deep breathing characteristics.

[0077] In this embodiment, two statistical measures, the absolute median difference and the absolute mean difference, of adjacent interval difference sequences are calculated as characteristics of deep breathing. Specifically, the difference between every two adjacent systolic heart beat intervals is first calculated and its absolute value is taken. Then, the median of these absolute difference sequences is calculated to obtain the absolute median difference; simultaneously, the average of these absolute difference sequences is calculated to obtain the absolute mean difference. Physiological heart rate fluctuations caused by deep breathing have a certain regularity and periodicity, and the changes in adjacent heart beat intervals are relatively mild and orderly. Therefore, the calculated absolute median difference and absolute mean difference are significantly small. In contrast, heart rate fluctuations caused by pathological reasons such as atrial fibrillation are completely random and disordered, with drastic differences in adjacent heart beat intervals, and their absolute median difference and absolute mean difference will be significantly larger.

[0078] This embodiment comprehensively characterizes cardiac rhythm by extracting multi-dimensional heart rate variability features, including time domain, frequency domain, and nonlinearity. By calculating the absolute median difference and absolute mean difference of adjacent intervals as deep breathing features, it effectively distinguishes between physiological and pathological heart rate fluctuations. This effectively quantifies the regularity differences in heart rate fluctuations, accurately distinguishes between regular physiological heart rate fluctuations caused by deep breathing and random and disordered pathological heart rate fluctuations caused by atrial fibrillation, and significantly reduces the false alarm rate of arrhythmia detection.

[0079] In some embodiments, step S620 may include, but is not limited to, steps S621 to S624: Step S621: Calculate the difference between each pair of adjacent intervals to obtain a difference sequence composed of the differences between multiple pairs of adjacent intervals; Step S622: Obtain the median of the difference sequence, and take the median of the absolute values ​​of the differences between each difference and the median as the absolute median difference; Step S623: Calculate the average value of the difference sequence, and calculate the absolute value of the difference between each difference and the average value, and take the average of the absolute values ​​of the differences between each difference and the average value as the absolute mean difference; Step S624: The difference between the absolute median and the absolute mean is used as the deep breathing feature.

[0080] In this embodiment, the median of the adjacent interval difference sequence is first calculated. Then, the absolute deviation between each difference and the median is calculated, and the median of all absolute deviations is obtained to obtain the absolute median difference. The absolute median difference is a robust statistic that can effectively resist the interference of outliers. Physiological heart rate fluctuations caused by deep breathing, which have a certain regularity and periodicity, have adjacent interval differences that fluctuate slightly around the median, resulting in a small absolute median difference. In contrast, pathological heart rate fluctuations exhibited by atrial fibrillation, which are completely random and disordered, have adjacent interval differences that are scattered, resulting in a large absolute median difference.

[0081] Simultaneously, the mean of the adjacent interval difference sequence is calculated, which is the arithmetic mean of all differences. Then, the absolute deviation between each difference and the mean is calculated, and the mean of all absolute deviations is taken to obtain the absolute mean difference. Similar to the absolute median difference, the absolute mean difference is also used to quantify the dispersion of the adjacent interval difference sequence. The absolute mean difference of physiological regular fluctuations is significantly smaller than the absolute mean difference of pathological disordered fluctuations.

[0082] This embodiment obtains the difference sequence between adjacent intervals by calculating the difference one by one. Based on the median and the mean, the absolute median difference and the absolute mean difference are calculated respectively. These two are used as deep breathing features and fused with heart rate variability features. This solves the problem that existing methods have difficulty distinguishing between regular physiological heart rate changes caused by deep breathing and pathological disordered heart rate changes such as atrial fibrillation, and reduces the false alarm rate of arrhythmia detection.

[0083] In some embodiments, step S700 may include, but is not limited to, step S710: Step S710: The heart rate variability features and the deep breathing features are classified and identified by the pre-trained machine learning model to obtain the heart rhythm detection result. The heart rhythm detection result includes normal heart rhythm or arrhythmia. The machine learning model has the ability to classify and identify based on the heart rate variability features and the deep breathing features.

[0084] In this embodiment, a pre-trained machine learning model is used to classify and identify heart rate features. This machine learning model is trained on a large number of labeled samples and possesses the ability to classify and identify features based on the numerical distribution patterns of heart rate variability and deep breathing characteristics. The training samples include heart rate feature data for known heart rhythm states and their corresponding heart rhythm labels. Supervised learning is used to enable the model to learn the distribution patterns and discrimination boundaries of various features under different heart rhythm states. The machine learning model can be any of the classification algorithms such as random forest, bagged trees, boosting trees, and support vector machines; no limitation is made here. The heart rate feature set is input into this pre-trained model, and the model performs pattern matching and classification decisions based on the numerical combinations of each feature.

[0085] The machine learning model comprehensively evaluates the overall level of heart rate variability and the fluctuation regularity reflected by deep breathing characteristics, performs fusion decision-making, and outputs the final heart rhythm test result. The heart rhythm test result includes two categories: normal heart rhythm or arrhythmia. A normal heart rhythm indicates that no obvious signs of arrhythmia were found in the current signal segment; an arrhythmia indicates that the current signal segment conforms to the characteristics of a certain type of arrhythmia, such as atrial fibrillation or premature beats. The test results can be presented to the user through the wearable device's display module in the form of text, symbols, or warning messages, or can be further uploaded to a cloud server for doctors to view and analyze remotely.

[0086] This embodiment uses a pre-trained machine learning model to fuse and classify heart rate variability features and deep breathing features, making full use of the complementary information of multi-dimensional features. This effectively solves the problems of low accuracy of single feature recognition and false alarms caused by respiratory interference. Through automated and intelligent classification and recognition methods, it effectively improves the accuracy, robustness and real-time performance of heart rhythm detection, providing reliable technical support for daily heart rhythm health monitoring on wearable devices.

[0087] This application's embodiments assess signal quality by introducing a pulse wave amplitude standard deviation threshold, filtering out noise interference signals at the source and enhancing data filtering capabilities under motion and environmental interference. It replaces traditional peak location with the maximum mean slope method for detecting systolic position, significantly reducing the measurement error of the heart beat interval and providing high-precision basic data for heart rate variability analysis. It innovatively extracts the absolute median and absolute mean differences between adjacent heart beat intervals as deep breathing features, effectively distinguishing between regular physiological heart rate fluctuations and disordered pathological heart rate fluctuations, significantly reducing false alarm rates, thereby achieving higher precision, stronger robustness, and better interpretability for wearable arrhythmia detection.

[0088] Please see Figure 7 This application also provides a heart rhythm detection device 800, which can implement the above-described heart rhythm detection method. The device includes: The acquisition module 10 is used to acquire the user's pulse wave signal and the acceleration signal of the user's wrist; The first evaluation module 20 is used to obtain a signal quality evaluation result based on the amplitude fluctuation information of the pulse wave signal; The second evaluation module 30 is used to obtain the motion state evaluation result based on the rate of change of the acceleration signal; The determination module 40 is used to determine the systolic position of each pulse wave in the pulse wave signal if the signal quality assessment result indicates that it is qualified and the motion state assessment result indicates that it is stationary. The systolic position is the time point when the systolic phase rises the fastest in a pulse wave. The first calculation module 50 is used to determine the systolic interval between each two pulse waves based on the time interval between each two adjacent systolic positions. The second calculation module 60 is used to obtain heart rate characteristics based on the systolic interval between every two pulse waves. The detection module 70 is used to determine the heart rhythm detection result based on the heart rate characteristics.

[0089] The specific implementation of this heart rhythm detection device is basically the same as the specific embodiment of the heart rhythm detection method described above, and will not be repeated here.

[0090] This application also provides an electronic device; please refer to [link / reference]. Figure 8 , Figure 8 The illustration shows the hardware structure of an electronic device according to one embodiment. The electronic device includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the aforementioned heart rhythm detection method. This electronic device can be any wearable smart terminal, including smartwatches, smart bracelets, smart rings, smart headphones, etc.

[0091] The electronic device also includes a photoplethysmography (PPG) module and an accelerometer sensor (ACC). The PPG module is used to collect pulse wave signals through an optical sensor on the skin surface, and the accelerometer sensor is used to simultaneously collect acceleration signals from the wrist. When the processor executes the computer program, it processes the pulse wave signals and acceleration signals to obtain the heart rhythm detection results.

[0092] The electronic device also includes a display module, which is used to output the heart rhythm detection results and present the user with the identification results of normal or irregular heart rhythm. The display module can be implemented in the form of an LCD screen, an organic light-emitting diode screen, or an e-ink screen.

[0093] Please see Figure 9 , Figure 9 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes: The processor 801 can be implemented using a general-purpose central processing unit (CPU), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory 802 can be implemented as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 802 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 802 and is called and executed by the processor 801 to execute the heart rhythm detection method of the embodiments of this application. The 803 input / output interface is used to implement information input and output. The communication interface 804 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 805 transmits information between various components of the device (e.g., processor 801, memory 802, input / output interface 803, and communication interface 804); The processor 801, memory 802, input / output interface 803, and communication interface 804 are connected to each other within the device via bus 805.

[0094] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described heart rhythm detection method.

[0095] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0096] The heart rhythm detection method, device, electronic equipment, and storage medium provided in this application acquire the user's pulse wave signal and wrist acceleration signal. They evaluate signal quality based on the fluctuation information of the pulse wave amplitude and determine the motion state based on the rate of change of the acceleration signal. Only when the signal quality is acceptable and the wrist is stationary, they further determine the point of fastest systolic rise in each pulse wave as the systolic position. Then, based on the time interval between adjacent systolic positions, they calculate a high-precision systolic interval and extract heart rate features to output the heart rhythm detection result. This application introduces amplitude fluctuation information to filter high-quality signals and uses the rate of change of acceleration to eliminate motion interference. By using the point of fastest systolic rise instead of the traditional peak for interval calculation, it avoids the positioning error caused by the smooth peak being easily interfered with, significantly improving the measurement accuracy of the heart rate interval, thereby enhancing the accuracy of heart rate feature extraction and the robustness of the heart rhythm detection results.

[0097] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0098] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0099] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0100] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0101] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0102] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0103] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0104] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0105] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0106] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0107] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A method for detecting heart rhythm, characterized in that, The method includes: Acquire the user's pulse wave signal and the user's wrist acceleration signal; Based on the amplitude fluctuation information of the pulse wave signal, the signal quality assessment result is obtained; The motion state assessment result is obtained based on the rate of change of the acceleration signal; If the signal quality assessment result indicates that it is qualified and the motion state assessment result indicates that it is stationary, then the systolic position corresponding to each pulse wave in the pulse wave signal is determined, and the systolic position is the time point when the systolic phase rises the fastest in a pulse wave. The systolic interval is determined based on the time interval between two adjacent systolic positions; Heart rate characteristics are obtained based on the systolic interval between every two pulse waves. The heart rate characteristics are used to determine the heart rhythm detection results.

2. The method according to claim 1, characterized in that, The step of obtaining the signal quality assessment result based on the amplitude fluctuation information of the pulse wave signal includes: The average amplitude of each pulse wave in the pulse wave signal is calculated to obtain the average amplitude. The standard deviation of each pulse wave amplitude relative to the mean amplitude is calculated to obtain the amplitude standard deviation; If the amplitude standard deviation is greater than the preset quality threshold, the signal quality assessment result indicates that it is unqualified, and the process jumps to the step of obtaining the user's pulse wave signal and the acceleration signal of the user's wrist. If the amplitude standard deviation is less than or equal to the quality threshold, the signal quality assessment result indicates that it is qualified.

3. The method according to claim 1, characterized in that, The step of obtaining the motion state assessment result based on the rate of change of the acceleration signal includes: The first derivative of the acceleration signal is calculated to obtain the user's motion amplitude; If the amplitude of the movement is greater than a preset static threshold, the motion state evaluation result indicates motion; If the amplitude of the movement is less than or equal to the static threshold, the motion state assessment result indicates static.

4. The method according to claim 1, characterized in that, Determining the systolic position corresponding to each pulse wave in the pulse wave signal includes: For each sampling point in the pulse wave signal, the average slope between the sampling point and multiple sampling points before the sampling point is calculated, and the maximum value of the average slope is taken as the maximum average slope value of the sampling point. The slope change sequence is obtained based on the maximum average slope value of all sampling points in the pulse wave signal, and the peak position in the slope change sequence is determined as the systolic position of the pulse wave.

5. The method according to claim 1, characterized in that, The heart rate characteristics include heart rate variability characteristics and deep breathing characteristics; The method of obtaining heart rate characteristics based on the systolic interval between every two pulse waves includes: Based on the systolic interval, the heart rate variability features are extracted, and the heart rate variability features include at least one of time-domain features, frequency-domain features, Poincaré scatter plot features, and sample entropy features; The deep breathing characteristics are obtained by calculating the absolute median difference and the absolute mean difference of the differences between adjacent systolic heart beats.

6. The method according to claim 5, characterized in that, The deep breathing characteristics are obtained by calculating the absolute median difference and the absolute mean difference of the differences between adjacent systolic heart beats, including: The difference between each pair of adjacent intervals is calculated to obtain a difference sequence composed of the differences between multiple pairs of adjacent intervals; Obtain the median of the difference sequence, and take the median of the absolute values ​​of the differences between each difference and the median as the absolute median difference; The average value of the difference sequence is calculated, and the absolute value of the difference between each difference and the average value is calculated. The average of the absolute values ​​of the differences between each difference and the average value is taken as the absolute mean difference. The difference between the absolute median and the absolute mean is used as the deep breathing feature.

7. The method according to claim 5, characterized in that, Determining the heart rhythm detection result based on the heart rate characteristics includes: The machine learning model, obtained through pre-training, classifies and identifies the heart rate variability features and the deep breathing features to obtain the heart rhythm detection results. The heart rhythm detection results include normal heart rhythm or arrhythmia. The machine learning model has the ability to classify and identify based on the heart rate variability features and the deep breathing features.

8. A heart rhythm detection device, characterized in that, The device includes: The acquisition module is used to acquire the user's pulse wave signal and the acceleration signal of the user's wrist. The first evaluation module is used to obtain a signal quality evaluation result based on the amplitude fluctuation information of the pulse wave signal; The second evaluation module is used to obtain the motion state evaluation result based on the rate of change of the acceleration signal; The determination module is used to determine the systolic position of each pulse wave in the pulse wave signal if the signal quality assessment result indicates that it is qualified and the motion state assessment result indicates that it is stationary. The systolic position is the time point when the systolic phase rises the fastest in a pulse wave. The first calculation module is used to determine the systolic interval between each two pulse waves based on the time interval between each two adjacent systolic positions. The second calculation module is used to obtain heart rate characteristics based on the systolic interval between every two pulse waves. The detection module is used to determine the heart rhythm detection result based on the heart rate characteristics.

9. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the heart rhythm detection method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the heart rhythm detection method according to any one of claims 1 to 7.

Images