Electrocardiographic monitoring method based on cardiac rehabilitation data

Abstract

The application discloses a heart recovery data-based electrocardio monitoring method, and relates to the technical field of electrocardio monitoring, solves the technical problem of high false positive rate of abnormal heart beat confirmation caused by lack of scene self-adaptive grading system, and introduces a signal quality index based on contact impedance and baseline drift variance, so that the system can identify invalid leads in real time and prompt the user to adjust, filters low-quality data from the source, avoids false analysis caused by poor contact, and improves the accuracy of waveform reconstruction by dynamically selecting the wavelet basis with the smallest error according to the recently confirmed QRS complex form, which is different from traditional fixed wavelet transform, and effectively identifies and eliminates artifacts caused by violent movement by combining cross-correlation analysis of acceleration signals and electrocardio signals, thereby reducing the false positive rate in a movement scene.

Description

Technical Field

[0001] This invention relates to the field of electrocardiogram (ECG) monitoring technology, specifically to an ECG monitoring method based on cardiac rehabilitation data. Background Technology

[0002] Cardiac rehabilitation is an important part of the postoperative care and management of chronic heart failure in patients with cardiovascular diseases. Wearable electrocardiogram monitoring devices can enable continuous monitoring outside the hospital and timely detection of arrhythmias and myocardial ischemia events.

[0003] According to patent publication number CN119454046A, a method and system for electrocardiogram (ECG) monitoring based on cardiac rehabilitation data are disclosed. The method includes the following steps: ECG signals under different exercise states are collected using a multi-lead ECG machine and a motion monitoring device, and fatigue-level ECG mapping is performed to obtain fatigue signals. Next, the heart rate changes of the fatigue-level ECG mapping signal are analyzed, the disordered scatter trend of heart rate is quantified, and a dynamic inflection point distribution is fitted to obtain trend dynamic data. Based on this data, multi-scale regularity intersection analysis is performed to extract the fatigue heart rate change pattern, and further quantitative assessment of rehabilitation deviation is conducted. Finally, the rehabilitation deviation assessment results are sent to a terminal for real-time ECG monitoring, supporting cardiac rehabilitation management. This invention improves ECG monitoring technology through optimization.

[0004] However, when monitoring, existing technologies rely solely on the relative change rate of the RR interval difference as a candidate abnormality marker, without combining it with compensatory pause characteristics for secondary confirmation. As a result, a large number of physiological variations or artifacts are mistakenly identified as abnormalities. At the same time, existing methods often use a fixed percentage of maximum heart rate as the upper limit of exercise heart rate, without dynamically adjusting it according to real-time exercise intensity and individual heart rate reserve, thus failing to achieve refined exercise guidance. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an electrocardiogram monitoring method based on cardiac rehabilitation data, which solves the problem of high false positives in the confirmation of abnormal heartbeats due to the lack of a scenario-adaptive grading system.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an electrocardiogram monitoring method based on cardiac rehabilitation data, which specifically includes the following steps: Step 1: Acquire the electrical activity signal of the heart and use the skin impedance detection module to monitor the contact impedance value of each lead in real time. Construct a signal quality index and compare the calculated lead signal quality index with a preset threshold. If it is less than the preset threshold, mark the lead as an invalid channel. Based on the above processing, obtain valid electrical activity signals. Step 2: Preprocess the effective electrical activity signal and use adaptive wavelet transform for multi-scale decomposition to accurately identify the QRS complex. At the same time, combine motion scene information to identify and eliminate candidate abnormal heartbeats to obtain confirmed abnormal heartbeats. Step 3: Based on the confirmed abnormal heartbeats, combined with the user's individualized abnormal heartbeat baseline, determine the immediate level according to the type of abnormal heartbeat, the occurrence scenario, and the preset individualized threshold. Step 4: In the resting scenario, mark the nighttime sleep period according to the user's historical sleep record, and maintain, downgrade or upgrade the level based on the heart rate drop and premature beat type. In the exercise scenario, obtain the acceleration variance and divide the exercise intensity into three levels: low, medium and high according to the preset threshold. Determine the upper limit of exercise heart rate based on different exercise intensities, obtain the heart rate when abnormal heartbeats occur, compare it with the upper limit of exercise heart rate, and dynamically adjust the level. Step 5: Based on the dynamic correction level, perform local recording, tactile alerts, or audible alerts corresponding to the level.

[0007] As a further aspect of the present invention, the specific method for constructing the signal quality index is as follows: ,in This indicates the preset high-quality contact resistance threshold. This represents the baseline drift variance estimated after high-pass filtering. Represents the variance of the original signal. and These are the weighting coefficients.

[0008] As a further aspect of the present invention, the method of using adaptive wavelet transform for multi-scale decomposition to accurately identify QRS groups is as follows: A wavelet basis library is established. Based on the 10 most recently confirmed QRS group morphologies, the wavelet basis with the smallest waveform reconstruction error is selected from the wavelet basis library as the wavelet basis for the current processing window, and then the formula is applied. ,in As a candidate small fundamental frequency, Let i be the candidate small fundamental wavelet set, and i be the sample index. This is the original signal segment. To reconstruct the signal segment, and simultaneously based on the obtained minimum fundamental frequency... The purified ECG signal was decomposed into four levels of multi-scale decomposition, and the decomposition scale containing the main frequency of the QRS complex was selected as the characteristic scale. The modulus maxima of wavelet coefficients are calculated at the characteristic scale, and a dynamic threshold is used. ,in and These are the mean and standard deviation of the wavelet coefficients within the sliding window, respectively. As an adaptive coefficient, after detecting the position of the R-wave peak, the QRS starts from the point where the modulus maxima appear on the feature scale and ends at the point where the modulus maxima decays to below the dynamic threshold.

[0009] As a further aspect of the present invention, the identification and elimination of candidate abnormal heartbeats by combining motion scene information specifically includes: Simultaneously, during the first 24 hours after the user first wears the device, individualized RR interval baselines are collected and established, the mean and standard deviation of the resting RR interval are calculated, the RR interval fluctuation range under exercise is calculated simultaneously, and then the difference between adjacent RR intervals is calculated. and relative rate of change ,in The average RR interval within the current sliding window is then used for anomaly detection processing in conjunction with the motion scene: If the scene is at rest, When the percentage is >25%, it is marked as a candidate abnormal heartbeat. If the scenario is motion, a dynamic threshold is used, and only when... It is only marked as an exception when >Th.

[0010] As a further aspect of the present invention, for the candidate abnormal heartbeats obtained by marking, the acceleration signals are extracted one second before and after time t of the candidate heartbeat, and the peak value of their amplitude change is calculated. and variance ,like >0.8 or If the value is greater than 2.5g, calculate the instantaneous energy of the effective electrical activity signal in the 30-100Hz frequency band and cross-correlate it with the acceleration amplitude sequence. The correlation coefficient with a time delay within ±50ms is calculated. If the correlation coefficient If the value is >0.7, it is determined to be a motion artifact. At the same time, the corresponding candidate abnormal heartbeats are removed, and the remaining candidate abnormal heartbeats are recorded as heartbeats to be analyzed. Next, obtain all the heartbeats to be analyzed and their local mean RR intervals. and the pre-ventricular interval and the interval after premature beats According to the formula The percentage of deviation was calculated. At the same time, the calculated percentage of deviation Compared with the preset ratio, if the deviation percentage If the ratio is greater than the preset ratio, the corresponding heartbeat to be analyzed will be removed, and the confirmed abnormal heartbeats will be obtained based on the above operations.

[0011] As a further aspect of the present invention, the specific method for determining the instantaneous level is as follows: Isolated ventricular premature beats (PVCs) or atrial premature beats (PACs) with a frequency of <10 times / hour for this type of premature beat in the past hour; a single pause with a duration of <2.0 seconds; a single aberrant conduction during exercise; any abnormal heartbeat occurring during a low-confidence period with a signal quality of 0.5 ≤ SQI < 0.6. If any of the above conditions are met, the corresponding classification is Level 1. If any of the following conditions are met, the patient is classified as Level 2: PVC or PAC frequency ≥10 beats / hour and <30 beats / hour in the past hour; 2 consecutive abnormal heartbeats, with a pair count ≥2 times in the past hour; 3-9 consecutive abnormal heartbeats, duration <10 seconds, and smooth heart rate changes during exercise; pause duration 2.0 seconds ≤ pause < 3.0 seconds. If any of the following conditions are met, the patient is classified as Grade III: ≥10 consecutive wide QRS complexes, heart rate ≥120 bpm, duration ≥10 seconds; completely irregular, amorphous ECG waveform; pause duration ≥3.0 seconds; short runs of VT ≥2 times / 24 hours; ST segment deviation ≥0.2mV and lasting ≥1 minute.

[0012] As a further aspect of the present invention, the method for maintaining, downgrading, or upgrading the level in a resting scenario is as follows: Based on the user's historical sleep records, the period is marked as nighttime sleep. If an abnormal heartbeat occurs during nighttime sleep and is an isolated Level 1 event, it is maintained as Level 1. If a Level 2 event occurs at night, it needs to be judged in conjunction with the heart rate trend: if the heart rate decreases by more than 20% compared to the daytime average and the premature beats are atrial, it may be benign and is downgraded to Level 1; if they are ventricular in pairs, it is maintained as Level 2. If a Level 3 event occurs at night, it is upgraded to a critical state.

[0013] As a further aspect of the present invention, the method for determining the safe upper limit of exercise heart rate based on different exercise intensities in a sports scenario is as follows: Based on acceleration variance Classify exercise intensity. <0.2 Classified as low-intensity exercise, 0.2 < <0.8 Classified as moderate-intensity exercise, ≥0.8 Classified as high-intensity exercise, obtain the user's corresponding resting heart rate. and heart rate reserve And calculate the safe upper limit of heart rate for different exercise intensities using the formula; Low-intensity exercise: ,in This is the upper limit of safety coefficient for low-intensity exercise; Moderate intensity exercise: ,in This represents the upper limit of safety coefficient for moderate-intensity exercise. High-intensity exercise: ,in This is the upper limit coefficient for safety during high-intensity exercise.

[0014] As a further aspect of the present invention, the specific method for dynamically adjusting the level is as follows: Obtain the heart rate corresponding to the occurrence of abnormal heartbeats. And compare it with the corresponding safe upper limit of exercise heart rate under different exercise intensities. If a comparison is made, <0.7× If the risk of abnormal heartbeats is relatively high, and the level remains unchanged, then... ≥0.7× If so, the level will be reduced by one level. > If it lasts for 10 seconds, it will be upgraded to level three.

[0015] As a further aspect of the present invention, the specific method for performing local recording, tactile alert, or audible alert processing corresponding to the dynamic correction level is as follows: For Level 1: Event logs are created in the device’s local storage without generating any vibration, sound, or screen notifications; For Level 2: The device generates a single, gentle vibration lasting 0.5 seconds, broadcasts an abnormal situation prompt, and displays a yellow warning icon for 3 seconds; For Level 3: The device generates strong vibrations with a frequency ranging from low to high for more than 3 seconds, emits a high-decibel intermittent buzzing sound, and continuously broadcasts warning information through audio equipment.

[0016] This invention provides an electrocardiogram (ECG) monitoring method based on cardiac rehabilitation data. Compared with existing technologies, it has the following advantages: This invention introduces a signal quality index based on contact impedance and baseline drift variance, enabling the system to identify invalid leads in real time and prompt users to adjust them. This filters low-quality data at the source and avoids misanalysis caused by poor contact. Unlike traditional fixed wavelet transform, this invention dynamically selects the wavelet basis with the smallest error based on the recently confirmed QRS complex morphology, improving the accuracy of waveform reconstruction. By combining the cross-correlation analysis of acceleration signals and ECG signals, it can effectively identify and eliminate artifacts caused by strenuous exercise, reducing the false positive rate in motion scenarios.

[0017] This invention establishes a personalized RR interval baseline based on the data from the first 24 hours of use, making the abnormal detection threshold more closely match the user's own physiological characteristics. It innovatively separates the processing of resting and exercise scenarios, and introduces the concept of a safe upper limit of heart rate in exercise scenarios to determine whether abnormal heartbeats occur within the physiological load limit, thereby deciding whether to downgrade or upgrade the risk level. Through differentiated processing of Level 1, Level 2, and Level 3, it ensures timely alarms for critical situations while avoiding excessive interference with benign or physiological variations, greatly improving the user experience and medical value. Attached Figure Description

[0018] Figure 1 This is a flowchart of the electrocardiogram monitoring method based on cardiac rehabilitation data according to the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please see Figure 1 This application provides a method for electrocardiogram monitoring based on cardiac rehabilitation data, which specifically includes the following steps: Step 1: Acquire electrical activity signals of the heart using a wearable device. Based on the acquired electrical activity signals, and using a skin impedance detection module to monitor the contact impedance values ​​of each lead in real time, construct a signal quality index. ,in This indicates the preset high-quality contact resistance threshold. This represents the baseline drift variance estimated after high-pass filtering. Represents the variance of the original signal. and The weighting coefficients are used, and the calculated lead signal quality index is also included. The lead signal quality index is compared with a preset value, which is typically 0.3. If the value is less than the preset value, the system automatically marks the lead as an invalid channel, resets its weight to zero in the subsequent waveform reconstruction, and triggers a voice prompt to guide the user to adjust it. Based on the above processing, valid electrical activity signals are obtained through filtering. Step 2: Preprocess the effective electrical activity signal, specifically including using a 50Hz notch filter to remove power frequency interference, then using a 0.5-150Hz bandpass filter to filter out electromyographic noise and low-frequency drift, and finally using wavelet transform to perform multi-scale decomposition of the signal, with the specific processing method as follows: A wavelet basis library is established. Based on the 10 most recently confirmed QRS group morphologies, the waveform reconstruction error under each wavelet basis is calculated. The wavelet basis with the smallest error is selected for the current processing window, and then the waveform reconstruction error is calculated according to the formula. The minimum fundamental wave was calculated. ,in As a candidate small fundamental frequency, Let i be the candidate small fundamental wavelet set, and i be the sample index. This is the original signal segment. To reconstruct the signal segment, and simultaneously based on the obtained minimum fundamental frequency... The purified ECG signal was decomposed into four levels of multi-scale decomposition, and the decomposition scale containing the main frequency of the QRS complex was selected as the characteristic scale.

[0021] Next, the modulus maxima of the wavelet coefficients are calculated on the feature scale, and a dynamic threshold is used instead of a fixed threshold. The formula for calculating the dynamic threshold is as follows: ,in and These are the mean and standard deviation of the wavelet coefficients within the sliding window, respectively. As an adaptive coefficient, after detecting the R-wave peak position, the reverse search is performed to the starting point of the modulus maxima on the feature scale as the QRS start point, and the forward search is performed to the point where the modulus maxima decays to below the threshold as the QRS end point. Simultaneously, during the first 24 hours after the user first wears the device, individualized RR interval baselines are collected and established, the mean and standard deviation of the resting RR interval are calculated, the RR interval fluctuation range under exercise is calculated simultaneously, and then the difference between adjacent RR intervals is calculated. and relative rate of change ,in The average RR interval within the current sliding window is then used for anomaly detection processing in conjunction with the motion scene: If the scene is at rest, When the percentage is >25%, it is marked as a candidate abnormal heartbeat. If the scenario is motion, a dynamic threshold is used, and only when... It is only marked as an exception when >Th; For the candidate abnormal heartbeats identified by the labeling, the acceleration signals are extracted one second before and after time t of the candidate heartbeat, and the peak amplitude changes are calculated. and variance ,like >0.8 or If the value is greater than 2.5g, calculate the instantaneous energy of the effective electrical activity signal in the 30-100Hz frequency band and cross-correlate it with the acceleration amplitude sequence. The correlation coefficient with a time delay within ±50ms is calculated. If the correlation coefficient If the value is >0.7, it is determined to be a motion artifact. At the same time, the corresponding candidate abnormal heartbeats are removed, and the remaining candidate abnormal heartbeats are recorded as heartbeats to be analyzed. Next, obtain all the heartbeats to be analyzed and their local mean RR intervals. and the pre-ventricular interval and the interval after premature beats According to the formula The percentage of deviation was calculated. At the same time, the calculated percentage of deviation Compared with the preset ratio, if the deviation percentage If the ratio is greater than the preset ratio, the corresponding heartbeat to be analyzed will be removed, and the confirmed abnormal heartbeats will be obtained based on the above operations.

[0022] Step 3: Based on the confirmed abnormal heartbeats, including timestamps, abnormality types, associated RR interval sequences, QRS waveform segments and morphological characteristics, and signal quality indices, an individualized abnormal heartbeat baseline is constructed according to the user's initial resting period. This baseline includes baselines for premature beat frequency, short-run VT episodes, abnormal load rate, and the proportion of motion-related abnormalities. An immediate level is assigned based on the type, occurrence scenario, and individualized threshold. The immediate level is determined as follows: Level 1: Isolated ventricular premature beats (PVCs) or atrial premature beats (PACs), with a frequency of <10 times / hour for this type of premature beat in the past hour; a single pause with a duration <2.0 seconds; a single aberrant conduction during exercise; any abnormal heartbeat occurring during a low-confidence period with a signal quality of 0.5 ≤ SQI < 0.6. If any of the above conditions are met, the corresponding level is determined to be Level 1. Level 2: PVC or PAC frequency ≥10 beats / hour and <30 beats / hour in the past hour; 2 consecutive abnormal heartbeats, with ≥2 pairs in the past hour; 3-9 consecutive abnormal heartbeats, duration <10 seconds, and smooth heart rate changes during exercise; pause duration 2.0 seconds ≤ pause < 3.0 seconds. If any of the above conditions are met, the corresponding level is Level 2. Level 3: ≥10 consecutive wide QRS complexes, heart rate ≥120 beats / min, duration ≥10 seconds; completely irregular, amorphous ECG waveform; pause duration ≥3.0 seconds; short runs of VT ≥2 times / 24 hours; ST segment deviation ≥0.2mV and lasting ≥1 minute. If any of the above conditions are met, the corresponding determination is Level 3.

[0023] Step 4: Next, dynamically optimize and adjust the real-time ratings obtained for both the resting and moving scenes. The adjustment method for the real-time ratings in the resting scene is as follows: Based on the user's historical sleep records, the period is marked as nighttime sleep. If an abnormal heartbeat occurs during nighttime sleep and is an isolated Level 1 event, it is maintained at Level 1. If a Level 2 event occurs at night, it needs to be judged in conjunction with the heart rate trend: if the heart rate decreases by more than 20% compared to the daytime average and the premature beats are atrial, it may be benign and is downgraded to Level 1; if they are ventricular and paired, it is maintained at Level 2. If a Level 3 event occurs at night, it is upgraded to a critical state. The method for adjusting the instantaneous level in motion scenarios is as follows: Obtain the acceleration variance. And based on the acceleration variance Exercise intensity is divided into three levels: low, medium, and high. Specifically... <0.2 If it is 0.2, then it is classified as low-intensity exercise. < <0.8 If it is classified as moderate intensity exercise, then it is considered moderate intensity exercise. ≥0.8 If the exercise intensity is high, it is classified as high-intensity exercise. Based on different exercise intensities, a safe upper limit for exercise heart rate is determined, and the user's corresponding resting heart rate is obtained. and heart rate reserve ,and ,in It represents the individualized maximum heart rate and obtains the safety upper limit coefficient corresponding to different exercise intensities. It calculates the safety upper limit of exercise heart rate for different exercise intensities according to different calculation formulas. Low-intensity exercise: ,in This is the upper limit of safety coefficient for low-intensity exercise; Moderate intensity exercise: ,in This represents the upper limit of safety coefficient for moderate-intensity exercise. High-intensity exercise: ,in This represents the upper limit of safety coefficient for high-intensity exercise. Next, obtain the heart rate corresponding to the occurrence of the abnormal heartbeat. And compare it with the corresponding safe upper limit of exercise heart rate under different exercise intensities. If a comparison is made, <0.7× If the risk of abnormal heartbeats is relatively high, and the level remains unchanged, then... ≥0.7× Then physiological stress is the primary cause, and the stress level decreases by one level. > If it lasts for 10 seconds, it will be upgraded to level three.

[0024] Step 5: Based on the dynamic correction levels obtained in Step 4, perform different monitoring and processing. For Level 1 anomalies, create an event record in the device's local storage, including: timestamp, anomaly type, original level and corrected level, correction reason, and confidence level. Level 1 events do not generate any vibration, sound, or screen prompts, and the device status indicator light remains solid green as in normal operation. For Level 2 abnormal situations, the device's built-in linear motor generates a gentle vibration for 0.5 seconds, vibrating only once, broadcasting an abnormal situation prompt and displaying a yellow warning icon, which disappears automatically after 3 seconds. In response to Level 3 abnormal situations, the device will vibrate continuously for more than 3 seconds, with the frequency increasing from low to high, generating a strong tactile warning. The device's built-in buzzer will emit a high-decibel intermittent buzzing sound, which will be continuously broadcast through Bluetooth headphones or the device's speaker.

[0025] Some of the data in the above formulas are numerical calculations with dimensions removed, and the contents not described in detail in this specification are all prior art known to those skilled in the art.

[0026] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. An electrocardiogram monitoring method based on cardiac rehabilitation data, characterized in that, The method specifically includes the following steps: Step 1: Acquire the electrical activity signal of the heart and use the skin impedance detection module to monitor the contact impedance value of each lead in real time. Construct a signal quality index and compare the calculated lead signal quality index with a preset threshold. If it is less than the preset threshold, mark the lead as an invalid channel. Based on the above processing, obtain valid electrical activity signals. Step 2: Preprocess the effective electrical activity signal and use adaptive wavelet transform for multi-scale decomposition to accurately identify the QRS complex. At the same time, combine motion scene information to identify and eliminate candidate abnormal heartbeats to obtain confirmed abnormal heartbeats. Step 3: Based on the confirmed abnormal heartbeats, combined with the user's individualized abnormal heartbeat baseline, determine the immediate level according to the type of abnormal heartbeat, the occurrence scenario, and the preset individualized threshold. Step 4: In the resting scenario, mark the nighttime sleep period according to the user's historical sleep record, and maintain, downgrade or upgrade the level based on the heart rate drop and premature beat type. In the exercise scenario, obtain the acceleration variance and divide the exercise intensity into three levels: low, medium and high according to the preset threshold. Determine the upper limit of exercise heart rate based on different exercise intensities, obtain the heart rate when abnormal heartbeats occur, compare it with the upper limit of exercise heart rate, and dynamically adjust the level. Step 5: Based on the dynamic correction level, perform local recording, tactile alerts, or audible alerts corresponding to the level.

2. The electrocardiogram monitoring method based on cardiac rehabilitation data according to claim 1, characterized in that, The specific method for constructing the signal quality index is as follows: ,in This indicates the preset high-quality contact resistance threshold. This represents the baseline drift variance estimated after high-pass filtering. Represents the variance of the original signal. and These are the weighting coefficients.

3. The electrocardiogram monitoring method based on cardiac rehabilitation data according to claim 1, characterized in that, The method of using adaptive wavelet transform for multi-scale decomposition to accurately identify QRS groups is as follows: A wavelet basis library is established. Based on the 10 most recently confirmed QRS group morphologies, the wavelet basis with the smallest waveform reconstruction error is selected from the wavelet basis library as the wavelet basis for the current processing window, and then the formula is applied. ,in As a candidate small fundamental frequency, Let i be the candidate small fundamental wavelet set, and i be the sample index. This is the original signal segment. To reconstruct the signal segment, and simultaneously based on the obtained minimum fundamental frequency... The purified ECG signal was decomposed into four levels of multi-scale decomposition, and the decomposition scale containing the main frequency of the QRS complex was selected as the characteristic scale. The modulus maxima of wavelet coefficients are calculated at the characteristic scale, and a dynamic threshold is used. ,in and These are the mean and standard deviation of the wavelet coefficients within the sliding window, respectively. As an adaptive coefficient, after detecting the position of the R-wave peak, the QRS starts from the point where the modulus maxima appear on the feature scale and ends at the point where the modulus maxima decays to below the dynamic threshold.

4. The electrocardiogram monitoring method based on cardiac rehabilitation data according to claim 1, characterized in that, The process of identifying and eliminating candidate abnormal heartbeats by combining motion scene information specifically includes: Simultaneously, during the first 24 hours after the user first wears the device, individualized RR interval baselines are collected and established, the mean and standard deviation of the resting RR interval are calculated, the RR interval fluctuation range under exercise is calculated simultaneously, and then the difference between adjacent RR intervals is calculated. and relative rate of change ,in The average RR interval within the current sliding window is then used for anomaly detection processing in conjunction with the motion scene: If the scene is at rest, When the percentage is >25%, it is marked as a candidate abnormal heartbeat. If the scenario is motion, a dynamic threshold is used, and only when... It is only marked as an exception when >Th.

5. The electrocardiogram monitoring method based on cardiac rehabilitation data according to claim 4, characterized in that, For the candidate abnormal heartbeats identified by the labeling, the acceleration signals are extracted one second before and after time t of the candidate heartbeat, and the peak amplitude changes are calculated. and variance ,like >0.8 or If the value is greater than 2.5g, calculate the instantaneous energy of the effective electrical activity signal in the 30-100Hz frequency band and cross-correlate it with the acceleration amplitude sequence. The correlation coefficient with a time delay within ±50ms is calculated. If the correlation coefficient If the value is >0.7, it is determined to be a motion artifact. At the same time, the corresponding candidate abnormal heartbeats are removed, and the remaining candidate abnormal heartbeats are recorded as heartbeats to be analyzed. Next, obtain all the heartbeats to be analyzed and their local mean RR intervals. and the pre-ventricular interval and the interval after premature beats According to the formula The percentage of deviation was calculated. At the same time, the calculated percentage of deviation Compared with the preset ratio, if the deviation percentage If the ratio is greater than the preset ratio, the corresponding heartbeat to be analyzed will be removed, and the confirmed abnormal heartbeats will be obtained based on the above operations.

6. The electrocardiogram monitoring method based on cardiac rehabilitation data according to claim 1, characterized in that, The specific method for determining the immediate level is as follows: Isolated ventricular premature beats (PVCs) or atrial premature beats (PACs) with a frequency of <10 times / hour for this type of premature beat in the past hour; a single pause with a duration of <2.0 seconds; a single aberrant conduction during exercise; any abnormal heartbeat occurring during a low-confidence period with a signal quality of 0.5 ≤ SQI < 0.

6. If any of the above conditions are met, the corresponding classification is Level 1. If any of the following conditions are met, the patient is classified as Level 2: PVC or PAC frequency ≥10 beats / hour and <30 beats / hour in the past hour; 2 consecutive abnormal heartbeats, with a pair count ≥2 times in the past hour; 3-9 consecutive abnormal heartbeats, duration <10 seconds, and smooth heart rate changes during exercise; pause duration 2.0 seconds ≤ pause < 3.0 seconds. If any of the following conditions are met, the patient is classified as Grade III: ≥10 consecutive wide QRS complexes, heart rate ≥120 bpm, duration ≥10 seconds; completely irregular, amorphous ECG waveform; pause duration ≥3.0 seconds; short runs of VT ≥2 times / 24 hours; ST segment deviation ≥0.2mV and lasting ≥1 minute.

7. The electrocardiogram monitoring method based on cardiac rehabilitation data according to claim 1, characterized in that, In a quiet environment, the methods for maintaining, downgrading, or upgrading a level are as follows: Based on the user's historical sleep records, the period is marked as nighttime sleep. If an abnormal heartbeat occurs during nighttime sleep and is an isolated Level 1 event, it is maintained as Level 1. If a Level 2 event occurs at night, it needs to be judged in conjunction with the heart rate trend: if the heart rate decreases by more than 20% compared to the daytime average and the premature beats are atrial, it may be benign and is downgraded to Level 1; if they are ventricular in pairs, it is maintained as Level 2. If a Level 3 event occurs at night, it is upgraded to a critical state.

8. The electrocardiogram monitoring method based on cardiac rehabilitation data according to claim 1, characterized in that, The method for determining the safe upper limit of exercise heart rate based on different exercise intensities in sports scenarios is as follows: Based on acceleration variance Classify exercise intensity. <0.2 Classified as low-intensity exercise, 0.2 < <0.8 Classified as moderate-intensity exercise, ≥0.8 Classified as high-intensity exercise, obtain the user's corresponding resting heart rate. and heart rate reserve And calculate the safe upper limit of heart rate for different exercise intensities using the formula; Low-intensity exercise: ,in This is the upper limit of safety coefficient for low-intensity exercise; Moderate intensity exercise: ,in This represents the upper limit of safety coefficient for moderate-intensity exercise. High-intensity exercise: ,in This is the upper limit coefficient for safety during high-intensity exercise.

9. The electrocardiogram monitoring method based on cardiac rehabilitation data according to claim 1, characterized in that, The specific method for dynamically adjusting the level is as follows: Obtain the heart rate corresponding to the occurrence of abnormal heartbeats. And compare it with the corresponding safe upper limit of exercise heart rate under different exercise intensities. If a comparison is made, <0.7× If the risk of abnormal heartbeats is relatively high, and the level remains unchanged, then... ≥0.7× If so, the level will be reduced by one level. > If it lasts for 10 seconds, it will be upgraded to level three.

10. The electrocardiogram monitoring method based on cardiac rehabilitation data according to claim 1, characterized in that, The specific method for performing local recording, tactile alerts, or audible alerts corresponding to the dynamic correction level is as follows: For Level 1: Event logs are created in the device’s local storage without generating any vibration, sound, or screen notifications; For Level 2: The device generates a single, gentle vibration lasting 0.5 seconds, broadcasts an abnormal situation prompt, and displays a yellow warning icon for 3 seconds; For Level 3: The device generates strong vibrations with a frequency ranging from low to high for more than 3 seconds, emits a high-decibel intermittent buzzing sound, and continuously broadcasts warning information through audio equipment.

Images