A wearable physiological monitoring system and its anti-fatigue data analysis method

Abstract

The application discloses a wearable physiological monitoring system and an anti-fatigue data analysis method thereof, and relates to the technical field of data analysis.The physiological data acquisition system is arranged in the wearable device of a user, and the ECG signal of the user is acquired as a first ECG signal to determine the motion state of the user; different sampling modes and motion artifact removal modes are set for the ECG signal of the user according to the motion state of the user to obtain a second ECG signal of the user; and it is determined whether the artifact of the ECG signal of the user is removed qualifiedly, if qualified, anti-fatigue data analysis is performed; if not qualified, the motion artifact removal of the first ECG signal is performed on the second ECG signal again until the second ECG signal is qualified. In this way, the physiological data of the user collected can be subjected to motion artifact removal, and the adverse effect on the anti-fatigue data analysis process of the user is reduced; the electrocardiogram signal of the user is ensured not to be distorted, the fatigue state evaluation is accurate, and the false health warning is avoided.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, specifically to a wearable physiological monitoring system and its anti-fatigue data analysis method. Background Technology

[0002] With the increasing demand for health management, wearable physiological monitoring systems (such as smartwatches and health bracelets) have become important health monitoring tools in people's daily lives. These devices integrate multiple sensors, such as electrocardiograms (ECGs), heart rate monitors, accelerometers, and gyroscopes, to collect users' physiological data and activity status in real time. In particular, ECGs and heart rate monitoring provide crucial data on a user's heart health and fatigue status. Furthermore, wearable devices can monitor users' exercise intensity, activity patterns, and sleep quality, providing comprehensive data support for health management.

[0003] Currently, fatigue data analysis methods based on wearable devices are gradually attracting the attention of researchers and developers. Utilizing physiological data collected from wearable devices, such as heart rate variability (HRV) and electrocardiogram signals, combined with the user's exercise status and environmental conditions, researchers have proposed several fatigue analysis models. These models, by analyzing indicators such as heart rate changes, exercise intensity, and recovery status, can determine the user's level of fatigue and thus provide personalized health recommendations. Many smart devices are now able to adjust monitoring strategies in real time based on this data and provide early warnings based on the user's fatigue level.

[0004] While existing fatigue data analysis methods based on wearable physiological monitoring systems have achieved some success, significant challenges remain in practical applications, particularly in eliminating motion artifacts and ensuring signal accuracy. Signal interference caused by movement (such as running and jumping) (e.g., ECG baseline drift, electromyography (EMG) artifacts) can adversely affect data acquisition and analysis; these motion artifacts can distort ECG signals, leading to inaccurate fatigue assessments and potentially even false health warnings. Summary of the Invention

[0005] The purpose of this invention is to solve the problems mentioned above and to provide a wearable physiological monitoring system and its anti-fatigue data analysis method.

[0006] In a first aspect of this invention, a wearable physiological monitoring system is first proposed, the system comprising:

[0007] Data acquisition module: The physiological data acquisition system is set up on the user's wearable device, and the user's ECG signal is collected by the physiological data acquisition system as the first ECG signal to determine the user's movement status;

[0008] Artifact Removal Module: Based on the user's motion state, different sampling methods and motion artifact removal methods are set for the user's ECG signal to obtain the user's second ECG signal;

[0009] First analysis module: Determine whether the artifact removal of the user's ECG signal is qualified based on the first ECG signal and the second ECG signal. If qualified, the second ECG signal is used as the final user's ECG signal for fatigue data analysis.

[0010] Second analysis module: If it fails, the motion artifact removal of the first ECG signal is performed again on the second ECG signal until the second ECG signal passes.

[0011] Optionally, the physiological data acquisition system in the user's wearable device includes:

[0012] The physiological data acquisition system is used in user wearable devices, and the physiological data acquisition system includes an accelerometer, an ECG sensor, and a PPG sensor;

[0013] The ECG sensor is used to collect the user's ECG signal; the accelerometer is used to determine the user's current motion state, which includes being at rest, walking, and strenuous exercise; the PPG sensor is used to detect the user's blood oxygen saturation and heart rate, and to perform anti-fatigue data analysis on the user in conjunction with the ECG signal.

[0014] Optionally, the accelerometer is used to determine the user's current motion state by:

[0015] The system collects the user's current acceleration amplitude and acceleration change frequency. If the acceleration amplitude is low, close to the gravitational acceleration value of 9.8 m / s², and the acceleration change frequency does not change significantly, then the user's current motion state is a stationary state.

[0016] If the acceleration amplitude changes to some extent, but the frequency of change is low and the periodicity is strong, then the user's current motion state is walking.

[0017] If the acceleration amplitude is large and the frequency of change is also large, then the user's current motion state is a state of intense motion.

[0018] Optionally, different sampling methods and motion artifact removal methods can be set for the user's ECG signal based on the user's motion state, including:

[0019] When the user's motion state is stationary, the sampling frequency of the ECG signal is set to no more than 250Hz, and the first ECG signal is filtered by a low-pass filter with a setting of 0.5-1Hz to remove low-frequency noise and motion artifacts; thus, the user's second ECG signal is obtained.

[0020] When the user's movement state is walking, the sampling frequency of the ECG signal is set to 500Hz, and the first ECG signal is processed by setting a high-pass filter, setting 1-2Hz to remove low-frequency noise, and then performing wavelet transformation on the removed low-frequency noise to remove high-frequency noise and motion artifacts; thus obtaining the user's second ECG signal.

[0021] When the user's motion state is in a state of intense motion, the sampling frequency of the ECG signal is set to 1000Hz, and the first ECG signal is separated into a mixed signal composed of motion artifacts and ECG signals by using blind source separation technology to extract the ECG signal. The extracted ECG signal is then subjected to wavelet transform to remove high-frequency noise and motion artifacts, thus obtaining the user's second ECG signal.

[0022] Optionally, the step of determining whether the artifact removal of the user's ECG signal is qualified based on the first ECG signal and the second ECG signal is as follows:

[0023] The second ECG signal is low-pass filtered to remove high-frequency components while retaining low-frequency baseline information;

[0024] The baseline of the ECG signal is fitted using the least squares method to obtain the fitted baseline. The baseline drift coefficient is then calculated based on the baseline of the ECG signal and the second ECG signal. The calculation formula is as follows:

[0025]

[0026] In the formula, GH is the baseline drift coefficient, ECG(t) is the value of the second ECG signal at time point t, the fitted baseline(t) is the fitted baseline at time point t, and T is the signal duration;

[0027] The artifact removal pass coefficient is calculated based on the baseline drift coefficient and the waveform of the second ECG signal, and the artifact removal pass coefficient is used to determine whether the artifacts of the user's ECG signal have been removed successfully.

[0028] Optionally, the step of calculating the artifact removal qualification coefficient based on the baseline drift coefficient and the waveform of the second ECG signal is as follows:

[0029] The Pan-Tompkins algorithm was used to extract the P wave, QRS complex, and T wave from the second ECG signal.

[0030] The extracted P-wave, QRS complex, and T-wave are compared with their corresponding normal waveforms. The similarity of amplitude, shape, and duration of each waveform is calculated, and the average similarity is used as the similarity of the corresponding waveform. The similarity of all waveforms is summed and divided by 3 to obtain the waveform integrity coefficient of the second ECG signal.

[0031] The baseline drift coefficient and waveform integrity coefficient are weighted and summed to obtain the artifact removal qualification coefficient.

[0032] Optionally, the steps for determining whether the artifact removal of the user's ECG signal is satisfactory based on the artifact removal pass coefficient are as follows:

[0033] The artifact removal pass coefficient is compared with the preset artifact removal pass coefficient threshold. If the artifact removal pass coefficient is not less than the preset artifact removal pass coefficient threshold, it means that the artifact removal of the user's ECG signal is qualified. Then the second ECG signal is used as the final user's ECG signal for fatigue data analysis.

[0034] If the artifact removal pass coefficient is less than the preset artifact removal pass coefficient threshold, it means that the artifact removal of the user's ECG signal is unqualified. Then, the motion artifact removal of the first ECG signal is performed again on the second ECG signal until the second ECG signal is qualified. The second ECG signal is then used as the end user's ECG signal for fatigue data analysis.

[0035] In a second aspect of this invention, a fatigue resistance data analysis method is proposed, the method comprising:

[0036] Heart rate variability and QT interval were extracted from the second ECG signal;

[0037] The user's blood oxygen saturation data is acquired through a PPG sensor. The blood oxygen saturation data includes the average blood oxygen saturation, fluctuation range, and hypoxia threshold.

[0038] The user's heart rate data is acquired through a PPG sensor, and the heart rate data includes heart rate fluctuations and recovery time.

[0039] The system combines heart rate variability, QT interval, average blood oxygen saturation, fluctuation amplitude and hypoxia threshold, heart rate fluctuation and recovery time into a comprehensive physiological feature vector, inputs it into a machine learning model to assess fatigue, outputs the user's fatigue level, and issues a rest alert based on the fatigue level.

[0040] The beneficial effects of this invention are:

[0041] This invention proposes a wearable physiological monitoring system and its anti-fatigue data analysis method. A physiological data acquisition system is installed in the user's wearable device, and the user's ECG signal is collected as the first ECG signal to determine the user's motion state. Different sampling methods and motion artifact removal methods are applied to the user's ECG signal based on the motion state to obtain the user's second ECG signal. The artifact removal of the user's ECG signal is determined based on the first and second ECG signals. If it is satisfactory, the second ECG signal is used as the final user's ECG signal for anti-fatigue data analysis. If it is unsatisfactory, the motion artifact removal process is repeated on the second ECG signal until the second ECG signal is satisfactory. This method removes motion artifacts from the collected physiological data, reducing adverse effects on the user's anti-fatigue data analysis process; it ensures that the user's ECG signal is not distorted, resulting in accurate fatigue state assessment and preventing false health warnings. Attached Figure Description

[0042] The invention will now be further described with reference to the accompanying drawings.

[0043] Figure 1 This is a framework diagram of a wearable physiological monitoring system;

[0044] Figure 2 This is a flowchart of a fatigue data analysis method. Detailed Implementation

[0045] 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.

[0046] Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this invention.

[0047] This invention provides a wearable physiological monitoring system. See also... Figure 1 , Figure 1 This is a framework diagram of a wearable physiological monitoring system provided in an embodiment of the present invention. The system includes:

[0048] Data acquisition module: A physiological data acquisition system is set up in the user's wearable device, and the user's ECG signal is collected by the physiological data acquisition system as the first ECG signal to determine the user's movement status;

[0049] Artifact Removal Module: Based on the user's motion state, different sampling methods and motion artifact removal methods are set for the user's ECG signal to obtain the user's second ECG signal;

[0050] First analysis module: Determine whether the artifact removal of the user's ECG signal is qualified based on the first ECG signal and the second ECG signal. If qualified, the second ECG signal is used as the final user's ECG signal for fatigue data analysis.

[0051] Second analysis module: If it fails, the motion artifact removal of the first ECG signal is performed again on the second ECG signal until the second ECG signal passes.

[0052] Based on the wearable physiological monitoring system provided in this embodiment of the invention, the above-mentioned method can remove motion artifacts from the collected user's physiological data, reduce the adverse effects on the user's anti-fatigue data analysis process, ensure that the user's electrocardiogram signal is not distorted, make the fatigue state assessment accurate, and avoid leading to false health warnings.

[0053] In one embodiment, the physiological data acquisition system is applied in a user wearable device, and the physiological data acquisition system includes an accelerometer, an ECG sensor, and a PPG sensor;

[0054] The ECG sensor is used to collect the user's ECG signal; the accelerometer is used to determine the user's current motion state, which includes being at rest, walking, and strenuous exercise; the PPG sensor is used to detect the user's blood oxygen saturation and heart rate, and to perform anti-fatigue data analysis on the user in conjunction with the ECG signal.

[0055] It should be noted that,

[0056] In this embodiment, the wearable device worn by the user is equipped with a physiological data acquisition system for real-time monitoring of the user's physiological state and analysis of anti-fatigue data. Specifically, it includes the following components:

[0057] 1. Accelerometer Function: The accelerometer is used to monitor the user's motion status. By detecting the magnitude and frequency of acceleration, it determines the user's motion type in real time.

[0058] Motion state judgment: stationary state: if the acceleration amplitude is close to 9.8 m / s2 (gravitational acceleration) and the acceleration change frequency does not fluctuate significantly, it is judged to be stationary.

[0059] Walking state: If the acceleration amplitude changes moderately and the periodic changes are strong, it is judged as walking state.

[0060] Vigorous motion state: If the acceleration amplitude changes significantly and the frequency fluctuates strongly, it is judged as a vigorous motion state.

[0061] 2. ECG Sensor Function: The ECG sensor is used to acquire the user's electrocardiogram (ECG) signal. The ECG signal reflects the user's cardiac electrical activity and is an important data source for assessing cardiac health and fatigue levels.

[0062] The role of ECG signals: to provide basic data for subsequent fatigue analysis and help determine the user's cardiac load and fatigue level.

[0063] 3. PPG Sensor Function: The PPG sensor is used to monitor the user's blood oxygen saturation (SpO2) and heart rate. PPG calculates blood oxygen levels and heart rate by analyzing the light absorption characteristics of blood flow in the blood vessels.

[0064] Blood oxygen saturation (SpO2): Monitors blood oxygen levels; low blood oxygen levels are often associated with excessive fatigue or lack of rest. Heart rate: Monitors heart rate fluctuations in real time; abnormal heart rate may be an indicator of fatigue.

[0065] 4. Correlation Analysis between Motion State and Signal

[0066] Accelerometer and Motion State Recognition: The accelerometer is used to determine the user's motion state, thereby providing motion intensity information. Depending on the motion state (stationary, walking, strenuous exercise), the system adjusts the signal acquisition and processing strategy to optimize ECG signal quality and avoid artifacts.

[0067] In one implementation, PPG is combined with ECG signals: combining blood oxygen saturation and heart rate data from ECG and PPG signals further enhances the accuracy of physiological data analysis and provides multi-dimensional support for fatigue state assessment.

[0068] In one embodiment, the accelerometer is used to determine the user's current motion state by:

[0069] The system collects the user's current acceleration amplitude and acceleration change frequency. If the acceleration amplitude is low, close to the gravitational acceleration value of 9.8 m / s², and the acceleration change frequency does not change significantly, then the user's current motion state is a stationary state.

[0070] If the acceleration amplitude changes to some extent, but the frequency of change is low and the periodicity is strong, then the user's current motion state is walking.

[0071] If the acceleration amplitude is large and the frequency of change is also large, then the user's current motion state is a state of intense motion.

[0072] It's important to note that the accelerometer determines a user's motion state in real time by collecting the user's acceleration amplitude and acceleration change frequency. Specifically, when the acceleration amplitude is close to the gravitational acceleration value of 9.8 m / s², and the acceleration change frequency does not fluctuate significantly, it indicates that the user is at rest, such as when the user is sitting or standing. The device's acceleration signal hardly changes, mainly due to the influence of Earth's gravity. If the acceleration amplitude changes, but the change frequency is low and highly periodic, it indicates that the user is engaged in low-intensity activity, such as walking. In this case, the acceleration signal shows relatively regular fluctuations, reflecting the undulations of the steps. Conversely, if the acceleration amplitude is large and the change frequency is also high, it indicates that the user is engaged in vigorous exercise, such as running or fast cycling. In this case, the acceleration signal fluctuates frequently and with large amplitudes, reflecting more intense physical activity. This method of determining motion state based on acceleration changes can effectively distinguish different types of motion states and provide a reliable basis for subsequent physiological data collection and analysis, ensuring more accurate fatigue assessment of users under different exercise conditions.

[0073] In one embodiment, setting different sampling methods and motion artifact removal methods for the user's ECG signal based on the user's motion state includes:

[0074] When the user's motion state is stationary, the sampling frequency of the ECG signal is set to no more than 250Hz, and the first ECG signal is filtered by a low-pass filter with a setting of 0.5-1Hz to remove low-frequency noise and motion artifacts; thus, the user's second ECG signal is obtained.

[0075] When the user's movement state is walking, the sampling frequency of the ECG signal is set to 500Hz, and the first ECG signal is processed by setting a high-pass filter, setting 1-2Hz to remove low-frequency noise, and then performing wavelet transformation on the removed low-frequency noise to remove high-frequency noise and motion artifacts; thus obtaining the user's second ECG signal.

[0076] When the user's motion state is in a state of intense motion, the sampling frequency of the ECG signal is set to 1000Hz, and the first ECG signal is separated into a mixed signal composed of motion artifacts and ECG signals by using blind source separation technology to extract the ECG signal. The extracted ECG signal is then subjected to wavelet transform to remove high-frequency noise and motion artifacts, thus obtaining the user's second ECG signal.

[0077] It's important to note that the system adapts its sampling frequency and artifact removal strategies to the ECG signal based on the user's different movement states, ensuring high-quality and accurate ECG signals under various exercise conditions. In a stationary state, the user's movement is minimal, and the ECG signal changes are relatively stable. Therefore, the system sets the sampling frequency to no more than 250Hz, sufficient to capture subtle changes in the ECG signal. At this time, a low-pass filter (0.5–1Hz) is used to remove low-frequency noise and baseline drift, effectively eliminating artifacts caused by minor skin vibrations or unstable device contact during rest. For example, when a user sits still in a chair for an extended period, signal fluctuations are minimal; the low-pass filter effectively removes these minor fluctuations, ensuring ECG signal stability and resulting in a clear second ECG signal.

[0078] During walking, the undulation of the gait and subtle body movements introduce noticeable low-frequency noise and artifacts into the ECG signal. Therefore, the system increases the sampling frequency to 500Hz to more accurately capture ECG changes. Simultaneously, a high-pass filter (1-2Hz) is used to remove low-frequency noise and eliminate low-frequency baseline drift caused by gait. Furthermore, the system uses wavelet transform to denoise the ECG signal, effectively removing high-frequency noise generated during walking. For example, when a user is walking easily, the ECG signal fluctuates relatively regularly, but the undulation of the gait introduces periodic low-frequency noise. A high-pass filter removes these low-frequency components, while wavelet transform effectively filters out high-frequency artifacts caused by gait, ensuring that signal quality remains unaffected.

[0079] During strenuous exercise, especially running or high-intensity workouts, users experience significant movement, leading to stronger motion artifacts in the ECG signal. This necessitates higher sampling frequencies and more complex artifact removal strategies. The system sets the sampling frequency to 1000Hz to ensure the capture of rapidly changing ECG signals. Simultaneously, the system employs blind source separation techniques (such as ICA) to separate motion-induced artifacts from the mixed signal, distinguishing them from the true ECG signal. This process effectively separates and removes motion artifacts, preserving the cardiac electrical activity signal. Next, the system uses wavelet transform to denoise the extracted ECG signal, further removing high-frequency noise and motion artifacts. For example, during strenuous running, the ECG signal is significantly affected by motion artifacts due to intense movement and muscle activity. Blind source separation technology can separate these artifacts from the true ECG waveform, ensuring the system acquires accurate ECG data and providing a reliable foundation for subsequent fatigue-related data analysis.

[0080] In one implementation, by dynamically adjusting the sampling frequency and artifact removal strategy based on the motion state, the system can ensure accurate and clear ECG signals under various motion environments, greatly improving the accuracy and reliability of fatigue analysis. Whether the user is stationary, walking, or engaged in strenuous activity, the system can flexibly adjust the signal processing method according to the real-time motion situation, effectively removing motion artifacts and ensuring that the fatigue state assessment results are not interfered with by motion artifacts.

[0081] In one embodiment, the step of determining whether the artifact removal of the user's ECG signal is satisfactory based on the first ECG signal and the second ECG signal is as follows:

[0082] The second ECG signal is low-pass filtered to remove high-frequency components while retaining low-frequency baseline information;

[0083] The baseline of the ECG signal is fitted using the least squares method to obtain the fitted baseline. The baseline drift coefficient is then calculated based on the baseline of the ECG signal and the second ECG signal. The calculation formula is as follows:

[0084]

[0085] In the formula, GH is the baseline drift coefficient, ECG(t) is the value of the second ECG signal at time point t, the fitted baseline(t) is the fitted baseline at time point t, and T is the signal duration;

[0086] The artifact removal pass coefficient is calculated based on the baseline drift coefficient and the waveform of the second ECG signal, and the artifact removal pass coefficient is used to determine whether the artifacts of the user's ECG signal have been removed successfully.

[0087] In one embodiment, the step of calculating the artifact removal qualification coefficient based on the baseline drift coefficient and the waveform of the second ECG signal is as follows:

[0088] The Pan-Tompkins algorithm was used to extract the P wave, QRS complex, and T wave from the second ECG signal.

[0089] The extracted P-wave, QRS complex, and T-wave are compared with their corresponding normal waveforms. The similarity of amplitude, shape, and duration of each waveform is calculated, and the average similarity is used as the similarity of the corresponding waveform. The similarity of all waveforms is summed and divided by 3 to obtain the waveform integrity coefficient of the second ECG signal.

[0090] The baseline drift coefficient and waveform integrity coefficient are weighted and summed to obtain the artifact removal qualification coefficient.

[0091] It should be noted that the second ECG signal undergoes low-pass filtering to remove high-frequency components. The low-pass filter can be set with an appropriate cutoff frequency (typically between 50Hz and 100Hz) to remove electrical interference and other high-frequency noise while preserving the low-frequency baseline information in the ECG signal. This process removes short-term noise caused by motion artifacts, resulting in a smoother signal.

[0092] It should be noted that the weighted summation of the baseline drift coefficient and the waveform integrity coefficient yields the artifact removal qualification coefficient as follows: GHY=a2×df-a1×fg, where GHY is the artifact removal qualification coefficient, fg and df are the baseline drift coefficient and the waveform integrity coefficient, respectively, and a1 and a2 are the preset proportional coefficients of fg and df, respectively, and both a1 and a2 are greater than 0;

[0093] It should be noted that a1 and a2 are set by professionals based on the actual situation. Generally, the sum of a1 and a2 is 1. For example, a1 and a2 can be 0.4 and 0.6 respectively, or other numbers. There are no specific restrictions. In addition, before applying the artifact removal qualification coefficient, the baseline drift coefficient and waveform integrity coefficient need to be normalized. Commonly used normalization methods include Min-Max normalization, Z-Score normalization, etc. The specific method is selected by professionals based on the actual situation. There are no specific restrictions or details.

[0094] It's important to note that the baseline drift coefficient refers to the degree of deviation between the second ECG signal and the fitted baseline, reflecting whether the low-frequency components of the ECG signal are affected by motion artifacts. A larger baseline drift coefficient indicates significant baseline fluctuations in the signal, usually caused by motion-induced artifacts (such as arm vibrations during running causing baseline drift in the ECG signal). The waveform integrity coefficient measures the similarity between the shape, amplitude, and duration of the P wave, QRS complex, and T wave in the second ECG signal and normal waveforms. A larger waveform integrity coefficient indicates that these key waveforms are closer to the characteristics of a normal ECG, with less artifact influence and higher signal quality.

[0095] A low baseline drift coefficient indicates that low-frequency noise and motion artifacts in the signal have been effectively removed, the baseline is more stable, and the signal response is more accurate. A high waveform integrity coefficient indicates that the morphology of the P wave, QRS complex, and T wave is closer to that of a normal ECG, suggesting that key signal features have not been distorted and can still be used to assess cardiac health. If both indicators are within the ideal range, it means that the ECG signal is not only unaffected by artifacts but also maintains high accuracy and reliability, thus it can be safely used for subsequent fatigue data analysis and health assessment.

[0096] For example, suppose a user wears a smartwatch during strenuous exercise. The wrist vibrations generated during exercise may cause significant baseline drift in the ECG signal, resulting in severe fluctuations in the low-frequency components. In this case, the baseline drift coefficient is large, while the waveform integrity coefficient is small, indicating that motion artifacts have not been completely removed. After a series of filtering and denoising processes, the baseline drift coefficient is reduced to a lower level, the waveform integrity coefficient is improved, and the P-wave, QRS complex, and T-wave shapes of the second ECG signal are close to normal. The artifacts are effectively removed, and the signal quality is restored to a higher level. At this point, this signal can be used as a valid ECG signal for the user, providing accurate information for assessing the user's fatigue state and health status.

[0097] In one implementation, analyzing the baseline drift coefficient and waveform integrity coefficient offers significant advantages in determining whether motion artifacts have been effectively removed from a user's ECG signal and whether subsequent fatigue-resistant data analysis is feasible. Firstly, the baseline drift coefficient, by quantitatively measuring the level of low-frequency noise in the signal, accurately assesses the impact of motion artifacts on the ECG signal. A smaller baseline drift coefficient indicates that low-frequency components have been effectively removed, ensuring signal quality and accurately reflecting the true activity of the heart. Secondly, the waveform integrity coefficient analyzes the morphology, amplitude, and duration of key waveforms such as the P wave, QRS complex, and T wave to ensure that the main physiological characteristics of the ECG signal are not distorted, which is crucial for assessing the user's health status. The combination of these two factors comprehensively evaluates the reliability of the ECG signal, avoids artifacts caused by motion artifacts, and ensures the authenticity and accuracy of the data. This method continuously monitors and optimizes ECG signals during dynamic exercise, providing accurate data support for fatigue analysis, reducing the risk of misjudgment due to data errors, thereby improving the reliability of health management and ensuring that cardiac health assessments of users under fatigue are not biased. Through this precise signal processing and analysis, it can better reflect the user's fatigue level, provide timely and effective health warnings, and help users take appropriate rest and recovery measures to avoid the potential health hazards of excessive fatigue.

[0098] In one embodiment, the step of determining whether the artifact removal of the user's ECG signal is satisfactory based on the artifact removal pass coefficient is as follows:

[0099] The artifact removal pass coefficient is compared with the preset artifact removal pass coefficient threshold. If the artifact removal pass coefficient is not less than the preset artifact removal pass coefficient threshold, it means that the artifact removal of the user's ECG signal is qualified. Then the second ECG signal is used as the final user's ECG signal for fatigue data analysis.

[0100] If the artifact removal pass coefficient is less than the preset artifact removal pass coefficient threshold, it means that the artifact removal of the user's ECG signal is unqualified. Then, the motion artifact removal of the first ECG signal is performed again on the second ECG signal until the second ECG signal is qualified. The second ECG signal is then used as the end user's ECG signal for fatigue data analysis.

[0101] It's important to note that the step of determining whether the artifact removal in a user's ECG signal is adequately removed based on the artifact removal pass coefficient is done by comparing the pass coefficient with a preset threshold to determine if the signal quality meets the standard. In this process, the artifact removal pass coefficient is first calculated by measuring the baseline drift coefficient and waveform integrity coefficient of the second ECG signal. If this coefficient is greater than or equal to the preset artifact removal pass coefficient threshold, it indicates that motion artifacts in the ECG signal have been effectively removed, the signal quality meets the analysis requirements, and it can then be used as the final user ECG signal for fatigue data analysis. This step ensures data accuracy and avoids analysis errors caused by motion artifacts. For example, when a user engages in strenuous exercise, significant motion artifacts may occur due to unstable device wear or intense muscle movement. After signal processing, if the artifact removal pass coefficient meets the standard, it indicates that the artifacts have been successfully removed, the signal restores the true waveform of the ECG, and thus effective fatigue assessment can be performed.

[0102] Conversely, if the artifact removal pass coefficient is less than the preset threshold, it indicates that there are still significant artifacts in the signal, causing ECG signal distortion and making it unsuitable for further analysis. In this case, the system will restart the motion artifact removal process, performing further signal processing until the signal meets the pass standard. For example, suppose a user is running; due to incomplete removal of motion artifacts, the P and T wave morphologies in the signal may be significantly distorted, resulting in a low artifact removal pass coefficient. In this situation, the system will readjust the signal processing method, re-performing denoising and filtering until the artifact removal pass coefficient meets the threshold requirement. Ultimately, the second ECG signal, after multiple optimizations, can provide accurate and reliable physiological data, ensuring that subsequent fatigue analysis is based on real and clear ECG signals, avoiding erroneous health assessments caused by artifacts.

[0103] Based on the same inventive concept, this invention also provides a method for fatigue data analysis. See [link to related document]. Figure 2 , Figure 2 A flowchart of an anti-fatigue data analysis method provided in an embodiment of the present invention, the method comprising:

[0104] Heart rate variability and QT interval were extracted from the second ECG signal;

[0105] The user's blood oxygen saturation data is acquired through a PPG sensor. The blood oxygen saturation data includes the average blood oxygen saturation, fluctuation range, and hypoxia threshold.

[0106] The user's heart rate data is acquired through a PPG sensor, and the heart rate data includes heart rate fluctuations and recovery time.

[0107] The system combines heart rate variability, QT interval, average blood oxygen saturation, fluctuation amplitude and hypoxia threshold, heart rate fluctuation and recovery time into a comprehensive physiological feature vector, inputs it into a machine learning model to assess fatigue, outputs the user's fatigue level, and issues a rest alert based on the fatigue level.

[0108] It should be noted that heart rate variability (HRV) and QT interval are extracted from the second ECG signal. Heart rate variability reflects the heart's self-regulation ability, and it usually decreases under fatigue. The QT interval represents the time interval on the electrocardiogram, which is normally related to the heart's health. An excessively long QT interval may be a sign of fatigue or overexertion.

[0109] Blood Oxygen Saturation Data Acquisition: Blood oxygen saturation data is acquired through a PPG sensor, including: Average Blood Oxygen Saturation: Reflects the user's overall oxygenation level. Low blood oxygen may indicate fatigue or discomfort. Fluctuation Range: Reflects the fluctuation in blood oxygen saturation. Large fluctuations may indicate abnormal changes in blood oxygen during exercise or at rest, suggesting potential signs of fatigue. Hypoxia Threshold: Refers to the duration or frequency of blood oxygen saturation falling below a certain threshold. Prolonged hypoxia can affect recovery and may be associated with over-fatigue or health problems.

[0110] Heart rate data acquisition: This involves obtaining the user's heart rate data via a PPG sensor, including:

[0111] Heart rate fluctuations: These indicate how heart rate changes at different points in time. Larger fluctuations may be related to fatigue, stress, or physical exertion.

[0112] Recovery time: This refers to the time required for the heart rate to return to its resting state after exercise. An excessively long recovery time may indicate slow physical recovery and suggest a higher level of fatigue.

[0113] Comprehensive physiological feature vector construction: The extracted physiological features (heart rate variability, QT interval, average blood oxygen saturation, fluctuation amplitude, hypoxia threshold, heart rate fluctuation, recovery time) are integrated into a comprehensive physiological feature vector, which covers multiple dimensions of information such as the user's cardiac health status, blood oxygen level, and heart rate changes.

[0114] Fatigue assessment using machine learning models: The physiological feature vector is input into a trained machine learning model (such as a support vector machine, decision tree, neural network, etc.) to perform fatigue assessment. Based on historical data and labeled fatigue levels, the model analyzes the user's current physiological state and outputs a fatigue level (such as low, medium, or high fatigue level).

[0115] Rest Alert: Based on the fatigue level results, if the model determines that the user's fatigue level has reached a certain threshold (for example, the fatigue level is "high"), the system will automatically issue a rest alert to remind the user to rest, avoid overwork, and reduce health risks.

[0116] For example, assuming a user wears a wearable device after a day's work, the system extracts heart rate variability and QT interval from the ECG signal. The results show a significant decrease in heart rate variability and a prolonged QT interval, indicating a decline in the user's cardiac self-regulation ability, possibly due to accumulated fatigue. Simultaneously, the PPG sensor detects that the user's average blood oxygen saturation is slightly below normal, with large fluctuations and a high hypoxia threshold, suggesting fluctuations in oxygenation levels, possibly a sign of fatigue from prolonged work. Combining heart rate fluctuations with recovery time, the user's large heart rate fluctuations and long recovery time further confirm a decline in the user's physical recovery ability.

[0117] In one implementation, this physiological data is integrated into a feature vector, which is then input into a machine learning model. The model determines that the user is at a high level of fatigue, triggering a rest alarm to remind the user to rest in time and avoid health damage from excessive fatigue. Through this comprehensive analysis of multi-dimensional data, the system can accurately assess the user's fatigue state and provide scientific health warnings.

[0118] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the present invention should still fall within the patent coverage of the present invention.

Claims

1. A wearable physiological monitoring system, characterized in that, The system includes: Data acquisition module: A physiological data acquisition system is set up in the user's wearable device, and the user's ECG signal is collected by the physiological data acquisition system as the first ECG signal to determine the user's movement status; Artifact Removal Module: Based on the user's motion state, different sampling methods and motion artifact removal methods are set for the user's ECG signal to obtain the user's second ECG signal; First analysis module: Determines whether the artifact removal of the user's ECG signal is satisfactory based on the first and second ECG signals. If satisfactory, the second ECG signal is used as the final user's ECG signal for fatigue data analysis. The steps for determining whether the artifact removal of the user's ECG signal is satisfactory based on the first and second ECG signals are as follows: The second ECG signal is low-pass filtered to remove high-frequency components while retaining low-frequency baseline information; The baseline of the ECG signal is fitted using the least squares method to obtain the fitted baseline. The baseline drift coefficient is then calculated based on the baseline of the ECG signal and the second ECG signal. The calculation formula is as follows: In the formula, GH is the baseline drift coefficient, ECG(t) is the value of the second ECG signal at time point t, the fitted baseline(t) is the fitted baseline at time point t, and T is the signal duration; The artifact removal pass coefficient is calculated based on the baseline drift coefficient and the waveform of the second ECG signal, and the artifact removal pass coefficient is used to determine whether the artifacts of the user's ECG signal have been removed successfully. Second analysis module: If it fails, the motion artifact removal of the first ECG signal is performed again on the second ECG signal until the second ECG signal passes.

2. The wearable physiological monitoring system according to claim 1, characterized in that, Setting up a physiological data acquisition system in a user's wearable device includes: The physiological data acquisition system is used in user wearable devices, and the physiological data acquisition system includes an accelerometer, an ECG sensor, and a PPG sensor; The ECG sensor is used to collect the user's ECG signal; the accelerometer is used to determine the user's current motion state, which includes being at rest, walking, and strenuous exercise; the PPG sensor is used to detect the user's blood oxygen saturation and heart rate, and to perform anti-fatigue data analysis on the user in conjunction with the ECG signal.

3. The wearable physiological monitoring system according to claim 2, characterized in that, The accelerometer is used to determine the user's current motion state, including: Collect the user's current acceleration amplitude and acceleration change frequency; if the acceleration amplitude is low, close to the gravitational acceleration value of 9.8 m / s², it indicates a potential acceleration problem. 2 If the frequency of acceleration change does not change significantly, then the user's current motion state is a stationary state. If the acceleration amplitude changes to some extent, but the frequency of change is low and the periodicity is strong, then the user's current motion state is walking. If the acceleration amplitude is large and the frequency of change is also large, then the user's current motion state is a state of intense motion.

4. The wearable physiological monitoring system according to claim 1, characterized in that, Different sampling methods and motion artifact removal methods are set for the user's ECG signal based on the user's motion status, including: When the user's motion state is stationary, the sampling frequency of the ECG signal is set to no more than 250Hz, and the first ECG signal is filtered by a low-pass filter with a setting of 0.5-1Hz to remove low-frequency noise and motion artifacts; thus, the user's second ECG signal is obtained. When the user's movement state is walking, the sampling frequency of the ECG signal is set to 500Hz, and the first ECG signal is processed by setting a high-pass filter, setting 1-2Hz to remove low-frequency noise, and then performing wavelet transformation on the removed low-frequency noise to remove high-frequency noise and motion artifacts; thus obtaining the user's second ECG signal. When the user's motion state is in a state of intense motion, the sampling frequency of the ECG signal is set to 1000Hz, and the first ECG signal is separated into a mixed signal composed of motion artifacts and ECG signals by using blind source separation technology to extract the ECG signal. The extracted ECG signal is then subjected to wavelet transform to remove high-frequency noise and motion artifacts, thus obtaining the user's second ECG signal.

5. A wearable physiological monitoring system according to claim 1, characterized in that, The steps for calculating the artifact removal pass factor based on the baseline drift factor and the waveform of the second ECG signal are as follows: The Pan-Tompkins algorithm was used to extract the P wave, QRS complex, and T wave from the second ECG signal. The extracted P-wave, QRS complex, and T-wave are compared with their corresponding normal waveforms. The similarity of amplitude, shape, and duration of each waveform is calculated, and the average similarity is used as the similarity of the corresponding waveform. The similarity of all waveforms is summed and divided by 3 to obtain the waveform integrity coefficient of the second ECG signal. The baseline drift coefficient and waveform integrity coefficient are weighted and summed to obtain the artifact removal qualification coefficient.

6. The wearable physiological monitoring system according to claim 1, characterized in that, The steps to determine whether the artifact removal of a user's ECG signal is satisfactory based on the artifact removal pass coefficient are as follows: The artifact removal pass coefficient is compared with the preset artifact removal pass coefficient threshold. If the artifact removal pass coefficient is not less than the preset artifact removal pass coefficient threshold, it means that the artifact removal of the user's ECG signal is qualified. Then the second ECG signal is used as the final user's ECG signal for fatigue data analysis. If the artifact removal pass coefficient is less than the preset artifact removal pass coefficient threshold, it means that the artifact removal of the user's ECG signal is unqualified. Then, the motion artifact removal of the first ECG signal is performed again on the second ECG signal until the second ECG signal is qualified. The second ECG signal is then used as the end user's ECG signal for fatigue data analysis.

7. A method for analyzing fatigue data, implemented using a wearable physiological monitoring system as described in any one of claims 1-6, characterized in that, The method includes: Heart rate variability and QT interval were extracted from the second ECG signal; The user's blood oxygen saturation data is acquired through a PPG sensor. The blood oxygen saturation data includes the average blood oxygen saturation, fluctuation range, and hypoxia threshold. The user's heart rate data is acquired through a PPG sensor, and the heart rate data includes heart rate fluctuations and recovery time. The system combines heart rate variability, QT interval, average blood oxygen saturation, fluctuation amplitude and hypoxia threshold, heart rate fluctuation and recovery time into a comprehensive physiological feature vector, inputs it into a machine learning model to assess fatigue, outputs the user's fatigue level, and issues a rest alert based on the fatigue level.

Images