Dynamic risk assessment method and device based on wearable device

Abstract

The invention relates to the technical field of exercise physiological monitoring, and particularly discloses a dynamic risk assessment method and equipment based on wearable equipment, the method and equipment are applied to wearable monitoring equipment, the wearable monitoring equipment is worn on exercise limbs of a user, and the method comprises the following steps: acquiring original electromyographic signals generated by limb muscles during exercise of the user; acquiring muscle change data of limbs in movement of the user, and acquiring muscle contractility data according to the muscle change data; preprocessing the original electromyographic signals to obtain digital electromyographic signals, and acquiring electromyographic characteristic data according to the digital electromyographic signals and the muscle change data; and generating a real-time dynamic risk assessment value according to the myoelectricity characteristic data and the muscle contractility data. According to the method, the electromyographic signals and the muscle mechanics data are collected in real time through the wearable device, and accurate assessment and early warning of the exercise fatigue risk are achieved in combination with the multi-source information fusion and dynamic analysis technology.

Description

Technical Field

[0001] This invention relates to the field of exercise physiological monitoring technology, and in particular to a dynamic risk assessment method and device based on wearable devices. Background Technology

[0002] With the increasing awareness of health among the general public, sports training, daily fitness, and high-intensity physical activities are becoming more and more popular, and users' demand for monitoring their own physiological state and fatigue risk during exercise has increased significantly. At the same time, the technology of smart wearable devices (such as smartwatches, heart rate belts, electromyography sensors, etc.) has developed rapidly and is now capable of collecting physiological and exercise data such as heart rate, muscle electrical activity, blood oxygen saturation, and range of motion in real time.

[0003] Existing dynamic risk assessment technologies based on wearable devices typically involve comprehensive analysis of multiple physiological and motion parameters, such as electrocardiogram (ECG) signals and blood oxygen saturation. While these methods can acquire relatively rich user status information, they suffer from significant efficiency drawbacks in practical applications. The need to process multiple high-frequency physiological signals simultaneously leads to a substantial increase in system computational load. Particularly in long-term motion monitoring scenarios, continuous multi-dimensional data acquisition and processing significantly shortens device battery life and increases the pressure on data transmission and storage. Furthermore, the complexity of multi-parameter fusion analysis is high, making it difficult for traditional algorithms to achieve accurate risk assessment while maintaining real-time performance. Summary of the Invention

[0004] The purpose of this invention is to provide a dynamic risk assessment method and device based on wearable devices to solve the technical problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A dynamic risk assessment method based on wearable devices, applied to wearable monitoring devices worn on a user's moving limbs, comprising: Acquire raw electromyographic signals generated by the user's limb muscles during exercise; Acquire muscle change data of the user's limbs during movement, and obtain muscle contraction force data based on the muscle change data; The raw electromyography (EMG) signal is preprocessed to obtain a digital EMG signal, and EMG feature data is obtained based on the digital EMG signal and muscle change data. A real-time dynamic risk assessment value is generated based on the electromyographic characteristic data and muscle contraction force data.

[0006] Preferably, the step of acquiring raw electromyographic signals generated by the user's limb muscles during exercise includes: Obtain the raised area when the user's limb muscles contract during exercise; Multiple muscle pressure feedback values ​​of the muscle bulge area to the wearable monitoring device are obtained, and the location of the maximum muscle pressure feedback value is used as the center point of muscle monitoring. Multiple muscle-skin electrical signals within a preset distance are obtained based on the muscle monitoring center point, and the largest and smallest muscle-skin electrical signals are obtained. The raw electromyographic (EMG) signals are obtained based on the largest and smallest muscle-skin EMG signals.

[0007] Preferably, the step of preprocessing the raw electromyography (EMG) signal to obtain a digital EMG signal includes: The low-frequency baseline drift interference in the original electromyography signal is filtered out by a high-pass filter circuit to obtain the first filtered original electromyography signal; The second filtered original electromyography signal is obtained by filtering out the power frequency interference signal generated by the wearable monitoring device in the first filtered original electromyography signal through a low-pass filter circuit. The second filtered raw electromyographic signal is amplified to obtain the amplified filtered raw electromyographic signal. The original electromyographic signal, after being amplified and filtered, is converted into a discrete digital signal to obtain a digital electromyographic signal.

[0008] Preferably, the step of obtaining electromyographic feature data based on the digital electromyographic signals and muscle change data includes: Based on the digital electromyography (EMG) signal, the EMG time-domain characteristics of the electrical activity during muscle contraction are obtained, and the digital EMG signal is converted into EMG frequency-domain characteristics. The digital electromyography (EMG) signals and muscle change data are aligned according to the acquisition time, and muscle change correlation data corresponding to the digital EMG signals are obtained based on the aligned muscle change data. The muscle change correlation data includes the change in muscle contraction amplitude, the duration of muscle activation state, and the synchronicity of muscle electrical activity. By simultaneously associating electromyographic time-domain features and electromyographic frequency-domain features with changes in muscle contraction amplitude, duration of muscle activation, and synchronicity of muscle electrical activity, an electromyographic-muscle association dataset of electromyographic features and muscle changes is obtained. Electromyographic feature data were obtained from the electromyographic-muscle association dataset.

[0009] Preferably, the step of aligning the digital electromyography (EMG) signal and muscle change data according to the acquisition time, and obtaining muscle change correlation data corresponding to the digital EMG signal based on the aligned muscle change data, includes: Multiple time markers for the acquisition time of digital electromyography (EMG) signals are obtained. The start and end times of a segment of digital EMG signal corresponding to each time marker are obtained. Time markers with the same time interval are obtained from muscle change data based on the start and end times. The digital electromyography signal and muscle change data are paired according to the extracted time marker points to generate paired data sets of signal and muscle change data. For each time marker, the paired data sets are extracted from the aligned muscle change data to obtain the dimensional information associated with the corresponding digital electromyography (EMG) signal. The morphological manifestation of muscle contraction in the muscle change data is obtained to obtain the change in muscle contraction amplitude. At the same time, the duration of muscle activation is obtained to obtain the duration of muscle activation. The activity performance of different parts of the muscle is compared to obtain the spatial synchronicity of muscle electrical activity. The changes in muscle contraction amplitude, duration of muscle activation, and spatial synchronicity of muscle electrical activity extracted from each pair of data sets are bound to the corresponding digital electromyography signal segments according to the acquisition time sequence of the time marker points to generate muscle change correlation data.

[0010] Preferably, the step of simultaneously associating electromyographic time-domain features and electromyographic frequency-domain features with changes in muscle contraction amplitude, duration of muscle activation, and synchronicity of muscle electrical activity to obtain an electromyographic-muscle association dataset of electromyographic features and muscle changes includes: The temporal characteristics of electromyography (EMG) are correlated with the changes in muscle contraction amplitude. When the intensity of electrical activity changes in the temporal characteristics of EMG, the corresponding changes in muscle contraction amplitude are recorded to obtain temporal-muscle contraction amplitude change data. The electromyography time-domain features are correlated with the duration of muscle activation, and the change trend of energy accumulation in the electromyography time-domain features is recorded during the duration of muscle activation to obtain time-domain muscle activation duration data. The correspondence between the frequency distribution of electromyography (EMG) frequency domain features and the synchronicity of muscle electrical activity is obtained, and the degree of concentration of the frequency distribution in the EMG frequency domain features is obtained based on the synchronicity of muscle electrical activity, thus obtaining frequency domain-muscle electrical activity synchronicity data. The electromyography-muscle association dataset is obtained by integrating time-domain data on changes in muscle contraction amplitude, time-domain data on duration of muscle activation, and frequency-domain data on synchronicity of muscle electrical activity.

[0011] Preferably, the step of generating a real-time dynamic risk assessment value based on the electromyographic feature data and muscle contraction force data includes: Based on the electromyographic characteristic data, the frequency distribution changes and electrical activity intensity changes are obtained; Based on the frequency distribution changes and muscle contraction force data, multiple muscle contraction degrees are obtained, and the smoothness of the movement is obtained based on the multiple muscle contraction degrees. Based on the changes in electrical activity intensity and muscle contraction force data, multiple muscle exertion levels are obtained, and exertion stability is obtained based on multiple muscle exertion levels; A real-time dynamic risk assessment value is generated based on the smoothness of the action and the stability of the force exertion. Obtain a historical dynamic risk assessment set, and generate a risk assessment result based on the historical dynamic risk assessment set and the real-time dynamic risk assessment value.

[0012] Preferably, the step of generating a risk assessment result based on the historical dynamic risk assessment set and the real-time dynamic risk assessment value includes: Multiple historical dynamic risk assessment values ​​are obtained based on the aforementioned historical dynamic risk assessment set; The movement change trend is generated based on multiple historical dynamic risk assessment values ​​and real-time dynamic risk assessment values; A trend difference value is generated based on the real-time dynamic risk assessment value and the historical dynamic risk assessment value with the closest recording time. If the trend is upward, it is determined that the user is at risk of exercise fatigue, and a risk assessment result for warning is generated based on the trend difference value. If the trend is downward or stable, it is determined that the user is not at risk of exercise fatigue, and the real-time dynamic risk assessment value is incorporated into the historical dynamic risk assessment set as the latest historical dynamic risk assessment value.

[0013] The present invention also provides a dynamic risk assessment system based on wearable devices, comprising: The muscle contraction force data acquisition module is used to acquire muscle change data of the user's limbs during movement, and to acquire muscle contraction force data based on the muscle change data. The electromyography feature data acquisition module is used to preprocess the raw electromyography signal to obtain a digital electromyography signal, and to acquire electromyography feature data based on the digital electromyography signal and muscle change data. The real-time dynamic risk assessment value acquisition module is used to generate a real-time dynamic risk assessment value based on the electromyographic characteristic data and muscle contraction force data.

[0014] Preferably, the raw electromyography signal acquisition module includes: The raised area acquisition unit is used to acquire the raised area when the user's limb muscles contract during movement; The muscle monitoring center point acquisition unit is used to acquire multiple muscle pressure feedback values ​​of the muscle bulge area to the wearable monitoring device, and to take the position of the maximum muscle pressure feedback value as the muscle monitoring center point. The muscle-skin electrical signal acquisition unit is used to acquire multiple muscle-skin electrical signals within a preset distance based on the muscle monitoring center point, and to acquire the largest and smallest muscle-skin electrical signals. The raw electromyography (EMG) signal acquisition unit is used to acquire raw EMG signals based on the maximum and minimum muscle-skin EMG signals.

[0015] The beneficial effects of this application are as follows: This invention acquires electromyographic signals and muscle biomechanical data in real time through wearable devices, and combines multi-source information fusion and dynamic analysis technology to achieve accurate assessment and early warning of exercise fatigue risk. Compared with traditional static assessment methods, the use of dynamic positioning and multi-sensor collaborative strategies effectively overcomes the signal distortion problem caused by changes in muscle contraction position; through hierarchical filtering and signal enhancement processing, the quality and anti-interference ability of electromyographic signals are significantly improved; based on the correlation analysis of time-frequency domain features and muscle biomechanical parameters, a multi-dimensional electromyographic-muscle correlation dataset is constructed, enhancing the comprehensiveness and accuracy of state recognition; the introduction of historical data trend analysis and personalized difference assessment realizes adaptive adjustment of dynamic risk thresholds, avoiding the limitations of a one-size-fits-all standard; by generating risk assessment values ​​in real time and combining them with historical trend judgments, early warnings can be issued in the early stages of fatigue accumulation. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of a method flow according to an embodiment of the present invention.

[0017] Figure 2 This is a schematic diagram of the system structure according to an embodiment of the present invention.

[0018] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0020] like Figure 1 As shown, this application provides a dynamic risk assessment method based on wearable devices, applied to wearable monitoring devices worn on a user's moving limbs, including: S1. Acquire the raw electromyographic signals generated by the user's limb muscles during exercise; S2. Obtain muscle change data of the user's limbs during movement, and obtain muscle contraction force data based on the muscle change data; S3. The original electromyographic signal is preprocessed to obtain a digital electromyographic signal, and electromyographic feature data is obtained based on the digital electromyographic signal and muscle change data. S4. Generate a real-time dynamic risk assessment value based on the electromyographic characteristic data and muscle contraction force data.

[0021] As described in steps S1-S4 above, this invention is applied to a wearable monitoring device. This device has a contact component that directly contacts the user's limbs. The contact component contains a data acquisition unit, such as an electromyography (EMG) sensor and a pressure sensor. When the user begins to exercise, the acquisition unit continuously collects raw EMG signals and muscle change data generated by the user's limb muscles during exercise. Since muscle fatigue is a key factor affecting athletic performance and causing sports injuries in exercise physiology, raw EMG signals, as a direct representation of muscle electrical activity, can sensitively reflect changes in the functional state of the neuromuscular system. When muscles are fatigued, the EMG signals will... Electromyography (EMG) exhibits a series of characteristic changes. For example, in the time domain, the signal amplitude typically increases, indicating a need for greater discharge intensity to maintain contractile force. In the frequency domain, the power spectrum shifts towards lower frequencies, and the median frequency decreases significantly, indicating a decrease in discharge rate and a recruitment shift from fast-twitch to slow-twitch fibers. Compared to indirect physiological indicators such as heart rate and blood oxygen saturation, EMG signals can reflect muscle fatigue more directly and earlier. Traditional exercise risk assessment relies heavily on indirect indicators such as heart rate monitoring and blood oxygen saturation, but these indicators have limited sensitivity to muscle-specific fatigue and often only detect changes when fatigue is already significant. By monitoring the time and frequency domain characteristics of EMG signals in real time, fatigue trends can be identified before a substantial decline in muscle function occurs, achieving true early warning. Therefore, this invention obtains digital electromyography (EMG) signals by acquiring raw EMG signals and performing low-pass filtering. Based on the digital EMG signals and muscle change data, EMG feature data is obtained, whereby the EMG feature data refers to feature vectors reflecting the functional state of the muscles. During exercise, the muscle change data is processed and analyzed using a linear regression analysis algorithm to obtain muscle contraction force data. Finally, multi-feature weighted fusion calculations are performed on the EMG feature data and muscle contraction force data to generate a real-time dynamic risk assessment value. This real-time dynamic risk assessment value is matched with a historical dynamic risk assessment set to obtain a risk assessment result. The risk assessment result includes risk level and fatigue state. For example, when the user's exercise state is low-intensity and the fatigue level is mild, the risk level is low; when the user's exercise state is high-intensity and the fatigue level is severe, the risk level is high. This allows for accurate prediction results to be generated in the early stages of user fatigue, reducing the complexity of the algorithm and facilitating timely warnings to the user.

[0022] In one embodiment, the step of acquiring raw electromyographic signals generated by a user's limb muscles during exercise includes: S101. Obtain the raised area when the user's limb muscles contract during exercise; S102. Obtain multiple muscle pressure feedback values ​​of the muscle bulge area to the wearable monitoring device, and take the location of the maximum muscle pressure feedback value as the center point of muscle monitoring. S103. Based on the muscle monitoring center point, acquire multiple muscle-skin electrical signals within a preset distance, and acquire the largest and smallest muscle-skin electrical signals. S104. Obtain the raw electromyographic signals based on the largest and smallest muscle-skin electrical signals.

[0023] As described in steps S101-S104 above, the muscle bulge area refers to the region where the body surface morphology changes during muscle contraction. Specifically, this can be achieved by using a three-dimensional pressure sensor array to detect skin surface deformation, dynamically locating the core area of ​​muscle activity to reduce redundant data acquisition. The muscle pressure feedback value refers to the muscle contraction pressure value sensed by the sensor, which can be achieved using a piezoelectric thin-film sensor array. The optimal signal acquisition location is determined by comparing pressure distributions. The muscle monitoring center point refers to the sensor location with the highest pressure feedback value, which can be determined through a real-time pressure data comparison algorithm to ensure that signal acquisition is focused on the core area of ​​muscle activity. The preset distance refers to the physical spatial range set with the monitoring center point as the center, for example, a circular area with a radius of 2 cm, used to limit the signal acquisition range and avoid data redundancy.

[0024] During dynamic movement, the location and intensity of muscle contraction constantly change. Traditional fixed-point acquisition methods struggle to adapt to this dynamic characteristic, easily leading to signal distortion. This invention utilizes a pressure sensor array built into a wearable device to detect real-time changes in muscle morphology. During movement, the algorithm compares the pressure readings of each sensor with the baseline value in real time. For example, when the pressure increment in a certain area exceeds 150% of the baseline value for five consecutive sampling cycles, it is determined to be a valid bulge area. For instance, when the arm is bent, the biceps brachii contraction causes the pressure value to rise from the baseline of 5 kPa to 12 kPa, which is then marked as the area to be monitored.

[0025] All pressure sensor feedback values ​​within the raised area are extracted, and a pressure distribution map is constructed using a spatial interpolation algorithm. The maximum value of the distribution field is found using the gradient descent method, and its coordinates are defined as the muscle monitoring center point. For example, when the leg is extended, the maximum pressure distribution of the quadriceps contraction is 28 kPa, located at a certain coordinate on the sensor array; this point is set as the monitoring center point.

[0026] Centered on the muscle monitoring center point, a surface electrode array is activated within a preset radius to form a data acquisition array. The electrodes are evenly distributed, and the maximum and minimum values ​​of the signal from each electrode are extracted. During the strong muscle contraction phase, the electrodes record the maximum and minimum electrical signals.

[0027] The multi-channel extreme value signals are fused: First, the maximum and minimum electrical signals collected by each electrode are sorted, and outliers are removed using the 3σ criterion; then, the average extreme value difference of the effective signals is calculated as the representative value of the time window; finally, these representative values ​​are combined in time series to form the original electromyographic signal.

[0028] In one embodiment, the step of preprocessing the raw electromyography (EMG) signal to obtain a digital EMG signal includes: S301. The low-frequency baseline drift interference in the original electromyography signal is filtered out by a high-pass filter circuit to obtain the first filtered original electromyography signal; S302. The power frequency interference signal generated by the wearable monitoring device in the first filtered original electromyography signal is filtered out by a low-pass filter circuit to obtain the second filtered original electromyography signal. S303. The second filtered raw electromyographic signal is amplified to obtain an amplified filtered raw electromyographic signal. S304. Based on the amplified and filtered original electromyographic signal, perform discrete digital signal conversion to obtain a digital electromyographic signal.

[0029] As described in steps S301-S304 above, the present invention transforms the raw electromyographic signals collected by wearable devices, which contain various interferences (low-frequency baseline drift caused by movement will mask the effective signal components, and power frequency interference generated by the device itself will distort the signal waveform) into high-quality digital electromyographic signals through a multi-level signal processing flow.

[0030] A high-pass filter circuit is an electronic circuit used to filter out low-frequency components in a signal. Specifically, it can be implemented using a second-order Butterworth filter with a cutoff frequency of 20Hz. Its function is to eliminate baseline drift caused by changes in skin contact impedance. A low-pass filter circuit is an electronic circuit used to filter out high-frequency noise. Specifically, it can be implemented using a Chebyshev filter with a cutoff frequency of 500Hz. Its function is to suppress power frequency interference generated by equipment circuits. Signal amplification refers to adjusting the gain of the filtered signal. This can be achieved using a differential amplifier circuit combined with a programmable gain amplifier. Its function is to compensate for signal attenuation and improve the signal-to-noise ratio. Discrete-to-digital conversion refers to converting analog signals into digital signals, for example, using a 12-bit analog-to-digital converter with a 1kHz sampling rate. Its function is to generate standardized digital signals for subsequent algorithm processing.

[0031] The raw electromyography (EMG) signal is input to the filtering circuit. When low-frequency drift caused by body movement exists in the signal, the filter can effectively attenuate these interfering components while maintaining the integrity of the mid-to-high frequency EMG signal. For example, in the presence of significant motion artifacts, the filter can significantly suppress low-frequency fluctuations without affecting the main characteristics of the EMG signal. The first filtered raw EMG signal passes sequentially through a main low-pass filter and a fixed-point notch filter. The notch filter uses a dual-T active network, and through precise quality factor control, it ensures effective suppression of power frequency interference and its adjacent components. For example, when the signal contains significant power frequency interference, this processing can suppress it to a negligible level without causing significant damage to the EMG signal. The filtered signal needs to be appropriately amplified to meet digitization requirements. The amplification circuit adopts an instrumentation amplifier architecture, and the gain setting is precisely calculated to ensure that the amplitude of the amplified signal matches the input range of the subsequent analog-to-digital converter. The second filtered raw EMG signal first passes through a high input impedance buffer stage to prevent load effects on the preceding circuitry. It then enters the differential amplification stage, utilizing its high common-mode rejection ratio to effectively suppress common-mode interference in the environment. For example, a microvolt-level input signal is amplified to the millivolt level, while common-mode noise is significantly suppressed. The amplified and filtered raw electromyographic signal first passes through an anti-aliasing filter to prevent high-frequency component aliasing. It then enters the analog-to-digital conversion channel, where each sample point is converted into a corresponding digital value. The conversion resolution ensures that subtle changes in the electromyographic signal can be captured.

[0032] In one embodiment, the step of acquiring electromyographic feature data based on the digital electromyographic signals and muscle change data includes: S305. Obtain the electromyographic time-domain characteristics of the electrical activity during muscle contraction based on the digital electromyographic signal, and convert the digital electromyographic signal into electromyographic frequency-domain characteristics. S306. Align the digital electromyography signal and muscle change data according to the acquisition time, and obtain muscle change association data corresponding to the digital electromyography signal based on the aligned muscle change data. The muscle change association data includes the change in muscle contraction amplitude, the duration of muscle activation state, and the synchronicity of muscle electrical activity. S307. Simultaneously correlate the electromyographic time-domain features and electromyographic frequency-domain features with the change in muscle contraction amplitude, the duration of muscle activation state, and the synchronicity of muscle electrical activity to obtain an electromyographic-muscle association dataset of electromyographic features and muscle changes. S308. Obtain electromyographic feature data based on the electromyographic-muscle association dataset.

[0033] As described in steps S305-S308 above, the present invention extracts effective electromyographic feature data from digital electromyographic signals and muscle change data, and establishes a correspondence between electromyographic activity and muscle mechanical performance through multi-dimensional feature extraction and data correlation analysis.

[0034] Electromyography (EMG) time-domain characteristics refer to the characteristic parameters describing the intensity of muscle electrical activity over time. These can be achieved using root mean square (RMS) values ​​or integrated EMG values, reflecting changes in electrical activity intensity during muscle contraction. EMG frequency-domain characteristics refer to the characteristic parameters that convert the time-domain signal into a frequency-domain distribution using Fourier transform. These can be achieved using median frequency or average power frequency, characterizing the frequency characteristics of muscle activity. Time alignment refers to synchronizing physiological signals and motion data acquired by different acquisition devices along the time axis. This can be achieved using timestamp matching or signal interpolation methods, ensuring temporal consistency in data correlation analysis. Changes in muscle contraction amplitude refer to the changes in muscle volume or thickness during contraction. These can be measured using pressure sensor arrays or optical volumetric plethysmography, quantifying muscle contraction intensity. The duration of muscle activation refers to the time span from muscle activation to relaxation. This can be calculated by the duration for which the EMG signal amplitude exceeds a threshold, reflecting muscle endurance levels. Synchronicity of muscle electrical activity refers to the correlation between electrical signals from different muscle regions in the time or frequency domain. This can be achieved using cross-correlation analysis or coherence analysis, used to assess muscle coordination.

[0035] Time-domain and frequency-domain features were extracted from digital electromyography (EMG) signals. Time-domain feature extraction employed a sliding window calculation method (differentiated window sizes were used for different data sources: 256-512 millisecond windows were used for EMG signal analysis to capture complete firing patterns; 100-200 millisecond windows were used for muscle mechanics data to achieve rapid response; and 1-2 second windows were used for risk assessment to smooth transient fluctuations. All windows had a 50% overlap rate to ensure temporal continuity). The extracted time-domain features included: root mean square (RMS) value, reflecting the average power level of the signal; average amplitude, characterizing the signal intensity; and zero-crossing rate, a simple indicator describing the frequency components of the signal.

[0036] Frequency domain features were obtained using Fast Fourier Transform (FFT) with a sampling rate of 1000Hz, 1024 FFT points, and a Hanning window to reduce spectral leakage. The focus was on analyzing the effective frequency band of electromyography (EMG) from 20-500Hz, accurately extracting key features such as the median frequency and average power frequency with a frequency resolution of 0.98Hz. The power spectral density distribution of the signal was obtained. The median frequency and average power frequency were extracted from the power spectrum as key frequency domain features. The median frequency is the frequency point that divides the power spectrum into two equal areas, and the average power frequency is the weighted average frequency of the power spectrum.

[0037] Time alignment employs a strategy combining hardware-synchronized timestamps with software dynamic normalization. A timestamp accurate to the millisecond is added to each data point.

[0038] Based on the aligned data, three aspects of muscle change correlation data are extracted: the change in muscle contraction amplitude is calculated by measuring the change in muscle bulge height using an optical sensor; the duration of muscle activation is identified from the angle change curve output by the inertial measurement unit; and the synchronicity of muscle electrical activity is calculated by the cross-correlation coefficient between multi-electrode signals. Then, a quantitative correlation between electromyographic characteristics and muscle change data is established. A feature-level fusion strategy was employed, pairing temporal and frequency domain features with muscle change-related data for analysis. Linear regression analysis was performed on the temporal features of electromyography (EMG) and changes in muscle contraction amplitude (using a multiple linear model, specifically Y=β0+β1*X1+β2*X2+ε, where a quantitative relationship was established between EMG temporal features (e.g., RMS value), frequency domain features (e.g., median frequency), and muscle contraction mechanical parameters, with regression coefficients estimated using the least squares method), establishing a correspondence between electrical activity intensity and mechanical output. Trend correlation analysis was conducted on the temporal features of EMG and the duration of muscle activation, generating an EMG-muscle association dataset. Based on this dataset, feature selection and dimensionality reduction were performed to generate the final EMG feature data. Correlation analysis and importance assessment methods were used to select the subset of features most relevant to muscle functional state from the association dataset. Selection criteria included the correlation strength between features and the target variable, feature redundancy, and feature stability. The final generated EMG feature data was then presented.

[0039] In one embodiment, the step of aligning the digital electromyography (EMG) signal and muscle change data according to acquisition time, and obtaining muscle change-related data corresponding to the digital EMG signal based on the aligned muscle change data, includes: S3061. Obtain multiple time markers for the acquisition time of digital electromyography (EMG) signals, obtain the start and end times of a segment of digital EMG signals corresponding to each time marker, and obtain time markers with the same time interval from muscle change data based on the start and end times. S3062. Pair the digital electromyography signal and muscle change data according to the extracted time marker points to generate paired data sets of signal and muscle change data. S3063. Extract the dimensional information associated with the corresponding digital electromyography signal from the paired data sets corresponding to each time marker point from the aligned muscle change data, and obtain the morphological manifestation of muscle contraction in the muscle change data to obtain the change in muscle contraction amplitude. At the same time, obtain the duration of muscle activation and obtain the duration of muscle activation. Compare the activity performance of different parts of the muscle to obtain the spatial synchronicity of muscle electrical activity. S3064. The changes in muscle contraction amplitude, duration of muscle activation, and spatial synchronicity of muscle electrical activity extracted from each pair of data sets are bound to the corresponding digital electromyography signal segments according to the acquisition time sequence of the time marker points to generate muscle change association data.

[0040] As described in steps S3061-S3064 above, this embodiment solves the spatiotemporal registration problem between digital electromyography (EMG) signals and muscle change data, establishing a precise correspondence between EMG activity and muscle biomechanical performance. This invention extracts the start and end times of each analysis window of the digital EMG signal to form a reference time series. Based on these time points, corresponding time markers are located from the muscle change data stream. When there is a difference in sampling frequencies between the two types of data, signal resampling technology is used to achieve frequency matching, ensuring the synchronization of the time series.

[0041] The time stamp refers to a unified timestamp generated by the built-in synchronization clock module of the wearable device. This can be achieved using a high-precision crystal oscillator in conjunction with a time synchronization protocol, eliminating clock deviations between the electromyography (EMG) signal acquisition unit and the muscle deformation sensor. Paired data sets refer to matching EMG signal segments with corresponding time-segment muscle deformation data using a sliding time window algorithm. This can be achieved using a dynamic time warping algorithm for time axis calibration, addressing data misalignment caused by differences in sensor sampling rates. Muscle contraction morphology refers to the dynamic change curves of the muscle bulge region captured by a three-dimensional pressure sensor array. This can be achieved using a three-dimensional reconstruction algorithm of pressure distribution data, quantifying the spatial distribution characteristics of force output during muscle contraction. Spatial synchronicity of muscle electrical activity is obtained through covariance analysis of multi-channel EMG signals and pressure distribution data. This can be achieved by using a cross-correlation function to calculate the phase consistency of activity in different muscle regions, evaluating the efficiency of muscle group coordination.

[0042] After time alignment, digital electromyography (EMG) signals and muscle change data are segmented and paired according to time markers. Each pair of data sets contains EMG signal segments and muscle change data segments within the same time period, with data segments for the corresponding time period centered on each time marker. Discrete sampling sequences are directly extracted from the digital EMG signals, while multidimensional feature sequences, including movement posture and morphological changes, are extracted from the muscle change data. Each data set is accompanied by complete timestamp information to ensure temporal accuracy. Three key dimensions of muscle change correlation data are extracted from the aligned pair of data sets. The change in muscle contraction amplitude is calculated by analyzing the trajectory of changes in muscle bulge height, reflecting the mechanical output intensity of muscle contraction. The duration of muscle activation is detected from kinematic data, characterizing the time characteristics of maintaining the muscle contraction state. Spatial synchronicity of muscle electrical activity is obtained by analyzing the consistency of multi-electrode signals, assessing the coordination degree of muscle activity in different parts of the body. The extracted feature dimensions are structurally bound to the corresponding EMG signal segments to generate a complete muscle change correlation dataset.

[0043] In one embodiment, the step of simultaneously associating electromyographic time-domain features and electromyographic frequency-domain features with changes in muscle contraction amplitude, duration of muscle activation, and synchronicity of muscle electrical activity to obtain an electromyographic-muscle association dataset of electromyographic features and muscle changes includes: S7011. Correlate the temporal characteristics of electromyography with the change in muscle contraction amplitude, and record the corresponding change in muscle contraction amplitude when the intensity of electrical activity changes in the temporal characteristics of electromyography, so as to obtain temporal-muscle contraction amplitude change data. S7012. Correlate the electromyographic time-domain features with the duration of muscle activation, and record the trend of energy accumulation in the electromyographic time-domain features during the duration of muscle activation to obtain time-domain-duration data of muscle activation. S7013. Obtain the correspondence between the frequency distribution of electromyography frequency domain features and the synchronicity of muscle electrical activity, and obtain the degree of concentration of frequency distribution in the electromyography frequency domain features based on the synchronicity of muscle electrical activity to obtain frequency domain-muscle electrical activity synchronicity data. S7014. The electromyography-muscle association dataset is obtained by integrating the time-domain data on changes in muscle contraction amplitude, the time-domain data on duration of muscle activation, and the frequency-domain data on synchronicity of muscle electrical activity.

[0044] As described in steps S7011-S7014 above, this invention establishes a quantitative relationship between electromyographic temporal characteristics and changes in muscle contraction amplitude. The electromyographic temporal characteristics are derived from the temporal analysis of digital electromyographic signals, including signal intensity characteristics; the changes in muscle contraction amplitude are derived from quantitative data on muscle morphological changes.

[0045] A sliding window correlation analysis method was used, with a fixed duration window, to analyze the correspondence between the electromyographic (EMG) time-domain feature sequence and the sequence of changes in muscle contraction amplitude. When the EMG signal intensity characteristics changed, the response characteristics of the muscle contraction amplitude changes were recorded simultaneously. The variation patterns of EMG time-domain features during the duration of muscle activation were analyzed. The duration of muscle activation was derived from kinematic data analysis, and the EMG time-domain features were derived from continuous time-domain analysis.

[0046] Using the start and end times of muscle activation as boundaries, the electromyographic (EMG) temporal feature sequences within this time period are extracted. The trends in energy accumulation within the EMG temporal features during the duration of muscle activation are analyzed, and the overall trajectory of signal intensity characteristics during contraction is examined.

[0047] Trend analysis is employed to calculate the gradient and curvature of temporal features during activation, characterizing the evolution of these features. Simultaneously, the cumulative amount of temporal features is calculated to quantify the energy accumulation throughout the activation period.

[0048] A correspondence between electromyographic (EMG) frequency domain features and muscle electrical activity synchronicity was established. EMG frequency domain features were derived from spectral analysis results, including frequency distribution characteristics; muscle electrical activity synchronicity was derived from consistency analysis of multi-electrode signals. The correlation data from these three dimensions were integrated to construct a unified EMG-muscle correlation dataset. A feature matrix fusion method was employed to align the time-domain-mechanical correlation, time-domain-temporal correlation, and frequency-domain-synchronicity correlation data according to time series, forming a multi-dimensional feature matrix.

[0049] Feature optimization is performed during the integration process. Based on the assessment of feature importance and redundancy, the most representative related features are selected.

[0050] In one embodiment, the step of generating a real-time dynamic risk assessment value based on the electromyographic data and muscle contractile force data includes: S401. Obtain the frequency distribution change and electrical activity intensity change based on the electromyographic characteristic data; S402. Based on the frequency distribution change and muscle contraction force data, obtain multiple muscle contraction degrees, and based on the multiple muscle contraction degrees, obtain the smoothness of the movement completion. S403. Obtain multiple muscle exertion levels based on the changes in electrical activity intensity and muscle contraction force data, and obtain exertion stability based on multiple muscle exertion levels; S404. Generate a real-time dynamic risk assessment value based on the smoothness of the action and the stability of the force exertion; S405. Obtain the historical dynamic risk assessment set, and generate a risk assessment result based on the historical dynamic risk assessment set and the real-time dynamic risk assessment value.

[0051] As described in steps S401-S405 above, frequency distribution change refers to the dynamic adjustment of the energy distribution of electromyographic signals in different frequency bands over time. This can be achieved using Fast Fourier Transform or Wavelet Transform, and is used to characterize muscle coordination. Electrical activity intensity change refers to the fluctuation characteristics of the electromyographic signal amplitude over time. This can be achieved by calculating the root mean square value or integrated electromyographic value, and is used to reflect the degree of muscle activation. Based on the frequency distribution change and muscle contraction force data, multiple muscle contraction degrees are obtained (frequency distribution weighting coefficients and multiple frequency distribution change indices are obtained based on the frequency distribution change, and contraction force weighting coefficients and muscle contraction force for multiple time windows are obtained based on the muscle contraction force data). ;in, This represents the degree of muscle contraction in the i-th time window. Indicators representing changes in frequency distribution Represents the frequency distribution weighting coefficients. This represents the muscle contraction strength during the i-th time window. (Indicates the contraction force weighting coefficient), and the smoothness of motion completion (obtained) The corresponding sequence smoothness coefficient and smoothness correction strength parameter, ,in, Indicates the smoothness of the action. The standard deviation of the contraction series. Indicates the number of windows. Sequence representing the degree of muscle contraction The maximum value in, Sequence representing the degree of muscle contraction The minimum value in, Indicates the smoothness correction strength parameter. The smoothness coefficient (representing the sequence smoothness) is an indicator of the consistency of mechanical output at each stage of muscle contraction, used to assess the stability of motor control. Force stability is a quantitative indicator of the degree of fluctuation in muscle contraction force output, specifically achieved by statistically analyzing the standard deviation of the force output sequence, used to determine muscle fatigue status.

[0052] Based on the synergistic analysis of frequency distribution changes and muscle contraction force data, the smoothness of movement execution is quantified. First, the various stages of muscle contraction, including initiation, duration, and relaxation, are identified according to frequency distribution characteristics. The correspondence between muscle contraction force and frequency distribution characteristics is analyzed to assess the coordination between the two. The smoothness characteristics of the contraction force trajectory are calculated using time series analysis to evaluate the coherence of stage transitions. Finally, by combining frequency distribution change patterns, the coordination ability of neuromuscular control is comprehensively analyzed to generate a movement completion smoothness score.

[0053] Multiple muscle exertion levels are obtained by analyzing changes in electrical activity intensity and muscle contraction force data. The formula for calculating muscle exertion level is as follows: ;in, This represents the degree of muscle exertion in the i-th time window. This represents the intensity of electrical activity in the i-th time window. This represents the weighting coefficient for the intensity of electrical activity. This represents the muscle contraction strength during the i-th time window. The formula for calculating force stability, representing the muscle contraction force weighting coefficient, is as follows: , Indicates the stability of force exertion. The coefficient of variation represents the degree of force exerted.

[0054] This study analyzes the variation patterns of force characteristics during continuous contraction cycles, assessing the fluctuation characteristics of peak force levels and the repeatability of force trajectories. Statistical analysis methods are used to identify the stability of force patterns, while also examining the temporal consistency characteristics of electrical activity intensity. By integrating the evaluation results from multiple dimensions, a force stability index is generated.

[0055] The evaluation results, considering both the smoothness of the movement and the stability of force application, are used to generate a real-time dynamic risk assessment value using a weighted summation algorithm. The weight coefficients of each indicator are determined based on the correlation strength between features and risk levels, and normalization is applied to ensure the consistency of the evaluation results.

[0056] Time series smoothing is introduced to comprehensively consider assessment results from multiple recent time points, improving the stability of risk assessment. Real-time dynamic risk assessment values ​​are updated at a fixed frequency to ensure timely reflection of current status changes. A confidence index for the assessment results is also provided to measure the reliability of the current assessment.

[0057] In one embodiment, the step of generating a risk assessment result based on the historical dynamic risk assessment set and the real-time dynamic risk assessment value includes: S501. Obtain multiple historical dynamic risk assessment values ​​based on the historical dynamic risk assessment set; S502. Generate movement change trends based on multiple historical dynamic risk assessment values ​​and real-time dynamic risk assessment values; S503. Generate a trend difference value based on the real-time dynamic risk assessment value and the historical dynamic risk assessment value with the closest recording time. S504. If the trend is upward, it is determined that the user is at risk of exercise fatigue. A risk assessment result for warning is generated based on the trend difference value. S505. If the trend is downward or stable, it is determined that the user does not have the risk of exercise fatigue, and the real-time dynamic risk assessment value is incorporated into the historical dynamic risk assessment set as the latest historical dynamic risk assessment value.

[0058] As described in steps S501-S505 above, this invention extracts valid assessment records from a historical dynamic risk assessment set. Historical data uses a circular buffer storage structure to retain the most recent complete assessment sequence. During data extraction, data validity is first verified to exclude abnormal records and invalid data. Assessment values ​​within consecutive time periods are selected according to chronological order to ensure the timeliness and continuity of the data. Simultaneously, considering the matching degree of movement types, historical data under the same movement pattern are preferentially selected as a reference benchmark.

[0059] Based on historical evaluation sequences and real-time evaluation values, time-series analysis is employed to identify trends in motion state changes. Historical data undergoes smoothing preprocessing to eliminate random fluctuations. A linear regression algorithm is used to calculate the trend line of evaluation values ​​over time, obtaining the trend direction and intensity. Simultaneously, a sliding window technique is used to analyze short-term change characteristics and identify trend persistence. The reliability of the analysis results is assessed by calculating trend confidence levels.

[0060] The trend difference is calculated by comparing the real-time assessment value with the most recent historical benchmark value. The most recent and reliable historical data is selected as the comparison benchmark. The difference is calculated using the relative rate of change method, comprehensively considering both absolute difference and relative magnitude of change. An individualized fluctuation benchmark model is established, and the sensitivity of the difference judgment is adjusted according to historical fluctuation characteristics. Statistical significance tests are used to distinguish between normal fluctuations and abnormal changes.

[0061] When a sustained upward trend is detected, risk levels are categorized based on trend strength and magnitude of the difference. The severity of the risk is assessed by considering both the trend's duration and magnitude of change. Corresponding warnings are generated for each risk level, including a risk description and response recommendations. Warning content considers the user's movement characteristics and historical responses, providing personalized guidance.

[0062] When the trend stabilizes or improves, real-time evaluation values ​​are incorporated into the historical dataset. Quality checks are performed before adding new data to ensure its representativeness. The dataset is regularly optimized, and statistical baseline parameters are updated to maintain the accuracy of the evaluation benchmark.

[0063] like Figure 2 As shown, the present invention also provides a dynamic risk assessment system based on wearable devices, comprising: Raw electromyography signal acquisition module 1 is used to acquire raw electromyography signals generated by the user's limb muscles during exercise; The muscle contraction force data acquisition module 2 is used to acquire muscle change data of the user's limbs during movement, and to acquire muscle contraction force data based on the muscle change data. Electromyographic feature data acquisition module 3 is used to preprocess the original electromyographic signal to obtain a digital electromyographic signal, and to acquire electromyographic feature data based on the digital electromyographic signal and muscle change data. The real-time dynamic risk assessment value acquisition module 4 is used to generate a real-time dynamic risk assessment value based on the electromyographic characteristic data and muscle contraction force data.

[0064] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in this application and in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-speed SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0065] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0066] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent results or equivalent process transformations made based on the content of the present invention specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A dynamic risk assessment method based on wearable devices, applied to wearable monitoring devices, wherein the wearable monitoring devices are worn on a user's moving limbs, characterized in that, include: Acquire raw electromyographic signals generated by the user's limb muscles during exercise; Acquire muscle change data of the user's limbs during movement, and obtain muscle contraction force data based on the muscle change data; The raw electromyography (EMG) signal is preprocessed to obtain a digital EMG signal, and EMG feature data is obtained based on the digital EMG signal and muscle change data. A real-time dynamic risk assessment value is generated based on the electromyographic characteristic data and muscle contraction force data.

2. The dynamic risk assessment method based on wearable devices according to claim 1, characterized in that, The steps for acquiring raw electromyographic signals generated by the user's limb muscles during exercise include: Obtain the raised area when the user's limb muscles contract during exercise; Multiple muscle pressure feedback values ​​of the muscle bulge area to the wearable monitoring device are obtained, and the location of the maximum muscle pressure feedback value is used as the center point of muscle monitoring. Multiple muscle-skin electrical signals within a preset distance are obtained based on the muscle monitoring center point, and the largest and smallest muscle-skin electrical signals are obtained. The raw electromyographic (EMG) signals are obtained based on the largest and smallest muscle-skin EMG signals.

3. The dynamic risk assessment method based on wearable devices according to claim 1, characterized in that, The step of preprocessing the raw electromyography (EMG) signal to obtain a digital EMG signal includes: The low-frequency baseline drift interference in the original electromyography signal is filtered out by a high-pass filter circuit to obtain the first filtered original electromyography signal; The second filtered original electromyography signal is obtained by filtering out the power frequency interference signal generated by the wearable monitoring device in the first filtered original electromyography signal through a low-pass filter circuit. The second filtered raw electromyographic signal is amplified to obtain the amplified filtered raw electromyographic signal. The original electromyographic signal, after being amplified and filtered, is converted into a discrete digital signal to obtain a digital electromyographic signal.

4. The dynamic risk assessment method based on wearable devices according to claim 1, characterized in that, The steps for obtaining electromyographic feature data based on the digital electromyographic signals and muscle change data include: Based on the digital electromyography (EMG) signal, the EMG time-domain characteristics of the electrical activity during muscle contraction are obtained, and the digital EMG signal is converted into EMG frequency-domain characteristics. The digital electromyography (EMG) signals and muscle change data are aligned according to the acquisition time, and muscle change correlation data corresponding to the digital EMG signals are obtained based on the aligned muscle change data. The muscle change correlation data includes the change in muscle contraction amplitude, the duration of muscle activation state, and the synchronicity of muscle electrical activity. By simultaneously associating electromyographic time-domain features and electromyographic frequency-domain features with changes in muscle contraction amplitude, duration of muscle activation, and synchronicity of muscle electrical activity, an electromyographic-muscle association dataset of electromyographic features and muscle changes is obtained. Electromyographic feature data were obtained from the electromyographic-muscle association dataset.

5. The dynamic risk assessment method based on wearable devices according to claim 4, characterized in that, The steps of aligning the digital electromyography (EMG) signals and muscle change data according to acquisition time, and obtaining muscle change correlation data corresponding to the digital EMG signals based on the aligned muscle change data, include: Multiple time markers for the acquisition time of digital electromyography (EMG) signals are obtained. The start and end times of a segment of digital EMG signal corresponding to each time marker are obtained. Time markers with the same time interval are obtained from muscle change data based on the start and end times. The digital electromyography signal and muscle change data are paired according to the extracted time marker points to generate paired data sets of signal and muscle change data. For each time marker, the paired data sets are extracted from the aligned muscle change data to obtain dimensional information associated with the corresponding digital electromyography (EMG) signal. The morphological manifestation of muscle contraction in the muscle change data is obtained to obtain the change in muscle contraction amplitude. At the same time, the duration of muscle activation is obtained to obtain the duration of muscle activation. The activity performance of different parts of the muscle is compared to obtain the spatial synchronicity of muscle electrical activity. The changes in muscle contraction amplitude, duration of muscle activation, and spatial synchronicity of muscle electrical activity extracted from each pair of data sets are bound to the corresponding digital electromyography signal segments according to the acquisition time sequence of the time marker points to generate muscle change correlation data.

6. The dynamic risk assessment method based on wearable devices according to claim 4, characterized in that, The steps for simultaneously associating electromyographic (EMG) time-domain and frequency-domain features with changes in muscle contraction amplitude, duration of muscle activation, and synchronicity of muscle electrical activity to obtain an EMG-muscle association dataset of EMG features and muscle changes include: The temporal characteristics of electromyography (EMG) are correlated with the changes in muscle contraction amplitude. When the intensity of electrical activity changes in the temporal characteristics of EMG, the corresponding changes in muscle contraction amplitude are recorded to obtain temporal-muscle contraction amplitude change data. The electromyography time-domain features are correlated with the duration of muscle activation, and the change trend of energy accumulation in the electromyography time-domain features is recorded during the duration of muscle activation to obtain time-domain muscle activation duration data. The correspondence between the frequency distribution of electromyography (EMG) frequency domain features and the synchronicity of muscle electrical activity is obtained, and the degree of concentration of the frequency distribution in the EMG frequency domain features is obtained based on the synchronicity of muscle electrical activity, thus obtaining frequency domain-muscle electrical activity synchronicity data. The electromyography-muscle association dataset is obtained by integrating time-domain data on changes in muscle contraction amplitude, time-domain data on duration of muscle activation, and frequency-domain data on synchronicity of muscle electrical activity.

7. The dynamic risk assessment method based on wearable devices according to claim 1, characterized in that, The steps for generating real-time dynamic risk assessment values ​​based on the electromyographic characteristic data and muscle contraction force data include: Based on the electromyographic characteristic data, the frequency distribution changes and electrical activity intensity changes are obtained; Based on the frequency distribution changes and muscle contraction force data, multiple muscle contraction degrees are obtained, and the smoothness of the movement is obtained based on the multiple muscle contraction degrees. Based on the changes in electrical activity intensity and muscle contraction force data, multiple muscle exertion levels are obtained, and exertion stability is obtained based on multiple muscle exertion levels; A real-time dynamic risk assessment value is generated based on the smoothness of the action and the stability of the force exertion. Obtain a historical dynamic risk assessment set, and generate a risk assessment result based on the historical dynamic risk assessment set and the real-time dynamic risk assessment value.

8. The dynamic risk assessment method based on wearable devices according to claim 7, characterized in that, The steps for generating risk assessment results based on the historical dynamic risk assessment set and real-time dynamic risk assessment values ​​include: Multiple historical dynamic risk assessment values ​​are obtained based on the aforementioned historical dynamic risk assessment set; The movement change trend is generated based on multiple historical dynamic risk assessment values ​​and real-time dynamic risk assessment values; A trend difference value is generated based on the real-time dynamic risk assessment value and the historical dynamic risk assessment value with the closest recording time. If the trend is upward, it is determined that the user is at risk of exercise fatigue, and a risk assessment result for warning is generated based on the trend difference value. If the trend is downward or stable, it is determined that the user is not at risk of exercise fatigue, and the real-time dynamic risk assessment value is incorporated into the historical dynamic risk assessment set as the latest historical dynamic risk assessment value.

9. A dynamic risk assessment system based on wearable devices, characterized in that, include: The raw electromyography (EMG) signal acquisition module is used to acquire the raw EMG signals generated by the user's limb muscles during exercise. The muscle contraction force data acquisition module is used to acquire muscle change data of the user's limbs during movement, and to acquire muscle contraction force data based on the muscle change data. The electromyography feature data acquisition module is used to preprocess the raw electromyography signal to obtain a digital electromyography signal, and to acquire electromyography feature data based on the digital electromyography signal and muscle change data. The real-time dynamic risk assessment value acquisition module is used to generate a real-time dynamic risk assessment value based on the electromyographic characteristic data and muscle contraction force data.

10. A dynamic risk assessment device based on wearable devices, characterized in that, It includes an electromyography (EMG) acquisition unit and a muscle change acquisition unit, and implements the dynamic risk assessment method based on a wearable device as described in any one of claims 1-8.

Images