Athlete muscle strength training and rehabilitation system based on electrical stimulation

Abstract

The application provides an athlete muscle strength training and rehabilitation system based on electric stimulation, and relates to the technical field of medical rehabilitation.The application synchronously collects electromyography and kinematic signals, extracts muscle coordination index and joint net output torque double-dimension features in real time, identifies three movement performance states of coordination dominant type, strength dominant type and mixed type, and dynamically generates differentiated electric stimulation parameter sets according to the three movement performance states, so that the precise typing and targeted intervention of the fatigue state of athletes are realized, the muscle coordination efficiency and strength output are synchronously optimized in training, compensatory injury is effectively prevented and functional recovery is accelerated in rehabilitation, and the individualization, safety and overall efficiency of training and rehabilitation are improved.

Description

Technical Field

[0001] This invention relates to the field of medical rehabilitation technology, specifically to an athlete muscle strength training and rehabilitation system based on electrical stimulation. Background Technology

[0002] Neuromuscular electrical stimulation (NMES), as a non-invasive bioelectrical intervention, has been widely used in the fields of muscle strength training and postoperative rehabilitation for athletes. Its basic principle is to apply low-frequency pulsed currents through electrodes on the body surface to stimulate motor nerves or muscle fibers, inducing controllable muscle contraction, thereby enhancing muscle strength, delaying atrophy, or promoting functional reconstruction. Current applications mostly rely on preset fixed programs or simple feedback based on a single physiological signal (such as the electromyographic amplitude of a particular muscle) to adjust stimulation parameters.

[0003] However, during dynamic, high-intensity specialized sports training or complex functional rehabilitation, athlete fatigue or performance decline is not a one-dimensional phenomenon. The underlying physiological mechanisms can be categorized into at least two types: first, neuromuscular coordination degeneration, manifested as disordered activation sequence and dyssynergistic effects between agonist and antagonist muscles; second, decreased muscle strength generation capacity, manifested as a gap between central driving intention and actual joint torque output. Current electrical stimulation techniques lack the ability to differentiate between these two mechanisms in real-time and synchronously, limiting intervention strategies to general approaches such as reducing stimulation intensity during fatigue. This results in a failure to address the root cause; when coordination patterns need correction, insufficient stimulation may render the intervention ineffective; and when strength needs to be supplemented, inappropriate stimulation patterns may reinforce erroneous compensatory movements, even increasing the risk of injury, thus hindering further improvements in training efficiency and rehabilitation safety. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides an athlete muscle strength training and rehabilitation system based on electrical stimulation.

[0005] The technical solution adopted in this invention is as follows:

[0006] An electrical stimulation-based muscle strength training and rehabilitation system for athletes, comprising:

[0007] The data acquisition and preprocessing module is used to simultaneously acquire the raw electromyographic signals of at least one pair of agonist muscles and their main antagonist muscles, as well as the kinematic data of the target joint, when the athlete performs the target training movement; process the raw electromyographic signals to obtain the standardized electromyographic envelopes of the agonist muscles and the main antagonist muscles respectively; process the kinematic data to obtain a clean kinematic time series; wherein the kinematic data includes at least joint angles and angular velocities;

[0008] A dual-dimensional feature extraction module is used to calculate a muscle coordination index to quantify the synergistic working state of the agonist muscle and the main antagonist muscle based on the standardized electromyographic envelope of the agonist muscle; at the same time, based on the clean kinematic time series and the amplitude of the standardized electromyographic envelope of the agonist muscle, the net output torque of the target joint during the execution of the target training action is estimated in real time.

[0009] The state recognition and classification module is used to compare the calculated muscle coordination index with a preset coordination baseline threshold, and determine whether coordination degradation has occurred based on the comparison result; at the same time, it compares the estimated net output torque with a preset target torque curve corresponding to the target training movement, and determines whether strength decay has occurred based on the comparison result; combining the determination results of coordination degradation and strength decay, the current sports performance is identified as one of the following three categories: coordination-dominated, strength-dominated, or mixed.

[0010] The intelligent decision-making and execution module is used to invoke different stimulation parameter decision logics according to the identified category: if it is identified as the coordination-dominant type, a first type of electrical stimulation parameter set with adjusting the muscle activation timing as the core is generated; if it is identified as the force-dominant type, a second type of electrical stimulation parameter set with compensating for muscle contraction intensity as the core is generated; if it is identified as the hybrid type, a third type of electrical stimulation parameter set integrating timing adjustment and intensity compensation is generated; finally, based on the generated electrical stimulation parameter set, electrical stimulation is applied to the corresponding target muscle group.

[0011] Furthermore, in the dual-dimensional feature extraction module, calculating the muscle coordination index specifically involves: calculating the peak value of the cross-correlation function of the standardized electromyographic envelopes of the agonist muscle and the main antagonist muscle during the key phase of the movement, or calculating the coherence coefficient of the two in a specific frequency band, or calculating the ratio of the electromyographic integral area of ​​the main antagonist muscle to that of the agonist muscle.

[0012] Furthermore, in the dual-dimensional feature extraction module, the real-time estimation of the net output torque of the target joint specifically involves inputting the clean kinematic time series and the standardized electromyographic envelope amplitude of the agonist muscle into a pre-calibrated simplified human musculoskeletal biomechanics model to calculate the net output torque in real time.

[0013] Furthermore, in the state recognition and classification module, determining whether force attenuation has occurred specifically involves: calculating the difference between the time series of the net output torque and the target torque curve to obtain the real-time torque gap; if the integral area or peak value of the real-time torque gap exceeds the preset force gap threshold, then force attenuation is determined to have occurred.

[0014] Furthermore, in the intelligent decision-making and execution module, the intensity compensation logic on which the second type of electrical stimulation parameter set is generated is specifically as follows: the instantaneous value or integral value of the real-time torque gap is dynamically converted into additional electrical stimulation intensity parameters for the agonist muscle through a predefined mapping function.

[0015] Furthermore, the timing adjustment logic for generating the first type of electrical stimulation parameter set is as follows: based on the time variation characteristics of the muscle coordination index, the sub-phase with the worst muscle coordination within the action cycle is identified, and the optimal timing parameters for applying corrective electrical stimulation to the main antagonist muscle or the agonist muscle are determined.

[0016] Furthermore, in the intelligent decision-making and execution module, the generation of the third type of electrical stimulation parameter set specifically involves: assigning dynamic weights to the temporal parameters in the first type of electrical stimulation parameter set and the intensity parameters in the second type of electrical stimulation parameter set based on the relative proportions of the degree of coordination degradation and the degree of strength attenuation determined in the dual-dimensional feature extraction module, and then performing weighted fusion.

[0017] Furthermore, in the intelligent decision-making and execution module, the timing for applying electrical stimulation is the moment within the next or current action cycle that corresponds to the decision sequence determined by the set of electrical stimulation parameters.

[0018] The beneficial effects of this invention are:

[0019] This invention simultaneously acquires electromyographic and kinematic signals, extracts the dual-dimensional features of muscle coordination index and joint net output torque in real time, identifies three types of motor performance states: coordination-dominated, strength-dominated, and mixed, and dynamically generates differentiated electrical stimulation parameter sets accordingly. This enables precise classification and targeted intervention of athletes' fatigue states, thereby simultaneously optimizing muscle synergy efficiency and force output during training, effectively preventing compensatory injuries and accelerating functional recovery during rehabilitation, and improving the personalization, safety, and overall effectiveness of training and rehabilitation. Attached Figure Description

[0020] Figure 1 This is a block diagram of an electrical stimulation-based athlete muscle strength training and rehabilitation system according to an embodiment of the present invention.

[0021] Figure 2 This is a flowchart of an embodiment of the present invention for an athlete's muscle strength training and rehabilitation based on electrical stimulation;

[0022] Figure 3 This is a flowchart of a two-dimensional feature extraction and state recognition method according to an embodiment of the present invention. Detailed Implementation

[0023] 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.

[0024] like Figures 1-3 As shown in the figure, an embodiment of the present invention provides an athlete muscle strength training and rehabilitation system based on electrical stimulation, which includes a data acquisition and preprocessing module, a two-dimensional feature extraction module, a state recognition and classification module, and an intelligent decision-making and execution module.

[0025] The data acquisition and preprocessing module is used to synchronously acquire the raw electromyographic signals of at least one pair of agonist muscles and their main antagonist muscles, as well as the kinematic data of the target joint, when the athlete performs the target training movement; process the raw electromyographic signals to obtain the standardized electromyographic envelopes of the agonist muscles and the main antagonist muscles; process the kinematic data to obtain a clean kinematic time series; wherein the kinematic data includes at least joint angles and angular velocities.

[0026] This module acquires muscle electrophysiological signals and joint kinematic signals when athletes perform target movements through high-precision synchronous acquisition. It also eliminates interference from individual differences and environmental noise through standardized preprocessing, and outputs standardized electromyographic envelopes and clean kinematic time series that can be directly used for subsequent feature calculations.

[0027] It should be noted that the core of muscle synergy is the coordination between the contraction of the agonist muscle and the relaxation / moderate contraction of the antagonist muscle. The comparison of their electrical signals can directly reflect the synergistic state. Therefore, the electromyography (EMG) signal acquisition targets at least one pair of agonist muscles and a major antagonist muscle. For example, during knee flexion and extension, the agonist muscle is the quadriceps femoris, and the major antagonist muscle is the hamstring. During elbow flexion and extension, the agonist muscle is the biceps brachii, and the major antagonist muscle is the triceps brachii. The kinematic data acquisition targets the joint angle and angular velocity of the target joint, such as the knee joint angle and knee joint angular velocity during knee flexion and extension.

[0028] To ensure the accuracy of the acquired signals, it is necessary to select appropriate acquisition equipment and set reasonable parameters. For electromyography (EMG) signal acquisition, a multi-channel surface electromyography (SEMG) sensor array can be used, with at least two channels corresponding to a pair of agonist-antagonist muscles. If multiple muscle pairs need to be acquired, the array can be expanded to 4-8 channels. Sampling rate... The frequency can be set to 1000-2000Hz, satisfying the sampling theorem for effective electromyography (EMG) signals of 20-500Hz, balancing accuracy and data redundancy control; the resolution is ≥12 bits, ensuring the quantization accuracy of weak EMG signals and avoiding signal distortion; the common-mode rejection ratio (CMRR) is ≥80dB, effectively suppressing common-mode signals such as power frequency interference and electrode contact noise, thus improving the signal-to-noise ratio. In a specific embodiment of this invention, during the acquisition operation, an Ag / AgCl surface electrode can be attached to the midpoint of the muscle belly (where the signal is strongest), and the reference electrode can be attached to the patella during acquisition of adjacent bones, such as the knee joint; before attachment, the skin is wiped with alcohol to remove grease, ensuring that the contact impedance between the electrode and the skin is ≤5kΩ, avoiding contact noise.

[0029] Kinematic data acquisition can utilize 6-axis or 9-axis inertial measurement units (IMUs). A 6-axis IMU includes a 3-axis gyroscope and a 3-axis accelerometer, while a 9-axis IMU adds a 3-axis magnetometer to improve static angle accuracy. The sampling rate can be set to 200-500Hz, covering the frequency range of 0.5-10Hz for joint movements, matching the electromyography (EMG) sampling rate. Angle accuracy must be ≤0.5° statically and ≤2° dynamically, with an angular velocity range ≥±2000° / s to meet the needs of athletes' rapid movements. During data acquisition, the IMU can be fixed to both ends of the target joint using elastic straps. For example, when acquiring data for the knee joint, it can be fixed to the lower thigh and upper calf. Static calibration is required before acquisition to define the joint's neutral position, such as 0° when the knee is extended and positive when flexed, ensuring comparability of angle data from different athletes. Simultaneously, hardware triggering or software timestamps are used to synchronize the two types of signals.

[0030] raw electromyographic signals Containing numerous interferences such as 50Hz power frequency, electrode noise, and motion artifacts, this invention eliminates interference and extracts effective features by preprocessing the original electromyography (EMG) signal. Specifically, it employs a four-step process: noise reduction filtering, full-wave rectification, low-pass filtering, and standardization. The first step, noise reduction filtering, targets power frequency interference and invalid frequency band noise in the original EMG signal using a combination of 50Hz notch filtering and 20-500Hz bandpass filtering. The notch filter can be a second-order infinite impulse response (IIR) notch filter with the following transfer function:

[0031] ;

[0032] In the formula, For Laplace variables; The center angular frequency of the notch filter. (Power frequency); The quality factor, with a value of 5-10, is used to determine the quality. The larger the Q value, the narrower the notch bandwidth and the stronger the suppression of 50Hz signals, while avoiding filtering out effective electromyographic signals at adjacent frequencies.

[0033] The final output is the electromyographic signal after removing power frequency interference, denoted as... .

[0034] A fourth-order Butterworth bandpass filter can be used for bandpass filtering, with the following transfer function:

[0035] ;

[0036] In the formula, This is the low-frequency cutoff angular frequency. ; It is the high-frequency cutoff angular frequency. ; This is the damping factor, with a value of 0.707 (a typical value for Butterworth filters, ensuring a flat passband).

[0037] The final output is a clean electromyographic signal after filtering out high and low frequency noise, denoted as .

[0038] The second step is full-wave rectification. Since electromyographic signals are bidirectional AC signals, and muscle activation intensity is only related to signal amplitude and not polarity, the bidirectional signal is converted into a unidirectional signal by taking the absolute value, which facilitates the subsequent extraction of amplitude change trends. Specifically, full-wave rectification can be used (it preserves signal amplitude information more completely than half-wave rectification), and the formula is as follows:

[0039] ;

[0040] In the formula, This indicates the absolute value operation.

[0041] The final output is a unidirectional rectified electromyographic signal, denoted as... The signal is pulsed, reflecting the activation pulse of a single motion unit.

[0042] The third step is low-pass filtering to extract the electromyographic envelope. The rectified signal still contains high-frequency activation pulses of individual motor units, which need to be smoothed to reflect the time-varying intensity of the overall muscle activation. A first-order RC low-pass filter can be used, with the following formula:

[0043] ;

[0044] In the formula, The electromyographic envelope value at time t; for Electromyographic envelope values ​​at any given time (historical values). These are the filter coefficients. , The filtering time constant ( ), (cutoff frequency); The sampling time interval of the electromyographic signal ( ,like hour, ).

[0045] The final output is the smoothed electromyographic envelope, denoted as . The signal exhibits a "slow fluctuation" pattern, directly reflecting the temporal changes in the overall muscle activation intensity.

[0046] The fourth step is standardization. Because the electromyographic amplitude varies greatly among athletes—for example, the maximum electromyographic amplitude of an 80kg athlete and a 50kg athlete may differ by 2-3 times—directly using the original envelope makes horizontal comparison or quantification of synergistic states impossible. Therefore, this invention uses the peak value of the electromyographic envelope at maximal voluntary contraction (MVC) as the benchmark. MVC is obtained by having the athlete apply maximum force to the target muscle in a static state. For example, during knee MVC, the athlete sits and extends their knee against a fixed resistance to reach maximum force, and the electromyographic envelope at this point is collected; the peak value is taken as the benchmark. The mathematical expression for the standardized benchmark is:

[0047] ;

[0048] In the formula, The standardized electromyographic envelope at time t, with values ​​ranging from 0 to 1; The peak value of the electromyographic envelope during the maximum voluntary contraction (MVC) of the target muscle.

[0049] The final output is the standardized electromyographic envelope of the agonist muscle. and standardized electromyographic envelopes of the main antagonist muscles .

[0050] Raw kinematic data (raw joint angles) Original joint angular velocity The data includes sensor jitter noise, such as slight IMU tremors, which needs to be removed from outliers and smoothed with a sliding window to obtain a clean kinematic time series.

[0051] Specifically, the joint angles are first calculated. If the IMU directly outputs the limb posture angles (such as the pitch angle between the thigh and the lower leg), the original joint angles are obtained by taking the absolute difference between the posture angles of the two ends of the limb. The calculation formula is:

[0052] ;

[0053] , The posture angles of the limbs at both ends of the target joint, for example The thigh pitch angle, The lower leg's flexion-extension angle; for example, when the knee is extended. 0° indicates straightening, and >0° indicates bending.

[0054] If the IMU only outputs angular velocity, then the angle is calculated by integration:

[0055] ;

[0056] In the formula, This refers to the initial joint angle, such as the angle at the start of the movement.

[0057] The second step is outlier removal. For instantaneous extreme values ​​caused by IMU collisions and jitter (such as a sudden jump in angular velocity to 1000° / s), this invention can use the 3σ criterion (under a normal distribution, 99.7% of the data fall within the normal range). If the value exceeds the limit, it is considered an outlier and is processed accordingly. Specifically, the original data (if any) is first calculated. or mean with standard deviation If the data at a certain moment satisfy or Then replace it with the mean of two adjacent time points:

[0058] ;

[0059] The third step is sliding window smoothing filtering. After outlier removal, the data still has high-frequency jitter, such as IMU's own electronic noise, with a frequency of approximately 50-100Hz. This invention uses a simple sliding window averaging method, with the following formula:

[0060] ;

[0061] ;

[0062] In the formula, Let be the clean joint angle at time t; t represents the clean joint angular velocity at time t; N is the sliding window length (values ​​range from 5 to 10; for example, when the IMU sampling rate is 500Hz, N=5 corresponds to a window duration of 10ms, which can smooth high-frequency jitter and prevent lag in action). The sampling time interval of the IMU ( ); This represents the i-th historical moment within the window.

[0063] The final output is a clean joint angle time series. and clean joint angular velocity time series .

[0064] The dual-dimensional feature extraction module is used to calculate the muscle coordination index, which quantifies the synergistic working state of the agonist and the main antagonist muscles, based on the standardized electromyographic envelopes of the agonist muscles. At the same time, based on the clean kinematic time series and the amplitude of the standardized electromyographic envelope of the agonist muscles, the net output torque of the target joint during the execution of the target training action is estimated in real time.

[0065] In one embodiment of the present invention, the muscle coordination index quantifies the degree of coordination between the agonist and antagonist muscles through the spatiotemporal characteristics of electromyographic signals. The calculation of the muscle coordination index specifically involves: calculating the peak value of the cross-correlation function of the standardized electromyographic envelopes of the agonist and the main antagonist muscles in the key phase of the movement, or calculating the coherence coefficient of the two in a specific frequency band, or calculating the ratio of the electromyographic integral area of ​​the main antagonist muscle to that of the agonist muscle.

[0066] In this embodiment of the invention, the first method is a calculation method based on the peak value of the cross-correlation function. This method is mainly for evaluating temporal synergy and is applicable to fast-moving action scenarios such as sprinting and jumping. Specifically, it first divides the key temporal phases of the action. (Not the entire movement cycle, but only the most critical phase of muscle synergy), for example, in knee flexion and extension, the key phase is the push-off phase. =The knee joint begins to extend. =Knee extension to maximum angle; elbow curl, the key phase is the upward phase. =The elbow joint begins to flex. =Elbow flexed to 90°. It should be noted that the key phase division is based on the extreme points of the joint angular velocity, such as... The moment when it changes from negative to positive is , The moment when it falls back to 0 after reaching its peak is .

[0067] Subsequently at the critical phase Within this context, calculate the cross-correlation function of the standardized electromyographic envelopes of the agonist and antagonist muscles. The formula is:

[0068] ;

[0069] In the formula, The time delay (in seconds) indicates the lag of the antagonist muscle signal. After a period of time, similarity to agonist muscle signals was calculated; The standardized activation intensity of the agonist muscle at time t; for Standardized activation intensity (hysteresis) of antagonistic muscles at any given time (A signal of time). The larger the value, the greater the delay. The more similar the activation patterns of agonist and antagonist muscles are, the better.

[0070] Finally, the maximum value of the cross-correlation function within a reasonable delay range is taken as the muscle coordination index. The formula is:

[0071] ;

[0072] In the formula, The maximum delay threshold (unit: seconds) is set to 10%-20% of the duration of the critical phase of the action. For example, if the critical phase is 0.5 seconds, then... ;like If positive, the antagonist muscle activation lags behind the agonist muscle (normal synergy, such as the quadriceps muscle activating first during extension); if A negative value indicates that the antagonist muscle is activated before the agonist muscle (abnormal synergy, which can easily lead to force cancellation).

[0073] It should be noted that, The value range is 0-1 (after standardization of the cross-correlation function), for example, This indicates excellent coordination, time synchronization, and activation pattern matching; This indicates good synergy; This indicates a decline in synergy, requiring time-adjusted electrical stimulation.

[0074] The second method for calculating the muscle coordination index is based on the coherence coefficient of a specific frequency band. Since the frequencies below 80Hz are mainly low-frequency noise caused by muscle fatigue, and the frequencies above 30Hz are mainly random activation signals of individual motor units, this invention selects 8-30Hz as the analysis frequency band. ( , This can reflect the synchronous activation of multiple motor units. Specifically, the time-domain electromyography (EMG) signal is first converted into a frequency-domain signal using Fast Fourier Transform (FFT), and the autopower spectral density of the agonist muscle EMG signal is calculated. (Reflecting the power distribution of the agonist muscle signal at frequency f), the autopower spectral density of the antagonist muscle electromyographic signal. (Reflecting the power distribution of the antagonist muscle signal at frequency f), cross-power spectral density of the agonist and antagonist muscles. (Reflecting the cross-power distribution between the two at frequency f); taking the agonist muscle as an example, the formula for calculating the self-power spectral density is:

[0075] ;

[0076] In the formula, T is the signal duration (taking the key time phase). (duration) The Hanning Window is used to reduce spectral leakage in the FFT. This is the Fast Fourier Transform operation.

[0077] The formula for calculating the cross power spectral density is:

[0078] ;

[0079] In the formula, It indicates complex conjugate.

[0080] Then in frequency band Within, calculate the coherence coefficient for each frequency f. The formula is:

[0081] ;

[0082] Finally, the average coherence coefficient within the frequency band is taken as the muscle coordination index. The formula is:

[0083] ;

[0084] In the formula, The value range is 0-1 (0 represents no frequency domain correlation, 1 represents perfect linear correlation), where, This indicates excellent frequency domain coordination (high synchronization of moving units). This indicates good frequency domain coordination; This indicates frequency domain coordination degradation (requiring adjustment of motor unit synchronicity via electrical stimulation). For explosive movements such as sprinting, the analysis frequency band can be adjusted to 15-35Hz to match a higher motor unit synchronization frequency.

[0085] The third method for calculating the muscle coordination index is based on the integral area ratio of electromyography (EMG). This method is simple and intuitive to operate, primarily assessing the synergy of activation intensity, and is suitable for slow strength exercises such as squats and deadlifts. By comparing the cumulative activation of the agonist and antagonist muscles during key phases, it determines whether their intensity matching is reasonable. For example, during knee flexion and extension, the activation intensity of the antagonist hamstring should be controlled at 20%-30% of that of the agonist quadriceps. Too high an activation intensity will hinder the movement, while too low an activation intensity will lead to joint instability. Firstly, during the key phases of the movement... Within this, the integrated areas of the standardized electromyographic envelopes of the agonist and antagonist muscles (reflecting the cumulative activation within the key time phase) are calculated separately. Specifically, the integrated area of ​​the agonist muscle is... The calculation formula is:

[0086] ;

[0087] Integral area of ​​antagonistic muscle The calculation formula is:

[0088] ;

[0089] The larger the area, the higher the total activation intensity of the muscle during the critical phase.

[0090] Then, the ratio of the integral area of ​​the antagonist muscle to the integral area of ​​the agonist muscle is used as the coordination index. The formula is:

[0091] ;

[0092] in, The baseline range needs to be calibrated for different movements and joints, for example: knee flexion and extension: (The antagonist muscle activation intensity is 20%-30% of the agonist muscle). If the knee joint flexion is >0.3, it indicates overactivation of the antagonist muscle (synergistic degeneration, requiring inhibition of the antagonist muscle). If the knee joint flexion is <0.2, it indicates insufficient activation of the antagonist muscle (synergistic degeneration, requiring enhancement of antagonist muscle stability). Elbow curl: (The antagonist muscle, the triceps brachii, has lower activation intensity.)

[0093] In this embodiment of the invention, the three methods each have their own emphasis. In practical applications, the appropriate adaptation scheme can be selected according to the action type (fast / slow, dynamic / static) to ensure the accuracy of collaborative evaluation.

[0094] In one embodiment of the present invention, the real-time estimation of the net output torque of the target joint specifically involves inputting a clean kinematic time series and the amplitude of the standardized electromyographic envelope of the agonist muscle into a pre-calibrated simplified human musculoskeletal biomechanics model to calculate the net output torque in real time.

[0095] Net output torque of joint To quantify the actual force output of a target joint during movement, its estimation needs to integrate muscle activation intensity (electromyography signals) and joint motion state (angular velocity, angle), achieved through a simplified human musculoskeletal biomechanics model, balancing real-time performance and accuracy. This simplified model can include active torque calculation sub-models and passive torque calculation sub-models. Among these, the net joint output torque... The expression is:

[0096] ;

[0097] In the formula, The active torque generated by the contraction of agonist muscles (the core force that propels joint movement); Passive torques generated by the joint connective tissues (ligaments, cartilage, fascia) (resistance to joint movement, such as the tensile resistance of ligaments during knee flexion). These are estimated using electromyography (EMG) signals. Calculated using kinematic data The difference between the two is the net output torque.

[0098] The active torque is the sum of the contractile force of each agonist muscle multiplied by its lever arm (this invention includes at least one agonist muscle, but can be extended to multiple synergistic agonist muscles). The expression for the active torque calculation sub-model is as follows:

[0099] ;

[0100] In the formula, The number of agonist muscles is used; for example, when the knee is flexed and extended, n=1, only the quadriceps; when the elbow is curled, n=1, only the biceps.

[0101] The contractile force of a single agonist muscle is linearly correlated with the standardized electromyographic envelope. The amplitude of the electromyographic signal directly reflects the number of activated motor units, as shown in the formula:

[0102] ;

[0103] In the formula, The electromyography-mechanical calibration coefficient of the j-th muscle can be determined through the MVC experiment. The athlete performs the MVC for that muscle and records the results. ,correspond Substituting into the formula, we get (After standardization, usually) ); The standardized electromyographic envelope of the j-th agonist muscle is derived from the S1 preprocessing results (values ​​0-1). The maximum contractile force of the j-th muscle can be determined through isometric muscle strength testing, such as measuring the maximum force when the knee is fully extended using a muscle dynamometer.

[0104] The lever arm of a single agonist muscle is the vertical distance from the muscle attachment point to the center of joint rotation. Its magnitude varies with the joint angle. For example, when the knee is flexed, the lever arm of the quadriceps femoris muscle first increases and then decreases. It can be calculated using the angle-lever arm fitting formula, which is:

[0105] ;

[0106] In the formula, , , The lever arm calibration coefficient for the j-th muscle can be obtained by acquiring the coordinates of the muscle attachment point through a human anatomy database (such as the OpenSim open-source model) or CT scan, and then fitting the curve of the relationship between the joint angle and the lever arm. The clean angle of the target joint is derived from the S1 preprocessing results (unit: rad, angles in degrees need to be converted to radians). For example, taking the quadriceps lever arm of the knee joint as an example, if... If the knee flexion angle is in rad, then , , (Unit: m), the range of the lever arm after fitting is 0.02-0.04 m.

[0107] Passive torque Determined by both joint stiffness (related to angle) and joint damping (related to angular velocity), the expression for the passive moment calculation sub-model is:

[0108] ;

[0109] In the formula, The passive stiffness coefficient related to the joint angle can be calibrated through passive joint movement experiments. By slowly rotating the joint and recording the resistance torque at different angles, a fitting can be obtained. curve (unit: ); The damping coefficient is related to the joint angular velocity. It can be obtained by recording the relationship between the drag torque and angular velocity through passive joint movement experiments at different velocities, and then fitting the data. curve (unit: ); The clean angular velocity of the target joint is derived from the S1 preprocessing results (unit: rad / s). For example, taking the passive torque of the knee joint as an example... , ,when (Approximately 57°) hour, .

[0110] In summary, substituting the active torque and passive torque into the formula...

[0111] ,get:

[0112] ;

[0113] Among them, net output torque The accuracy can be directly verified by measuring joint torque directly using a torque sensor. For example, installing a torque sensor on training equipment can compare the estimated torque with the measured torque. Compare the results to ensure the error is ≤10%; if the error is too large, recalibration is required. (Electromyographic-mechanical coefficient) or , , (Lever arm coefficient) to eliminate interference factors such as muscle fatigue and sensor loosening.

[0114] The state recognition and classification module compares the calculated muscle coordination index with a preset coordination baseline threshold and determines whether coordination degradation has occurred based on the comparison result. At the same time, it compares the estimated net output torque with a preset target torque curve corresponding to the target training movement and determines whether strength decay has occurred based on the comparison result. Combining the determination results of coordination degradation and strength decay, the current athletic performance is identified as one of the following three categories: coordination-dominated, strength-dominated, or mixed.

[0115] This module uses a two-dimensional state determination and comprehensive type classification to extract the muscle coordination index from the two-dimensional feature extraction module. and joint net output torque This can be transformed into a motor performance category that can directly guide electrical stimulation strategies.

[0116] Specifically, the baseline threshold for coordination needs to be determined by combining the characteristics of the specific sport and benchmark data of elite athletes to calibrate specific threshold ranges for different types of muscle coordination indices. Specifically, for indices reflecting temporal coordination... Its baseline threshold range can be set to Its calibration can be achieved by collecting motion data from 20 elite athletes in the same specialty and extracting key time phases. Take the 25th percentile of the data as The 75th percentile is This threshold is applicable to rapid explosive movements such as sprinting and jumping; it reflects frequency domain coordination. The baseline interval can be set to Its calibration can be achieved by collecting electromyographic signals from elite athletes during the steady-state phase of their movements, calculating the mean coherence coefficient in the 8-30Hz frequency band, and taking the 30th percentile as the mean. 70th percentile Suitable for precise control actions such as shooting and archery; reflects strength coordination The knee joint can be set as different depending on the type of joint. The elbow joint can be set as The shoulder joint can be set as Its calibration can be performed by recruiting healthy subjects to complete standard movements, measuring the ratio of the electromyographic integral area of ​​the antagonist muscle to the agonist muscle, and taking the mean ± 1 standard deviation as the baseline interval. It is applicable to slow strength movements such as squats and deadlifts.

[0117] After establishing the baseline threshold for coordination, a determination of coordination deterioration can be made based on the rule that an index exceeding the baseline range constitutes deterioration. However, the meaning of deterioration differs for different types of muscle coordination indices. Specifically, for… , (0.5) indicates that the activation sequence of the agonist and antagonist muscles is asynchronous (e.g., the antagonist muscle is activated earlier). (0.8) Most of these are abnormal synchronizations (such as a "locked-in" state caused by excessive muscle tension), all of which are considered degenerative; for , (0.3) indicates that the motion unit lacks coordinated activation, resulting in poor motion control accuracy, and is thus classified as degenerate; for , This indicates overactivation of the antagonist muscle (such as excessive force exerted by the hamstrings during knee flexion and extension, which counteracts the force of the quadriceps). Insufficient activation of antagonistic muscles (poor joint stability) is considered a sign of degeneration.

[0118] To assign weights to subsequent mixed electrical stimulation parameters, the degree of degradation needs to be normalized to a 0-1 range (0 for no degradation, 1 for severe degradation). corresponding The calculation formula and derivation logic are as follows:

[0119] ;

[0120] In the formula, with or Using the denominator, the absolute difference is transformed into a relative proportion, eliminating individual differences in the absolute values ​​of the indices of different athletes; for example, if , ,but This indicates that the degree of temporal co-degradation is 40%.

[0121] corresponding The calculation formula and derivation logic are as follows:

[0122] ;

[0123] It should be noted that, The absence of excessive degradation indicates higher frequency domain synchronization and more stable operation, requiring only the identification of insufficient degradation.

[0124] corresponding The calculation formula and derivation logic are as follows:

[0125] ;

[0126] in, The higher the value, the further the strength matching between the antagonist and agonist muscles deviates from the optimal ratio, and the higher the risk of joint injury.

[0127] In one embodiment of the present invention, determining whether force attenuation has occurred specifically involves: calculating the difference between the time series of net output torque and the target torque curve to obtain the real-time torque gap; if the integral area or peak value of the real-time torque gap exceeds a preset force gap threshold, then force attenuation is determined to have occurred.

[0128] In this embodiment of the invention, the core of force attenuation determination is the overall assessment of the torque gap, which requires first constructing the target torque curve. As a criterion for judgment, its construction needs to be combined with the athlete's individual condition and training or rehabilitation goals, and there are three main methods. Among them, for strength improvement training for professional athletes, the individual peak state fitting method can be used. This method involves collecting joint torque data during the athlete's peak health period and obtaining the torque curve within the movement cycle through polynomial fitting. For novice athletes, basic training can rely on specialized standard models based on biomechanical databases for track and field, weightlifting, and other sports, and be personalized by taking into account the athlete's weight and height. For rehabilitation after sports injuries, it is necessary to build a rehabilitation stage target model, with rehabilitation physicians setting stage torque upper limits to avoid secondary injuries, and constructing a progressive target torque curve as the rehabilitation progresses.

[0129] Based on the target torque curve, first calculate the torque gap at a single moment. It reflects the difference between the actual torque and the target torque at that point in time, and the calculation formula is:

[0130] ;

[0131] In the formula, This indicates that the actual torque is insufficient, and there is a decrease in force. This indicates that the actual torque exceeds the target, which can easily lead to injury during the rehabilitation phase and is also considered an abnormal situation. To uniformly assess the size of the gap, its absolute value should be used for subsequent analysis.

[0132] To avoid misjudgment caused by a single torque fluctuation, the torque gap needs to be evaluated over the entire motion cycle. The core indicator is the integral area of ​​the torque gap. and peak torque gap .in, The calculation formula is:

[0133] ;

[0134] In the formula, It reflects the cumulative degree of torque deviation from the target within a single action cycle; the larger the area, the more unstable the overall force output; its threshold... The calibration method involves collecting three normal movements from the athlete. The threshold is set to 1.5 times the mean.

[0135] peak torque gap The formula is:

[0136] ;

[0137] In the formula, This reflects the instantaneous maximum torque gap within the motion cycle; an excessively high peak value can easily lead to motion distortion; its threshold... Take 20% of the target torque peak value as the threshold, for example, the target torque peak value is ,but .

[0138] Among them, if or If so, it is determined to be a force decay.

[0139] Degree of strength decline The expression is:

[0140] ;

[0141] This formula takes the maximum of the ratio of the integral area exceeding the threshold and the ratio of the peak value exceeding the threshold to ensure coverage of both attenuation types. If the threshold is not exceeded, then... (No attenuation); The value ranges from 0 to 1, with larger values ​​indicating more severe power attenuation. If , then take 1, representing extreme decay.

[0142] In completing the degree of coordination degradation and degree of strength decline After quantification, a comprehensive classification of the athletic performance type needs to be completed using dominant thresholds. These dominant thresholds include coordination dominant thresholds. and the power dominance threshold The value ranges from 0 to 1 and needs to be calibrated in conjunction with specific characteristics. Specifically, for precision control sports such as gymnastics and shooting... A value of 0.2-0.25 is acceptable, with priority given to addressing coordination issues. A value of 0.2-0.3 is acceptable; for explosive power sports such as sprinting and weightlifting, A value of 0.15-0.2 is acceptable; priority should be given to determining the force involved. A dose of 0.25-0.3 is acceptable; during the rehabilitation phase... and Both can be set to 0.2 to balance the issue of coordination and strength.

[0143] The rules for judging the three types of athletic performance are as follows: and The current type is coordination-driven, and the core problem is poor muscle synergy but the power output is basically up to standard. The subsequent intervention should focus on adjusting the muscle activation sequence. and The type is strength-driven, and the main problem is insufficient joint torque but normal muscle coordination, which needs to be improved by compensating for muscle contraction strength. and The current situation is mixed, with both coordination and strength failing to meet standards, requiring adjustment according to... Weights are allocated, and interventions combining time-series adjustments and intensity compensation are implemented.

[0144] For example, taking athletes recovering from knee injuries as an example, the rehabilitation phase... , If measured (Overactivation of antagonist muscles) If the torque is slightly insufficient but does not exceed the threshold, it is determined to be a coordination-dominant type, and subsequent electrical stimulation is needed to inhibit the overactivation of the antagonist muscle (hamstring).

[0145] It should be noted that an athlete's performance status will change during training / rehabilitation and needs to be retested and recalibrated every two weeks. / , , / This avoids misjudgments caused by outdated thresholds. Furthermore, to eliminate random errors from a single action, three consecutive action cycles must meet a certain judgment condition before the action is finally classified into the corresponding type, improving the reliability of the judgment.

[0146] The intelligent decision-making and execution module is used to invoke different stimulation parameter decision logics according to the identified category: if it is identified as coordination-dominant, it generates a first type of electrical stimulation parameter set with the core of adjusting the muscle activation timing; if it is identified as force-dominant, it generates a second type of electrical stimulation parameter set with the core of compensating for muscle contraction intensity; if it is identified as hybrid, it generates a third type of electrical stimulation parameter set that integrates timing adjustment and intensity compensation; finally, based on the generated electrical stimulation parameter set, electrical stimulation is applied to the corresponding target muscle group.

[0147] This module takes the type of motor performance and the corresponding degree of degeneration determined by the state recognition and classification module as input, and outputs appropriate electrical stimulation commands. It can not only target and solve the problems of muscle coordination disorder or insufficient force output, but also avoid muscle damage caused by overstimulation through multiple safety thresholds.

[0148] In one embodiment of the present invention, the intensity compensation logic on which the second type of electrical stimulation parameter set is based is specifically as follows: the instantaneous value or integral value of the real-time torque gap is dynamically converted into additional electrical stimulation intensity parameters for the agonist muscle through a predefined mapping function.

[0149] For strength-dominant motor performance, the core issue of the second type of electrical stimulation parameter set is to convert the net joint output torque gap into electrical stimulation intensity, achieving precise force intervention that compensates for the exact amount of torque deficiency. In this invention, regarding the intensity mapping method, the electrical stimulation intensity is correlated with the real-time torque gap. They are positively correlated. A linear or nonlinear mapping function can be selected depending on the intervention scenario (training / rehabilitation). Linear mapping is suitable for conventional strength training, and the formula is:

[0150] ;

[0151] In the formula, The intensity of the additional electrical stimulation of the agonist muscle used to compensate for the intensity of muscle contraction at time t; The baseline electrical stimulation intensity, which is the intensity that maintains the basic excitability of muscles, can be calibrated using the athlete's tolerance level, taking 30%-40% of the maximum tolerated intensity. For example, if the maximum tolerance is 20mA, then... ; The torque notch-strength mapping coefficient is determined by calibrating the muscle lever arm; muscles with small lever arms (such as fingers) are used. Muscles with large lever arms (such as the thigh) are selected. ; This is the real-time torque gap.

[0152] It should be noted that, ( (To the athlete's maximum tolerable intensity, to avoid muscle burns) If the calculated value exceeds... If ), then take For example, if , , ,but That is, through strength increment compensation of 3mA Torque gap.

[0153] For nonlinear mapping to be suitable for rehabilitation / power training, where sudden changes in intensity need to be avoided during rehabilitation and rapid increases in intensity are required during power training, an exponential mapping is used, with the formula as follows:

[0154] ;

[0155] In the formula, The intensity of the additional electrical stimulation of the agonist muscle used to compensate for the intensity of muscle contraction at time t; The target torque peak value, i.e., the baseline value of the normalized torque gap, comes from the target torque curve in the state recognition and classification module; It is a natural constant used to achieve smooth / rapid growth of intensity. When the intensity is low, the strength increases slowly (to protect muscles during the recovery period). When the intensity is high, the strength increases rapidly (meeting the needs of explosive power training).

[0156] Furthermore, the pulse width dynamically adjusts with the degree of force decay; the greater the decay, the wider the pulse width, thereby enhancing the muscle penetration depth of the stimulus and activating more motor units. The pulse width affects the penetration depth of the stimulus, as shown in the formula:

[0157] ;

[0158] In the formula, The base pulse width is 200-300μs. The attenuation level is a pulse width coefficient (0.5-0.8). To determine the degree of force decay, ensure that the more severe the decay, the deeper the stimulus penetrates (activating more motor units).

[0159] The stimulation target of this type of parameter set is only to stimulate the agonist muscle, thereby increasing the net output torque of the joint by enhancing the contractile force of the agonist muscle and avoiding stimulation of the antagonist muscle which would cause force cancellation.

[0160] In one embodiment of the present invention, the timing adjustment logic on which the first type of electrical stimulation parameter set is based is as follows: based on the time change characteristics of the muscle coordination index, the sub-phase with the worst muscle coordination within the action cycle is identified, and the optimal timing parameters for applying corrective electrical stimulation to the main antagonist muscle or agonist muscle are determined.

[0161] For coordination-driven motor performance, the core issue is misalignment of muscle activation timing (e.g., premature activation of antagonist muscles and delayed activation of agonist muscles). Therefore, the parameter set must focus on correcting the activation timing, reshaping the synergistic timing of agonist and antagonist muscles through precise stimulation timing and duration. Specifically, this first requires analyzing the time series of muscle coordination indices. In the middle, the sub-phase with the worst coordination within the positioning action cycle.

[0162] The basis for judgment is The slope of change:

[0163] ;

[0164] when Reaching the maximum value and continuously exceeding the preset slope threshold At that time, this period is the worst-case sub-stage of collaboration. .

[0165] For example, during knee flexion and extension movements, if , During the period The largest absolute value indicates that the temporal misalignment between the agonist and antagonist muscles is most severe during that period.

[0166] Key parameters in the first type of electrical stimulation parameter set include the optimal stimulation time. Duration of stimulation and stimulation frequency Among them, the optimal stimulation time point is determined based on the start and end times of the worst-case cooperative sub-phase. Its mathematical model is:

[0167] ;

[0168] In the formula, The start time of the worst-case sub-stage is the start time of the time sequence disorder. The worst-case sub-stage termination time is the termination time of the disordered sequence. Assigning coefficients to time points, i.e., the relative position of the stimulus within the disordered sub-phase, such as premature activation of antagonist muscles. (Early intervention); delayed activation of agonist muscles (Precise replacement). For example, if , , ,but That is, applying stimulation at 30% of the time in the disordered sub-stage.

[0169] Duration of stimulation It is positively correlated with the degree of coordination degradation, and the mathematical model is as follows:

[0170] ;

[0171] In the formula, The basal stimulus duration is the basal correction duration without regression. This is the degradation degree-duration mapping coefficient, i.e., the gain coefficient of degradation degree on duration; The degree of coordination degradation, i.e., the severity of temporal disorder, is a quantized value (0-1) derived from the state recognition and classification module. For example, if... , , ,but That is, due to a 40% degree of degradation, the stimulation duration is extended by 40%.

[0172] To avoid muscle adaptation to a fixed frequency, a stimulation frequency synchronized with the movement frequency is used. Its mathematical model is:

[0173] ;

[0174] In the formula, The duration of a single movement cycle ensures that the stimulus is synchronized with the movement rhythm. For example, if the knee flexion-extension cycle is 0.5 seconds, then... .

[0175] In this embodiment of the invention, the selection of the target muscle group to be stimulated will be adjusted according to the specific type of temporal disorder. If the antagonist muscle is activated prematurely ( corresponding If the agonist muscle is activated late ( ), then the stimulation target is the antagonist muscle, and premature excitation is inhibited by electrical stimulation; if the agonist muscle activation is delayed ( If the stimulation target is the agonist muscle, its contraction is triggered in advance through electrical stimulation; if the intensity coordination is unbalanced ( If the antagonist muscle is overactivated (abnormal), it will inhibit the antagonist muscle, while underactivation will slightly enhance the antagonist muscle.

[0176] In one embodiment of the present invention, generating the third type of electrical stimulation parameter set specifically involves: assigning dynamic weights to the temporal parameters in the first type of electrical stimulation parameter set and the intensity parameters in the second type of electrical stimulation parameter set based on the relative proportions of the degree of coordination degradation and the degree of strength attenuation determined in the state recognition and classification module, and then performing weighted fusion.

[0177] Hybrid athletic performance corresponds to a situation where muscle synergy disorder and strength deficiency coexist. Its parameter set needs to integrate the core logic of the first two types of parameters, achieving a balance between temporal adjustment and intensity compensation through dynamic weight allocation. The weight allocation is based on the relative proportion of the degree of coordination decline to the degree of strength loss; the proportion of these two directly determines the weight of the temporal and intensity parameters after integration. Specifically, based on the degree of coordination decline... With the degree of strength decline The relative proportions are used to calculate the weights of time-series parameters. and intensity parameter weights The calculation formula is:

[0178] ;

[0179] ;

[0180] in, =1, ensuring the normalization of weight allocation; for example, if , ,but , This indicates that intensity compensation accounts for 60% of the weight, while timing adjustment accounts for 40%.

[0181] In terms of specific parameter generation, the intensity of the fused stimulus will be biased towards the compensation ratio of force attenuation, the timing of the fused stimulus will take into account the need for correction of coordination disorder, and the duration of the stimulus will be balanced between the timing correction duration and the force compensation duration according to weight.

[0182] Specifically, the intensity of stimulation after fusion The expression is:

[0183] ;

[0184] Among them, weight The higher the value, the greater the proportion of strength compensation; The proportion of basic strength retained should be maintained to avoid overcompensation that could further disrupt synergy.

[0185] Post-fusion stimulation time point The expression is:

[0186] ;

[0187] In the formula, The baseline stimulation point, i.e., the regular stimulation time without synergistic disturbances, can be calibrated using the movement cycle and taken as the midpoint of the force exertion phase of the movement, such as the force exertion point in a squat. Among them, weight The higher the value, the more the stimulation time is biased towards the worst synergy sub-stage; conversely, the lower the value, the closer it is to the normal exertion time, taking into account the timeliness of force compensation.

[0188] Post-fusion stimulation duration The expression is:

[0189] ;

[0190] in, The base duration for strength compensation (taken) This achieves a balance between timing correction and intensity compensation in terms of duration.

[0191] In this embodiment of the invention, if during the training / rehabilitation process or Changes, weights , It can be updated in real time, for example, in the later stages of recovery. reduce, If it rises, it will automatically rise. Strengthen the proportion of force compensation.

[0192] In one embodiment of the present invention, in the intelligent decision-making and execution module, the timing of applying electrical stimulation is the moment corresponding to the decision sequence in the set of electrical stimulation parameters within the next or current action cycle.

[0193] In the precise application of electrical stimulation, if the data for the current action cycle has been collected and the determination is complete (sufficient real-time performance), the current cycle can be... Stimulation is applied continuously; if real-time performance is insufficient, such as due to lengthy data preprocessing, stimulation is applied at the corresponding moment in the next action cycle to ensure precise timing alignment. The multi-channel electrical stimulation execution module must ensure timing synchronization between channels to avoid exacerbating muscle synergy disorder due to stimulation misalignment between channels. Simultaneously, to ensure athlete safety, this invention sets three types of termination thresholds: intensity threshold, duration threshold, and feedback threshold. For the intensity threshold, when… Automatically descends Regarding the duration threshold, To avoid overstimulation; for feedback thresholds, if athletes report muscle soreness / numbness, stimulation pause can be manually triggered.

[0194] In this invention, the parameter system in the intelligent decision-making and execution module needs to be specifically calibrated for different sports and intervention scenarios. For example, the basic stimulus intensity is higher and the duration is shorter for explosive sports such as sprinting, while the stimulus intensity is lower and the duration is longer for fine control sports such as shooting. In the knee joint rehabilitation scenario, safety and effectiveness need to be considered, and a parameter benchmark of moderate intensity and duration should be selected. In addition, this invention also has a dynamic optimization mechanism, such as re-collecting data after every 5-10 sets of movements. and If the degree of degradation / If the decrease is more than 15%, the intervention intensity of the corresponding parameter will be automatically reduced, such as... Reduce by 10% Shorten by 10% to achieve gradual intervention and avoid muscle dependence on stimulation.

[0195] According to an embodiment of the present invention, the athlete muscle strength training and rehabilitation system based on electrical stimulation simultaneously collects electromyographic and kinematic signals, extracts the dual-dimensional features of muscle coordination index and joint net output torque in real time, identifies three types of motor performance states: coordination-dominant, strength-dominant, and mixed, and dynamically generates differentiated electrical stimulation parameter sets accordingly. This enables precise classification and targeted intervention of the athlete's fatigue state, thereby simultaneously optimizing muscle synergy efficiency and strength output during training, effectively preventing compensatory injuries and accelerating functional recovery during rehabilitation, and improving the personalization, safety, and overall effectiveness of training and rehabilitation.

[0196] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0197] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of the invention pertain.

[0198] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0199] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0200] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0201] Furthermore, the functional units in the various embodiments of the present invention can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0202] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A muscle strength training and rehabilitation system for athletes based on electrical stimulation, characterized in that, include: The data acquisition and preprocessing module is used to simultaneously acquire the raw electromyographic signals of at least one pair of agonist muscles and their main antagonist muscles, as well as the kinematic data of the target joint, when the athlete performs the target training movement; process the raw electromyographic signals to obtain the standardized electromyographic envelopes of the agonist muscles and the main antagonist muscles respectively; process the kinematic data to obtain a clean kinematic time series; wherein the kinematic data includes at least joint angles and angular velocities; A dual-dimensional feature extraction module is used to calculate a muscle coordination index to quantify the synergistic working state of the agonist muscle and the main antagonist muscle based on the standardized electromyographic envelope of the agonist muscle; at the same time, based on the clean kinematic time series and the amplitude of the standardized electromyographic envelope of the agonist muscle, the net output torque of the target joint during the execution of the target training action is estimated in real time. Specifically, the real-time estimation of the net output torque of the target joint involves inputting the clean kinematic time series and the standardized electromyographic envelope amplitude of the agonist muscle into a pre-calibrated simplified human musculoskeletal biomechanics model to calculate the net output torque in real time. The state recognition and classification module is used to compare the calculated muscle coordination index with a preset coordination baseline threshold, and determine whether coordination degradation has occurred based on the comparison result; at the same time, it compares the estimated net output torque with a preset target torque curve corresponding to the target training movement, and determines whether strength decay has occurred based on the comparison result; combining the determination results of coordination degradation and strength decay, the current sports performance is identified as one of the following three categories: coordination-dominated, strength-dominated, or mixed. The intelligent decision-making and execution module is used to invoke different stimulation parameter decision logics according to the identified category: if it is identified as the coordination-dominant type, a first type of electrical stimulation parameter set with adjusting the muscle activation timing as the core is generated; if it is identified as the force-dominant type, a second type of electrical stimulation parameter set with compensating for muscle contraction intensity as the core is generated; if it is identified as the hybrid type, a third type of electrical stimulation parameter set integrating timing adjustment and intensity compensation is generated; finally, based on the generated electrical stimulation parameter set, electrical stimulation is applied to the corresponding target muscle group. The timing adjustment logic used to generate the first type of electrical stimulation parameter set is as follows: based on the time change characteristics of the muscle coordination index, the sub-phase with the worst muscle coordination within the action cycle is identified, and the optimal timing parameters for applying corrective electrical stimulation to the main antagonist muscle or the agonist muscle are determined. The generation of the third type of electrical stimulation parameter set is specifically as follows: based on the relative proportion of the degree of coordination degradation and the degree of strength attenuation determined in the dual-dimensional feature extraction module, dynamic weights are assigned to the temporal parameters in the first type of electrical stimulation parameter set and the intensity parameters in the second type of electrical stimulation parameter set, and then weighted fusion is performed.

2. The electrostimulation-based athlete muscle strength training and rehabilitation system according to claim 1, characterized in that, In the dual-dimensional feature extraction module, the calculation of the muscle coordination index specifically involves: calculating the peak value of the cross-correlation function of the standardized electromyographic envelopes of the agonist muscle and the main antagonist muscle in the key phase of the movement, or calculating the coherence coefficient of the two in a specific frequency band, or calculating the ratio of the electromyographic integral area of ​​the main antagonist muscle to that of the agonist muscle.

3. The electrostimulation-based athlete muscle strength training and rehabilitation system according to claim 2, characterized in that, In the state recognition and classification module, determining whether force decay has occurred specifically involves: The difference between the time series of the net output torque and the target torque curve is calculated to obtain the real-time torque gap; if the integral area or peak value of the real-time torque gap exceeds the preset force gap threshold, it is determined that force attenuation has occurred.

4. The electrostimulation-based athlete muscle strength training and rehabilitation system according to claim 3, characterized in that, In the intelligent decision-making and execution module, the intensity compensation logic for generating the second type of electrical stimulation parameter set is as follows: the instantaneous value or integral value of the real-time torque gap is dynamically converted into additional electrical stimulation intensity parameters for the agonist muscle through a predefined mapping function.

5. The electrostimulation-based athlete muscle strength training and rehabilitation system according to claim 1, characterized in that, In the intelligent decision-making and execution module, the timing for applying electrical stimulation is the moment within the next or current action cycle that corresponds to the decision sequence in the set of electrical stimulation parameters.

Images