Facial paralysis electrical stimulation system based on VR and dynamic facial myoelectric threshold feedback

Abstract

This invention provides a facial paralysis electrical stimulation system based on VR and dynamic facial electromyography (EMG) threshold feedback. The system includes a VR module for constructing a virtual rehabilitation scenario and dynamically adjusting the task difficulty within that scenario; an EMG acquisition module for acquiring physiological parameters of the patient's target muscle groups during operation, based on a resting baseline mode or a task interaction mode; an analysis and control module for extracting features from the physiological parameters, performing threshold comparisons, matching and executing corresponding intervention plans according to a preset selection strategy, and dynamically updating the judgment threshold for the next acquisition cycle based on the comparison results of the previous acquisition cycle; and a data recording module for reproducing the patient's baseline parameter set to obtain a rehabilitation trend assessment as a basis for adjusting the task difficulty.

Description

Technical Field

[0001] This invention belongs to the field of bioelectrical signal detection and feedback technology, specifically relating to a facial paralysis electrical stimulation system based on VR and dynamic facial electromyography threshold feedback. Background Technology

[0002] In key medical fields such as rehabilitation medicine, neurology, and bioengineering, neuromuscular electrical stimulation (NMES) systems serve as a core rehabilitation method to promote muscle function reconstruction. The accuracy of their stimulation parameters and the real-time interactivity directly determine the patient's rehabilitation efficiency, the degree of motor function recovery, and clinical treatment compliance. Therefore, intelligent electrical stimulation technology based on biofeedback has become a core research direction in the field of neurorehabilitation engineering.

[0003] With the rapid iteration of sensor technology, virtual reality (VR) technology, and digital signal processing technology, the electrostimulation rehabilitation system has evolved from early open-loop stimulation relying on fixed parameters, such as muscle training with preset fixed frequencies and amplitudes, to a closed-loop interactive system integrating electromyography (EMG) monitoring. The monitoring dimensions have also expanded from single muscle contraction intensity to multimodal physiological parameter co-sensing, covering the frequency characteristics, temporal amplitude, muscle fatigue indicators, and motion feedback in the virtual environment of surface electromyography (sEMG). Simultaneously, to enhance patients' training immersion and willingness to actively move, VR technology, with its advantages of strong interactivity, rich scenarios, immediate feedback, and high rehabilitation enjoyment, has been widely integrated into rehabilitation programs for motor dysfunction, becoming a key interactive node connecting the digital virtual environment and biophysical stimulation. By setting specific movement tasks in a VR scene, patients can be guided to perform targeted muscle group training.

[0004] Existing VR-assisted electrical stimulation solutions generally adopt independent processing modes with preset parameters or lagging feedback logic. This makes it impossible to dynamically adjust the stimulation threshold according to the patient's real-time muscle state during treatment. This not only easily leads to muscle over-fatigue or insufficient treatment intervention, but also easily results in low training efficiency due to the logical disconnect between electromyographic signals and VR tasks. Furthermore, due to the lack of in-depth multimodal data fusion and real-time closed-loop control, it cannot meet the needs of refined facial paralysis rehabilitation and complex motor function reconstruction. Summary of the Invention

[0005] To address the shortcomings of existing technologies, such as fixed electrical stimulation parameters, delayed feedback logic, and a lack of deep multimodal fusion leading to low rehabilitation efficiency, this invention aims to achieve intelligent, real-time monitoring and closed-loop intervention in the rehabilitation process of facial paralysis and motor dysfunction by synergistically combining virtual reality task guidance, dynamic electromyography threshold analysis, and a closed-loop electrical stimulation feedback scheme. A facial paralysis electrical stimulation system based on VR and dynamic facial electromyography threshold feedback is provided, including: The VR module is used to build virtual rehabilitation scenarios and can dynamically adjust the difficulty of tasks in the virtual rehabilitation scenarios. The electromyography (EMG) acquisition module acquires physiological parameters of the patient's target muscle group during operation, based on either a resting baseline mode or a task interaction mode. The analysis and control module extracts features from the physiological parameters and performs threshold comparisons. It matches and executes the corresponding intervention plan according to the preset selected strategy. The analysis and control module dynamically updates the judgment threshold for the next collection cycle based on the comparison results of the previous collection cycle. The data recording module is used to reproduce the baseline parameter set of patients and obtain a rehabilitation trend assessment as a basis for comparison in adjusting the difficulty of the task.

[0006] The analysis and control module is configured with selected strategies, which include: The pre-built database stores multiple rehabilitation stage types, and each rehabilitation stage type is pre-matched with corresponding baseline electromyographic features and associated stimulation parameters. The baseline electromyographic features reflect the muscle contraction ability under the corresponding rehabilitation stage, and the associated stimulation parameters reflect the upper limit of the corresponding electrical stimulation intensity. The analysis and control module acquires muscle operating parameters as regular real-time RMS values, extracts features of regular real-time RMS values ​​and compares them with dynamic thresholds to obtain the deviation relationship, uses the ratio of the RMS value of the current acquisition cycle to the dynamic threshold as the judgment basis, and judges the degree of threshold deviation. If the real-time RMS value is lower than the dynamic threshold, then execute the first analysis plan; If the coefficient of variation corresponding to the real-time RMS value is greater than the preset coefficient of variation threshold, then execute the second analysis scheme; If the real-time RMS value meets the triggering conditions and the stimulus waveform needs to be optimized, then the third analysis scheme is executed.

[0007] The first analysis scheme includes: The patient's resting and maximum contraction electromyographic values ​​are obtained through an initial threshold setting procedure, and the corresponding initial threshold is initially determined based on the baseline electromyographic characteristics. According to the acquisition cycle, the real-time RMS value is obtained from the electromyography acquisition module, and the resting reference mode is switched to the task interaction mode. In the task interaction mode, the real-time threshold is adjusted by the dynamic coefficient to obtain the reinforcement feedback threshold. The reference parameter after adjusting the dynamic coefficient is reproduced by the data recording module and compared with the real-time RMS value to calculate the degree of deviation. By comparing the degree of deviation in the selected strategies, if the real-time RMS value is continuously lower than the reinforcement feedback threshold, the intensity of electrical stimulation is increased; if the coefficient of variation corresponding to the real-time RMS value exceeds the preset value, the second analysis scheme is executed; if the real-time RMS value meets the triggering conditions and the frequency needs to be optimized, the third analysis scheme is executed.

[0008] Using the selected strategy, in the first analysis scheme, the intensity of the electrical stimulation current is adjusted in a step-by-step manner based on the comparison between the real-time RMS value and the reinforcement feedback threshold. If the real-time RMS value is continuously lower than the reinforcement feedback threshold, the intensity of the electrical stimulation current is gradually increased. If the real-time RMS value reaches the reinforcement feedback threshold, the current intensity of the electrical stimulation current is maintained unchanged.

[0009] The second analysis scheme includes: Establish fatigue assessment benchmark patterns in a pre-built database; The fluctuation pattern of the electromyography waveform to be analyzed is compared with the fatigue assessment benchmark pattern to determine whether it conforms to the fatigue state. If it does not conform, it is judged as normal fluctuation; if it conforms, it is judged as muscle fatigue, and the abnormality confirmation sub-strategy is executed. The VR module is used to reproduce the task difficulty reduction instruction determined to be in a fatigued state. The deviation is compared with the real-time muscle performance. If the muscle recovery level is determined to be lower than the recovery threshold, the resting baseline mode is switched; otherwise, the third analysis scheme is executed.

[0010] The anomaly confirmation sub-strategy includes: Feature extraction is performed on the electromyographic waveforms to be analyzed that exceed the fatigue threshold to generate fatigue feature vectors; Based on the task interaction mode, the weight or resistance parameters of objects in the VR scene are adjusted downwards to obtain the difference in electromyographic response before and after adjustment, and the fatigue recovery index is calculated. If the fatigue recovery index is lower than the preset response threshold, the third analysis scheme will be executed based on the fatigue recovery index.

[0011] The third analysis scheme includes: The pre-built database contains pre-stored combinations of electromyographic spectral features and corresponding frequency optimization rules, as well as electrical stimulation optimization parameters associated with each combination of electromyographic spectral features. The characteristics of the waveform to be analyzed are combined and compared with the characteristics of the electromyography spectrum, and then... Wave power variation is used as the criterion. like If the reduction in wave power is greater than the first preset threshold, then the electrical stimulation frequency will be increased. like If the fluctuation of wave power within the preset monitoring period is within the preset stability threshold, then the spectral feature is used as the reference feature and stored. like When the wave power fluctuation is in the dynamic range, the VR task interaction mode is entered. By adjusting the action frequency of the VR task, the amplified electromyographic spectrum features are obtained and compared with the electromyographic spectrum feature combination in the database. If the similarity of all feature combinations is lower than the second preset threshold, the electrode adaptive layout parameters are recalculated and written into the database.

[0012] The electrode adaptive layout parameters are recalculated based on abnormal spectral features with similarity below a preset range, and are used to optimize the signal acquisition quality of subsequent acquisition cycles.

[0013] The resting reference mode includes: The electromyography (EMG) acquisition module acquires actual physiological parameters at rest according to a preset sampling frequency, and records the average values ​​of the parameters within a preset time period, including the average, maximum, and minimum values ​​of the EMG potentials. Based on the records, an initial rehabilitation baseline report for the patient is generated.

[0014] The task interaction modes include scenarios that only monitor without intervention and scenarios that monitor and intervene. The scenario of monitoring without intervention includes: the electromyography acquisition module records the electromyography signals point by point during the preparation phase of the VR task and calculates the real-time RMS waveform in real time; The monitoring and intervention scenario includes: when the triggering condition is met, an electrical stimulation operation is initiated. The electrical stimulation operation is achieved by adjusting the magnitude or frequency of the electrical stimulation current and simultaneously performing periodic electromyography (EMG) acquisition during the exertion phase of the VR task, recording the parameters point by point, and recording the real-time contraction waveform after the electrical stimulation intervention. The triggering conditions include: the real-time RMS value is lower than the dynamic threshold in the first analysis scheme; the system is in the abnormal confirmation sub-strategy execution phase; and in the third analysis scheme... The wave power has shifted significantly.

[0015] The advantages and positive effects of this invention are: This invention addresses the problems of low patient participation and unintuitive movement guidance in traditional rehabilitation by setting synchronous movement tasks in a VR scene and utilizing the virtual interactive feedback it provides to simulate a real rehabilitation training environment. Under task guidance, the system can change the timing of electrical stimulation intervention through a real-time dynamic threshold algorithm, enabling functions such as amplifying muscle contraction characteristics and verifying the root causes of motor function. This ensures the fun of training and provides an operable biofeedback basis for multimodal monitoring and progressive analysis, realizing real-time dynamic control in the context of refined facial paralysis rehabilitation.

[0016] This invention uses a resting baseline mode to adapt to the initial state to establish a personalized model, a task interaction mode to focus on the training state to capture muscle recruitment details, and an analysis and control module to link the first, second, and third analysis schemes in a hierarchical manner based on the logical judgment of RMS and coefficient of variation values. This forms a closed-loop logic from threshold triggering to fatigue protection and then to parameter optimization, which not only avoids the training ineffectiveness caused by fixed parameters, but also ensures that resources are prioritized for high-value muscle activation areas, thus greatly improving the accuracy of motor function reconstruction and rehabilitation efficiency.

[0017] This invention solves the problem that simply changing the current intensity cannot effectively regulate the type of muscle fiber recruitment by optimizing the linkage between spectral characteristics and electrical stimulation frequency. Furthermore, the adaptive electrode layout and the new feature storage mechanism endow the system with adaptive optimization capabilities, which can continuously optimize the patient's personalized rehabilitation parameters as the training process progresses. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0019] In the attached diagram: Figure 1 This is a flowchart of dynamic facial electromyography threshold feedback and graded electrical stimulation intervention.

[0020] Figure 2 This is a logic diagram for switching operating modes and fatigue verification. Detailed Implementation

[0021] 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.

[0022] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0023] In the rehabilitation training of facial paralysis patients, the facial muscles often exhibit extremely high sensitivity and easy fatigue due to impaired nerve innervation. Traditional electrical stimulation systems typically employ fixed stimulation frequencies and intensities. This application, through research, found that the low rehabilitation efficiency and poor patient compliance are due to the following: When a constant current pulse is continuously applied to the target muscle bundle, the lack of dynamic matching between the movement intention and stimulation parameters leads to an inability of acetylcholine release at the neuromuscular junction to keep up with the recruitment demand induced by high-frequency electrical stimulation, resulting in significant chemical fatigue. If the system cannot adjust the charge density according to the real-time physiological state of the muscles, the electrical stimulation energy will accumulate linearly, causing a polarization shift in local tissue impedance and producing discomfort. However, simultaneously… In VR virtual environments, visual feedback tasks continuously guide patients to attempt high-intensity facial movements. This is because the active intention driven by the visual and auditory tasks and the passive contraction assisted by electrical stimulation have a logical discontinuity in time and phase, causing the remodeling of neural pathways to be forced in an asynchronous state. This can easily lead to the misrecruitment of synergistic muscles or the ineffective antagonism of antagonistic muscles, i.e., an imbalance in the interaction between the neuromuscular and VR environment. Under the combined effect of high-intensity virtual tasks and fixed-parameter electrical stimulation, the local motor units consume too much energy under high-frequency drive and fail to obtain effective functional compensation, which can cause fatigue symptoms such as tremors, sudden decrease in contraction force, or soreness in facial muscles. If the response of some estimation methods is too slow, muscle damage or ineffective training can occur during the training process.

[0024] To address the aforementioned technical problems, this invention provides a facial paralysis electrical stimulation system based on VR and dynamic facial electromyography threshold feedback, comprising: The VR module, which runs in a virtual reality device, is used to construct and display immersive sports rehabilitation scenarios to provide interactive visual guidance for rehabilitation training. It enables the setting of specific sports tasks in a virtual environment and guides patients to generate active sports intentions through dynamic adjustment of scenario difficulty. It also provides a logically synchronized physical triggering basis for electromyography signal acquisition and electrical stimulation feedback. The electromyography (EMG) acquisition module acquires physiological parameters of the patient's target muscle group during operation according to the resting baseline mode or the task interaction mode. It acquires muscle operation parameters through two monitoring modes and collects data on the time domain characteristics, frequency domain characteristics and coefficient of variation of muscle contraction intensity. The analysis and control module extracts features from the physiological parameters and performs threshold comparison. It matches and executes the corresponding intervention plan according to the preset feedback strategy. The analysis and control module dynamically updates the judgment threshold of the next collection cycle based on the comparison results of the previous collection cycle. Through feature extraction, dynamic threshold calculation and feedback strategy matching, it realizes the analysis closed loop from action perception to stimulus intervention. It dynamically selects the first analysis plan, the second analysis plan or the third analysis plan according to the electromyographic signal characteristics, and links subsequent adjustments based on the previous threshold to form a progressive closed-loop logic. This avoids excessive fatigue caused by fixed parameters and reduces the probability of false triggering through dynamic thresholds. The data recording module is used to reproduce and obtain baseline parameters in the rehabilitation program, obtain rehabilitation trend assessment as a comparison basis for adjusting the training difficulty, and realize the degree verification of clinical assessment data and real-time training data by recording the coupling relationship between electromyographic signals and electrical stimulation parameters.

[0025] Specifically, the electromyography acquisition module also includes an 8-channel surface electrode, a Butterworth filter, a notch filter, and an A / D converter; The input terminals of the 8-channel surface electrodes are attached to the target muscle surface, and signal coupling is achieved using conductive gel to obtain electromyographic amplitude and spectral parameters of muscle operation. The Butterworth filter is used to filter out low-frequency baseline drift and high-frequency electromagnetic noise, and the notch filter is used to filter low power frequency interference. The A / D converter is used to convert analog potential signals into digital real-time electromyographic waveforms. The root mean square (RMS) value and coefficient of variation of the muscle are calculated using the real-time electromyographic waveforms through a sliding window algorithm. The RMS value is used to characterize the muscle contraction intensity, and the coefficient of variation is used to characterize the stability of motor unit recruitment, providing multi-physical quantity data support for the analysis and control module and avoiding the limitations of single amplitude judgment.

[0026] Specifically, the resting baseline mode includes: the electromyography (EMG) acquisition module acquires actual physiological parameters in the resting state according to a preset sampling frequency, and records the average values ​​of the parameters within a preset time period, including the average, maximum, and minimum values ​​of the EMG potentials. Based on the records, an initial rehabilitation baseline report for the patient is generated. This mode is used to assess the patient's basic muscle strength before training and to reduce the energy consumption of the electrical stimulation module when there is no task trigger. In this embodiment, the preset duration is the duration for acquiring a stable resting EMG baseline, preferably 18 seconds. In addition, to meet the effective bandwidth requirement of 10-450Hz for surface EMG signals, according to the Nyquist sampling theorem, the sampling rate needs to be no less than 900Hz. In this embodiment, the preset sampling frequency is preferably 1200Hz, which can avoid signal aliasing, avoid 50Hz power frequency harmonic interference, and also take into account the system's computational load.

[0027] Task interaction modes include: monitoring-only-no-intervention scenarios and monitoring-and-intervention scenarios; The monitoring-only-without-intervention scenario includes: the electromyography (EMG) acquisition module records the EMG signals point by point during the preparation phase of the VR task and calculates the real-time RMS waveform as a real-time reference for the background noise. The preparation phase of the VR task is the preparatory period after the VR module issues the action prompt, before the patient actively contracts the target facial muscles. This period is used to collect the basic EMG signals of the patient in the preparatory state of the action, providing a benchmark for judging the subsequent force exertion phase. For example, if the VR prompts: "Please prepare to puff out your cheeks," the patient concentrates and prepares to exert force on their face, but has not yet puffed out their cheeks.

[0028] The monitoring and intervention scenarios include: when the triggering conditions are met, an electrical stimulation operation is initiated. The intervention operation is carried out by adjusting the magnitude or frequency of the electrical stimulation current and simultaneously collecting electromyography data for a preset period during the exertion phase of the VR task. The parameters are recorded point by point, and the real-time contraction waveform after the electrical stimulation intervention is recorded. The preset period is the duration for which electromyographic characteristics can be stably acquired and real-time feedback can be achieved. In this embodiment, each period is preferably 3 seconds. The exertion phase of the VR task is as follows: the patient actively exerts force and contracts the target facial muscle groups according to the VR task guidance to complete the specified rehabilitation movement.

[0029] Specifically, the triggering conditions include the real-time value falling below the dynamic threshold in the first analysis scheme, the execution of the anomaly confirmation sub-strategy, and the third analysis scheme. The parameters to be analyzed where the wave power has shifted significantly.

[0030] Furthermore, the analysis and control module is configured with a selected strategy, which includes: The pre-built database stores various rehabilitation stage types, such as: edema stage, linkage stage, recovery stage, etc., and each rehabilitation stage type is pre-matched with corresponding baseline electromyographic characteristics and associated stimulation parameters. The analysis and control module acquires muscle operating parameters as regular real-time RMS values, extracts features of regular real-time RMS values ​​and compares them with dynamic thresholds to obtain the deviation relationship, and uses the ratio of the RMS value of the current acquisition cycle to the dynamic threshold as the judgment basis. If the real-time RMS value is lower than the dynamic threshold, then execute the first analysis plan; If the coefficient of variation corresponding to the real-time RMS value is greater than the preset coefficient of variation threshold, then execute the second analysis scheme; If the real-time RMS value meets the triggering conditions and the stimulus waveform needs to be optimized, for example, if the RMS meets the standard but the spectral efficiency is low, then the third analysis scheme is executed.

[0031] For example, the database contains pre-stored information such as: Rehabilitation stage A is the edema stage of facial paralysis symptoms, and the baseline electromyographic characteristics are resting potential fluctuations <5. The maximum shrinkage RMS threshold is 20. The upper limit of the associated stimulus parameter is 15mA; And the dynamic threshold is initially set to 5. The real-time RMS value obtained through the electromyography acquisition module was 12. Below 15 This precisely guides the analysis to the first analysis plan, using increased electrical stimulation intensity to assist muscle contraction and achieve dynamic switching of the analysis path.

[0032] In this embodiment, the preset coefficient of variation threshold is set to 0.3, which is a conventionally preferred critical value determined in the art based on facial electromyographic characteristics, muscle contraction stability and clinical rehabilitation experience, and is used to distinguish between normal contraction and fatigue or abnormal contraction states.

[0033] Specifically, the first analysis plan includes: The patient's 18-second resting and maximum contraction electromyography (EMG) values ​​are obtained through an initial threshold setting procedure, and the initial threshold is calculated. In this embodiment, the initial threshold is the resting mean value + 30% of the maximum contraction value. Real-time RMS values ​​are obtained from the EMG acquisition module every 3 seconds according to the acquisition cycle, and the system switches from resting reference mode to task interaction mode. In task interaction mode, the real-time threshold is adjusted by a dynamic coefficient to obtain the reinforcement feedback threshold. The dynamic coefficient is automatically corrected according to the trend of the previous 3 seconds of signal. Using the selected strategy, in the first analysis scheme, the intensity of the electrical stimulation current is adjusted in a step-by-step manner based on the comparison between the real-time RMS value and the reinforcement feedback threshold. If the real-time RMS value remains below the reinforcement feedback threshold, the intensity of the electrical stimulation current is gradually increased. If the real-time RMS value reaches the reinforcement feedback threshold, the current intensity of the electrical stimulation current is maintained, i.e., square wave electrical stimulation is output using the electrical stimulation module. Furthermore, by comparing the real-time RMS value with the reinforcement feedback threshold in the selected strategy, if the real-time RMS value remains below the reinforcement feedback threshold, the intensity of the electrical stimulation is increased by 1 mA. If the real-time RMS value reaches the reinforcement feedback threshold, the current stimulation is maintained.

[0034] For example, in the selected strategy, the initial threshold for entering the first analysis scheme is set to 20. The first data acquisition was performed over a 3-second cycle, with a real-time RMS value of 18μV < 20. The system automatically outputs a 5mA stimulus; In the second cycle, due to the patient's increased effort and the upward trend from the previous cycle, the dynamic coefficient was set to 1.05, and the reinforcement feedback threshold was 20. ×1.05=21 ; If the real-time RMS is only 19 at this time The electrical stimulation intensity automatically increases from 5mA to 6mA, achieving precise compensation for weak contractions through dynamic adjustment.

[0035] Specifically, the second analysis scheme includes: A fatigue assessment benchmark pattern is established in a pre-built database, wherein the fatigue assessment benchmark pattern is the synergistic relationship between the slope of the decrease in the median frequency of electromyography and the coefficient of variation. The fluctuation pattern of the electromyography waveform to be analyzed is compared with the fatigue assessment benchmark pattern to determine whether it conforms to the fatigue state. If it does not conform, it is judged as normal fluctuation; if it conforms, it is judged as muscle fatigue, and the abnormality confirmation sub-strategy is executed. The VR module is used to reproduce the task difficulty reduction instruction determined to be in a fatigued state. The deviation is compared with the real-time muscle performance. If the muscle recovery level is determined to be lower than the recovery threshold, the resting baseline mode is switched; otherwise, the third analysis scheme is executed.

[0036] In this embodiment, the fatigue assessment benchmark is obtained through statistical analysis of a large amount of clinical electromyography data. It is used to assess muscle fatigue status and is the slope of the decrease in median electromyography frequency as fatigue progresses. There is a corresponding change relationship between the slope and the coefficient of variation. Through this synergistic pattern, we can comprehensively judge whether the muscle is in a state of normal contraction, mild instability, or true fatigue, thereby improving the accuracy of fatigue identification. The specific judgment method is as follows: compare the coefficient of variation of the electromyography waveform to be analyzed with the preset fatigue threshold of 0.3. If the slope of the decrease in the median frequency of electromyography is close to 0 and the coefficient of variation is less than 0.3, the muscle is judged to be in a normal contraction state. If the slope of the decrease in the median frequency of electromyography is small and the coefficient of variation is between 0.3 and 0.5, the muscle is judged to be in a mildly unstable or pseudo-fatigue state. If the slope of the decrease in the median frequency of electromyography is significantly lower than the preset slope threshold and the coefficient of variation is greater than the preset coefficient of variation threshold, the muscle is judged to be in a true fatigue state.

[0037] The anomaly confirmation sub-strategy includes: Feature extraction is performed on the electromyography waveforms to be analyzed that exceed the fatigue threshold. The extracted features include the time-domain RMS features of electromyography, the median frequency and the descent slope features in the frequency domain, the stability features of the signal coefficient of variation, and the difference features of the electromyography response before and after the VR task difficulty adjustment, and a fatigue feature vector is generated. Based on the task interaction mode, the weight or resistance parameters of objects in the VR scene are adjusted downwards to obtain the difference in electromyographic response before and after adjustment, and the fatigue recovery index is calculated. in The degree to which the difficulty of VR tasks has been reduced. This is the root mean square value after the difficulty level has been lowered. The root mean square value after the difficulty level is reduced. If the response exceeds the threshold, it is determined to be pseudo-fatigue, and the parameters are optimized in the third analysis scheme. If the response is below the threshold, it is confirmed to be true fatigue, and the system is forcibly put into rest mode.

[0038] The third analysis scheme includes: The pre-built database contains pre-stored combinations of electromyographic spectral features and corresponding frequency optimization rules, including... wave and The power spectrum ratio of the wave; The spectral characteristics of the waveform to be analyzed are compared with the frequency optimization rules in the database, and then... Wave power variation is used as the criterion. like If the reduction ratio of wave power is greater than the preset high threshold, the electrical stimulation frequency will be increased to stimulate rapid glycolytic fiber. like If the wave power is stable, the reference stimulation frequency is maintained; like When the wave power fluctuation is in the dynamic range, the amplified electromyographic spectrum characteristics are obtained by adjusting the action frequency of the VR task in the task interaction mode, and then compared again.

[0039] Case 1: When the patient is undergoing rehabilitation training, The proportion of wave power to total power dropped from 45% to 30%, exceeding the threshold. The system determined this to be a decrease in muscle recruitment efficiency and automatically adjusted the electrical stimulation frequency from 30Hz to 40Hz. Subsequently, observations were made... When the wave power recovers to 42%, real-time optimization of the electrical stimulation waveform is achieved. Example 2: If the similarity between the waveform features to be analyzed and the database is lower than the threshold, it is determined that the patient has produced a new linkage error, such as a twitching of the corner of the mouth when blinking. The system automatically records the spectral features of the abnormal waveform, generates a new abnormal type and stores it, providing data support for subsequent precise suppression of linkage.

[0040] The working principle and process of this invention are as follows: After the system starts, the electromyography (EMG) acquisition module first enters the resting baseline mode, and obtains the physiological parameters of the target facial muscle groups in the resting state through sampling frequency. After filtering by Butterworth filter and notch filter and conversion by A / D converter, the mean, maximum and minimum values ​​of EMG potentials are recorded to generate the patient's initial rehabilitation baseline report. The analysis and control module obtains the patient's resting and maximum contraction EMG values ​​through the initial threshold setting program, and preliminarily determines the initial dynamic judgment threshold by adding 30% of the maximum contraction value to the resting mean value, thus completing the patient's personalized baseline parameter modeling.

[0041] The system switches from resting baseline mode to task interaction mode. The VR module constructs a virtual rehabilitation scenario and dynamically adjusts the task difficulty according to the patient's rehabilitation status. During the preparation phase of the virtual scenario, the electromyography (EMG) acquisition module enters a monitoring-only, non-interventional scenario, recording EMG signals point by point and calculating the real-time RMS waveform. During the exertion phase, the system enters a monitoring and intervention scenario. The EMG acquisition module acquires the physiological parameters of the patient's facial target muscle groups during operation according to a preset acquisition cycle, extracts the real-time RMS value, coefficient of variation, and EMG spectral characteristics, and transmits them to the analysis and control module.

[0042] After receiving physiological parameters, the analysis and control module compares the routine real-time RMS value with the dynamic threshold. The ratio of the RMS value to the dynamic threshold in the current acquisition cycle is used as the judgment criterion. At the same time, it combines the benchmark electromyographic characteristics and associated stimulation parameters corresponding to different rehabilitation stages in the pre-built database. Based on the comparison results of the previous acquisition cycle, the real-time threshold is adjusted by the dynamic coefficient to obtain the reinforcement feedback threshold. The judgment threshold for the next acquisition cycle is dynamically updated, and the corresponding intervention plan is executed according to the selected strategy.

[0043] When the real-time RMS value is lower than the reinforcement feedback threshold, the first analysis scheme is executed. The system adjusts the electrical stimulation current intensity in a step-by-step manner. If the real-time RMS value remains below the reinforcement feedback threshold, the electrical stimulation intensity is gradually increased. If the real-time RMS value reaches the reinforcement feedback threshold, the current electrical stimulation intensity is maintained. If the coefficient of variation corresponding to the real-time RMS value is greater than the preset coefficient of variation threshold, the second analysis scheme is executed. The analysis and control module judges the muscle fatigue state based on the synergistic law of the slope of the decrease in the median frequency of electromyography and the coefficient of variation in the pre-built database. An abnormality confirmation sub-strategy is executed, fatigue waveform features are extracted to generate a fatigue feature vector, the VR scene task resistance parameter is lowered, and the fatigue recovery index is calculated. If true fatigue is confirmed, the system switches to the resting baseline mode; if false fatigue is confirmed, the system jumps to execute the third analysis scheme. When the real-time RMS value meets the triggering conditions and the stimulation waveform needs to be optimized, the third analysis scheme is executed. The spectral features of the waveform to be analyzed are compared with the electromyographic spectral features in the database. Wave power variation is used as the criterion. If the reduction in wave power exceeds a preset threshold, the electrical stimulation frequency will be increased. If the wave power is stable, it is stored as the baseline feature. If the spectral feature similarity is lower than a preset threshold, the electrode adaptive layout parameters are recalculated and written into the database to optimize the signal recruitment quality.

[0044] Throughout the entire operation, the data recording module fully reproduces the patient's baseline parameter set, records electromyographic signals, electrical stimulation parameters, VR task difficulty, and execution data of each analysis plan, and forms rehabilitation trend assessment results. This provides a basis for comparison for the VR module to dynamically adjust the task difficulty and for the analysis and control module to correct the judgment threshold.

[0045] The embodiments of the present invention have been described in detail above, but the content described is only a preferred embodiment of the present invention and should not be considered as limiting the scope of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of this patent.

Claims

1. A facial paralysis electrical stimulation system based on VR and dynamic facial electromyography threshold feedback, characterized in that, include: A head-mounted helmet, which is equipped with a VR module, an electromyography (EMG) acquisition module, and an analysis and control module; The VR module is used to construct relevant virtual rehabilitation scenarios for different stages of rehabilitation and provide a set of rehabilitation instruction tasks. The electromyography (EMG) acquisition module is used to capture the physiological parameters of the user when executing the rehabilitation instruction set, including real-time RMS value, coefficient of variation, and EMG spectrum. The analysis and control module extracts and compares features of physiological parameters, and performs the following judgments and interventions according to the selected configured strategy; When the real-time RMS value is lower than the dynamic judgment threshold, the first analysis scheme is executed, and the intensity of electrical stimulation is adjusted in stages. When the coefficient of variation is greater than the preset coefficient of variation threshold, the second analysis plan is executed to assess muscle fatigue and protect against fatigue. When the electromyographic spectrum characteristics meet the waveform optimization conditions, the third analysis scheme is executed to optimize the electrical stimulation frequency or electrode parameters.

2. The facial paralysis electrical stimulation system based on VR and dynamic facial electromyography threshold feedback according to claim 1, characterized in that, The analysis and control module is equipped with a database that stores various rehabilitation stage types. Each rehabilitation stage type is pre-matched with: baseline electromyographic characteristics, muscle stability baseline characteristics, electromyographic spectrum baseline characteristics, and associated stimulation safety parameters. The database also has preset threshold adjustment dynamic coefficients that match each rehabilitation stage type. These dynamic coefficients are preset adjustment weight coefficients used to adaptively correct the dynamic judgment threshold for each acquisition cycle under the corresponding rehabilitation stage. The dynamic judgment threshold is generated based on the baseline electromyographic characteristics corresponding to the rehabilitation stage, and is the critical judgment value of the instantaneous muscle contraction intensity under a single sliding sampling window; The preset coefficient of variation threshold is generated based on the muscle stability benchmark features corresponding to the rehabilitation stage, and is the critical judgment value for the degree of fluctuation of the real-time RMS value data corresponding to the continuous sampling time sequence signal. The waveform optimization conditions are generated based on the electromyographic spectrum baseline features corresponding to the rehabilitation stage, and can be triggered if any of the following conditions are met: The real-time RMS value is lower than the dynamic threshold corresponding to the first analysis scheme; The system is in the execution phase of the exception confirmation sub-policy; Electromyography (EMG) using continuously sampled time-series signals in the third analysis scheme The power drop reached the preset threshold.

3. The facial paralysis electrical stimulation system based on VR and dynamic facial electromyography threshold feedback according to claim 2, characterized in that, The first analysis scheme includes: Collect the user's facial electromyography (EMG) values ​​at rest and at maximum voluntary contraction. Based on the numerical range of resting electromyography (EMG) values ​​and maximum voluntary contraction EMG values, and combined with the proportional coefficient matched with the baseline EMG characteristics corresponding to the current rehabilitation stage type, the initial dynamic judgment threshold is calculated. Real-time RMS values ​​are acquired through the electromyography acquisition module according to the preset acquisition cycle, and the working mode of the electromyography acquisition module is switched from resting reference mode to task interaction mode. In task interaction mode, the dynamic judgment threshold is adaptively corrected based on the initial dynamic judgment threshold and the dynamic coefficient, and the dynamic judgment threshold of the current period is obtained iteratively. The baseline parameters after dynamic coefficient correction are reproduced by the data recording module. The baseline parameters are compared with the real-time RMS value according to the selected strategy and the degree of data deviation is calculated. If the real-time RMS value is continuously lower than the dynamic judgment threshold, the intensity of electrical stimulation is increased until the real-time RMS value rises and stabilizes above the dynamic judgment threshold.

4. The facial paralysis electrical stimulation system based on VR and dynamic facial electromyography threshold feedback according to claim 3, characterized in that, When the real-time RMS value of a single sliding sampling window remains below the dynamic judgment threshold, the current level is progressively increased for each preset sampling cycle to achieve step-by-step current increase; until the real-time RMS value rises and remains above the dynamic judgment threshold for a continuous preset number of sampling cycles, the current electrical stimulation current level is locked.

5. The facial paralysis electrical stimulation system based on VR and dynamic facial electromyography threshold feedback according to claim 3, characterized in that, The database also contains pre-built fatigue assessment benchmark rules and preset coefficient of variation thresholds; The second analysis scheme includes: Continuous sampling time-series signals are acquired, and multiple waveform data segments are extracted through a sliding window. The effective value of the entire domain is calculated on a single time-domain waveform to generate an RMS segment feature quantity that represents the muscle contraction state of that segment. The fluctuation pattern of the electromyographic waveform is characterized by the multi-frame RMS segment feature quantity corresponding to multiple waveform segments. The coefficient of variation of electromyography waveform is calculated based on the feature quantity of multi-frame RMS segments, and the coefficient of variation of electromyography waveform is compared with the preset coefficient of variation threshold. The analysis is based on a comprehensive comparison and judgment of the fluctuation patterns of the electromyographic waveform to be analyzed and the fatigue assessment benchmark patterns. If the coefficient of variation of the electromyographic waveform is lower than the preset coefficient of variation threshold and does not conform to the fatigue assessment benchmark, it is judged as normal fluctuation. If the coefficient of variation of the electromyographic waveform is not lower than the preset coefficient of variation threshold and conforms to the fatigue assessment benchmark, it is initially determined to be muscle fatigue, and the abnormal confirmation sub-strategy is executed for verification. The anomaly confirmation sub-strategy is used to perform a secondary verification of the initial fatigue assessment results, distinguishing between temporary fluctuations and true, persistent muscle fatigue.

6. The facial paralysis electrical stimulation system based on VR and dynamic facial electromyography threshold feedback according to claim 5, characterized in that, The anomaly confirmation sub-strategy includes: Feature extraction is performed on the electromyographic waveforms to be analyzed that exceed the fatigue threshold to generate fatigue feature vectors; Based on the preset task interaction mode, the difficulty of the rehabilitation instruction task set in the VR scene is adjusted down level by level; Continuous sampling time-series electromyographic signals were collected before and after difficulty adjustment, the difference in electromyographic response before and after adjustment was calculated, and the fatigue recovery index was obtained by quantification. The fatigue recovery index is compared with a preset response threshold, which is generated based on the muscle and nerve response benchmark features corresponding to the rehabilitation stage. If the fatigue recovery index is lower than the preset response threshold, it is determined to be true fatigue; the working mode of the electromyography acquisition module is switched to resting reference mode. If the fatigue recovery index is not lower than the preset response threshold, it is determined to be a temporary electromyographic fluctuation, and the third analysis plan is executed.

7. The facial paralysis electrical stimulation system based on VR and dynamic facial electromyography threshold feedback according to claim 6, characterized in that, The database also contains a first preset threshold, a second preset threshold, and a preset stable threshold, each of which is generated based on the facial electromyography spectrum benchmark features corresponding to the rehabilitation stage. The database also stores combinations of electromyographic spectral features and corresponding frequency optimization rules, as well as electrical stimulation optimization parameters associated with each combination of electromyographic spectral features. The third analysis scheme includes: The electromyography (EMG) acquisition module acquires continuously sampled time-series EMG waveforms to be analyzed. This module reuses the signal acquisition hardware of the second analysis scheme and performs spectral decomposition on the acquired waveforms to extract the EMG signals. Waves are the object of analysis; Extract to be analyzed The waveform characteristics of the wave are compared with those of electromyography (EMG) spectrum, and the waveform characteristics are combined with those of the EMG spectrum. The change in wave power is used as the criterion for judgment; like If the reduction ratio of wave power is greater than the first preset threshold, then the electrical stimulation frequency increase operation is performed; like If the wave power fluctuation within the preset monitoring period is within the preset stability threshold, then the spectral feature is used as the reference feature and stored. like The wave power fluctuation is in the dynamic range. When entering the task interaction mode, the amplified electromyographic spectrum features are obtained by adjusting the action frequency of the VR task. These features are then compared with the electromyographic spectrum feature combinations in the database. If the similarity of all feature combinations is lower than the second preset threshold, the electrode adaptive layout parameters are recalculated and written into the database.

8. The facial paralysis electrical stimulation system based on VR and dynamic facial electromyography threshold feedback according to claim 7, characterized in that, The electrode adaptive layout parameters are recalculated based on abnormal spectral features with similarity below a preset range, and are used to optimize the signal acquisition quality of subsequent acquisition cycles.

9. The facial paralysis electrical stimulation system based on VR and dynamic facial electromyography threshold feedback according to claim 7, characterized in that, The resting reference mode includes: The electromyography (EMG) acquisition module acquires actual physiological parameters at rest according to a preset sampling frequency, and summarizes and records the average values ​​of the parameters within a preset time period to construct the baseline characteristics of the patient's muscle and nerve response at rest.

10. The facial paralysis electrical stimulation system based on VR and dynamic facial electromyography threshold feedback according to claim 7, characterized in that, The task interaction mode includes a monitoring-only-no-intervention scenario and a monitoring-and-intervention scenario. The monitoring-only-no-intervention scenario includes: the electromyography acquisition module records the electromyography signals in the preparation stage of the VR task point by point and calculates the real-time waveform of the RMS value in real time, thereby establishing a muscle reference waveform in the preparation stage without intervention. The monitoring and intervention scenario includes: when the triggering condition is met, an electrical stimulation operation is initiated. The electrical stimulation operation adjusts the magnitude or frequency of the electrical stimulation current and simultaneously performs periodic electromyography (EMG) acquisition during the exertion phase of the VR task, records the parameters point by point, and records the real-time contraction waveform after the electrical stimulation intervention. This provides waveform basis for the iterative correction of electrode adaptive layout parameters and electrical stimulation optimization parameters, thereby realizing dynamic feedback adjustment of stimulation parameters.

Images