Active rehabilitation training method and system based on brain-computer interaction and myoelectricity

Abstract

The invention provides an active rehabilitation training method and system based on brain-computer interaction and myoelectricity. The method comprises the following steps: S1, performing electroencephalogram evaluation and myoelectricity evaluation before training to obtain an optimal electroencephalogram threshold value and an optimal myoelectricity threshold value; s2, acquiring real-time electroencephalogram data and real-time myoelectricity data after training is started; s3, inputting the real-time myoelectricity data and the optimal myoelectricity threshold value into the myoelectricity direction / action evaluation model to calculate scores in all directions, and taking the direction with the maximum score as the motion direction; s4, the electroencephalogram motion intention scoring model calculates a real-time electroencephalogram score according to the real-time electroencephalogram data and the optimal electroencephalogram threshold value, whether the real-time electroencephalogram score is larger than a preset value or not is judged, if yes, the multi-degree-of-freedom rehabilitation training module is driven to move in the motion direction, the step S2 is executed, and if not, training is ended; the motion intention of the training object is recognized through the electroencephalogram, so that the limbs are assisted to do corresponding motion, and multi-degree-of-freedom active rehabilitation training is carried out.

Description

Technical Field

[0001] This invention relates to the field of rehabilitation training technology, and in particular to an active rehabilitation training method and system based on brain-computer interface and electromyography. Background Technology

[0002] Spinal cord injury (SCI) refers to a serious condition caused by trauma, disease, or degenerative changes that damages the structure or function of the spinal cord, resulting in partial or complete loss of motor, sensory, and autonomic functions below the level of injury. Complete high-level spinal cord injury, in particular, is the most severe type of SCI, typically referring to a complete injury to the cervical spine (C1-C4 segments) (ASIA grade A), leading to the loss of all motor, sensory, and autonomic functions below the level of injury. Current rehabilitation methods include manual therapy, passive rehabilitation equipment, and brain-computer interfaces (BCIs). Manual therapy and passive rehabilitation equipment are both passive limb training methods with limited patient participation. For complete high-level spinal cord injuries, active participation is completely impossible, thus limiting rehabilitation effectiveness. Existing BCI training methods, due to limited degrees of freedom of control, struggle to perform complex rehabilitation movements. Referring to the robotic arm control method, control device, electronic device, and storage medium disclosed in Chinese Patent Publication No. CN117484489A, the robotic arm control method includes: acquiring the user's eye movement signals and facial electromyography (EMG) signals; determining the target position of a target object based on the eye movement signals, wherein the target position represents the position of the target object in the robotic arm coordinate system; determining the user's activity intention based on the facial EMG signals; and controlling the robotic arm to complete a target task based on the target position of the target object and the activity intention, wherein the target task is to reach the target position and grasp or release the target object. The above technical solution only uses facial EMG signals as the basis for motion control and can only be used for daily living assistance; it cannot identify the patient's active movement intention for multi-degree-of-freedom active rehabilitation training. Summary of the Invention

[0003] This invention addresses the problem that spinal cord injury patients are unable to perform multi-degree-of-freedom active rehabilitation training. It proposes an active rehabilitation training method and system based on brain-computer interface and electromyography (EMG). By recognizing the movement intentions of the training subject through EEG, it assists the limbs in performing corresponding movements, enabling multi-degree-of-freedom active rehabilitation training.

[0004] To achieve the above objectives, the following technical solution is proposed: An active rehabilitation training method based on brain-computer interface and electromyography includes: S1. Before training, EEG and EMG assessments are performed to obtain the optimal EEG and EMG thresholds. S2, after training begins, acquires real-time EEG data and real-time EMG data; S3: Input real-time electromyography data and the optimal electromyography threshold into the electromyography direction / movement assessment model to calculate the score for each direction, and take the direction with the highest score as the direction of motion or synthesize the scores into a direction vector as the direction of motion. S4: The EEG motor intention scoring model calculates the real-time EEG score based on real-time EEG data and the optimal EEG threshold. It then determines whether the real-time EEG score is greater than the preset value. If so, it drives the multi-degree-of-freedom rehabilitation training module to move in the direction of movement and returns to S2. If not, it continues to calculate the electromyography and EEG scores in real time until the training ends.

[0005] This invention combines brain-computer interfaces, facial electromyography (EMG), and multi-degree-of-freedom rehabilitation devices. By pre-collecting EEG and facial EMG information of the trainee's imagined limb movements, an EEG and facial EMG model is established. During training, the movement pattern or direction is determined by recognizing facial EMG, and the trainee's movement intention is recognized by EEG. When the trainee wants to move the corresponding limb in a specified pattern or direction, robots, electrical stimulation, magnetic stimulators, etc. are used to drive / stimulate the nerves and muscles of the relevant limb, thereby assisting the limb to perform the corresponding movement and conducting multi-degree-of-freedom active rehabilitation training.

[0006] Preferably, the optimal EEG threshold and optimal EMG threshold are manually adjusted according to the training intensity.

[0007] The present invention S1 further includes the following steps: determining whether the optimal EEG threshold and the optimal EMG threshold need to be adjusted; if adjustment is needed, manually adjusting the thresholds; the purpose of manually adjusting the thresholds is to control the intensity of training.

[0008] Preferably, the optimal EEG threshold is obtained by decoding the EEG information collected from the training subject when imagining / attempting to move limbs and when at rest, to distinguish whether the training subject is in a state of imagining limb movement or a state of rest; the optimal EMG threshold is obtained by decoding the EMG information collected from the training subject when imagining / attempting to move limbs and when at rest, to distinguish whether the training subject is in a state of imagining limb movement or a state of rest.

[0009] Preferably, the process for obtaining the optimal electromyographic threshold is as follows: Calculate the threshold values ​​for each direction: Set the characteristic coefficients for muscle contraction and rest in each direction; Acquire resting electromyographic characteristics and electromyographic characteristics of muscle contraction in various directions at a set data length; Calculate the mean value of electromyographic characteristics of muscle contraction in each direction under a set data length; Threshold for each direction = Characteristic coefficient of muscle contraction in each direction × Mean of electromyographic characteristics of muscle contraction in each direction under the set data length - Characteristic coefficient at rest × Electromyographic characteristics at rest; The optimal electromyography (EMG) threshold is obtained by extracting EMG frequency band features from the thresholds in each direction.

[0010] Preferably, the process for obtaining the optimal EEG threshold is as follows: Optimal EEG threshold = (1 + feature coefficient) × preset standard threshold, where the feature coefficient is greater than -1; Set the coefficients for each different EEG channel and the first coefficients for various EEG characteristics during imagination / attempt to move limbs and in the resting state; Acquire various EEG characteristics at a set data length during imagining / attempting to move limbs and at rest. Calculate the first cumulative value of the difference between the EEG characteristics of various types of EEG features when imagining / attempting to move limbs and the EEG characteristics at rest, under a set data length; and the second cumulative value of the sum of the EEG characteristics of various types of EEG features when imagining / attempting to move limbs and the EEG characteristics at rest, under a set data length. The ratio of the first cumulative value to the second cumulative value of each type of EEG feature is calculated and then multiplied by the first coefficient of each type of EEG feature to obtain the first reference value of each type of EEG feature; The sum of the first reference values ​​of various EEG characteristics is multiplied by the coefficients of each different EEG channel, and the cumulative value of all EEG channels is used as the characteristic coefficient.

[0011] Preferably, the specific process by which the electromyographic direction / movement assessment model calculates the score for each direction is as follows: Preset auxiliary scores and characteristic coefficients of electromyography in each direction; Calculate the mean value of the electromyographic characteristics of muscles in each direction during contraction, within a set data length; Set the manual adjustment parameters for each direction and obtain the threshold values ​​for each direction; Muscle contraction score in each direction = preset auxiliary score + characteristic coefficient of electromyography in each direction × (mean of electromyographic characteristics of muscle contraction in each direction under the set data length - manual adjustment parameter of each direction × score threshold of each direction).

[0012] Preferably, the S4 EEG motor intention scoring model calculates the real-time EEG score based on real-time EEG data and the optimal EEG threshold, specifically including the following steps: Preset auxiliary scores and characteristic coefficients of various EEG patterns; The first calculated value is the product of the cumulative value of the EEG characteristics of various types of EEG when imagining / attempting to move limbs over a set data length and the characteristic coefficients of various types of EEG. The first mean value is obtained by summing the first calculated values ​​of various EEG types and then dividing by the set data length. Obtain the coefficients and number of brainwave channels for each different brainwave channel; Obtain manually adjustable parameters and optimal EEG thresholds; The second calculated value is obtained by subtracting the product of the coefficients of each different EEG channel and the first mean from the product of the manually adjusted parameters and the optimal EEG threshold. The real-time EEG score is equal to the preset auxiliary score plus the second calculated value, which is the sum of the number of EEG channels.

[0013] An active rehabilitation training system based on brain-computer interface and electromyography (EMG) employs the aforementioned active rehabilitation training method based on brain-computer interface and EMG. The system includes a control module electrically connected to an EEG assessment module, an EEG motion prediction module, an EMG assessment module, and an EMG prediction module. The EEG assessment module and the EEG motion prediction module are electrically connected to an EEG acquisition module. The EEG assessment module and the EEG motion prediction module are electrically connected, as are the EMG assessment module and the EMG prediction module. The EMG assessment module and the EMG prediction module are also electrically connected to the EMG acquisition module. The control module is electrically connected to a multi-degree-of-freedom rehabilitation training module.

[0014] Preferably, the control module is electrically connected to an audiovisual feedback module. The audiovisual feedback module of this invention includes a display, projector, audio system, VR glasses, etc.; wherein visual feedback includes images of limb movements provided by a computer, examples of models, expert guidance, etc., or movements presented using virtual reality technology; auditory feedback includes movement voice guidance provided by the system, start / stop prompts, success or failure feedback, etc.

[0015] Preferably, the EEG assessment module includes an EEG threshold calculation model; the EEG motion prediction module includes an EEG motion intention scoring model; the EMG assessment module includes an EMG threshold calculation model; and the EMG prediction module includes an EMG direction / movement assessment model.

[0016] The beneficial effects of this invention are as follows: During training, electromyography (EMG) accurately identifies the direction or movement the user wants to move, and brain-computer interface actively identifies the movement intention of the training subject when imagining / attempting to move the limbs. Furthermore, the multi-degree-of-freedom rehabilitation training module can operate after reaching the optimal EEG threshold. This not only enables the user to train complex movements with multiple degrees of freedom, but also allows the brain to actively participate in imagining the training movements when the user starts training. At the same time, the multi-degree-of-freedom rehabilitation training module drives limb movement, achieving higher intensity activation and movement of the limbs. This allows the brain and limbs to be activated synchronously, realizing convenient and rapid active movement rehabilitation and recovery training, and accelerating the rehabilitation process. Attached Figure Description

[0017] Figure 1 This is a flowchart of the method of the present invention.

[0018] Figure 2 This is a simplified structural diagram of the system of the present invention.

[0019] The module includes: 1. EEG acquisition module; 2. EEG assessment module; 3. EEG motion prediction module; 4. EMG acquisition module; 5. EMG assessment module; 6. EMG prediction module; 7. Control module; 8. Audiovisual feedback module; 9. Multi-degree-of-freedom rehabilitation training module. Detailed Implementation

[0020] Example 1: This embodiment proposes an active rehabilitation training method based on brain-computer interface and electromyography, referencing... Figure 1 This includes the following steps: S1. Before training, electromyography (EMG) and electroencephalography (EEG) assessments are performed to obtain the optimal EMG and EEG thresholds for reference calculation. EMG assessment: By collecting EMG information from the training subject when imagining / attempting to move a limb and at rest, the most suitable optimal EMG threshold is decoded to distinguish whether the training subject is in a state of imagining limb movement or at rest. EEG assessment: By collecting EEG information from the training subject when imagining / attempting to move a limb and at rest, the most suitable optimal EEG threshold is decoded to distinguish whether the training subject is in a state of imagining limb movement or at rest.

[0021] The present invention S1 further includes the following steps: determining whether the optimal EEG threshold and the optimal EMG threshold need to be adjusted; if adjustment is needed, manually adjusting the thresholds; the purpose of manually adjusting the thresholds is to control the intensity of training.

[0022] The optimal electromyography (EMG) threshold of this invention is calculated by an EMG threshold calculation model, which is a matrix, specifically: T EMG =[T up T down T left T right T front T back ] Among them, T EMG The optimal electromyographic threshold; T up T down T left T right T front T back The threshold values ​​for up, down, left, right, front, and back; a up a down aleft a right a front a back a represents the characteristic coefficients of muscle contraction in all directions (up, down, left, right, forward, backward), and lateral. rest The characteristic coefficient during rest; F EMG-up F EMG-down F EMG-left F EMG-right F EMG-front F EMG-back F represents the electromyographic characteristics of muscles contracting in all directions (up, down, left, right, forward, backward). EMG-rest The values ​​represent the resting electromyographic (EMG) characteristics; m represents the data length. EMG frequency band characteristics include: mean, standard deviation, average, standard deviation, power spectral density, and asymmetry coefficient.

[0023] The optimal EEG threshold described in this invention is calculated using an EEG threshold calculation model, which is as follows: T EEG = (1+ε)*T std Among them, F EEG The optimal EEG threshold; ε is the characteristic coefficient, ε > -1; T std The preset standard threshold; h i Here, represents the coefficients for each different EEG channel, and ch represents the number of EEG channels. k1 / k2 / k3 / k4 / k5 are coefficients representing the EEG characteristics of delta waves, theta waves, alpha waves, beta waves, and SMR waves during imagined / attempted limb movements and at rest; the sum of these five coefficients is 1. F δ-image For the characteristics of delta waves in brainwaves when imagining / attempting to move limbs, F θ-image For the theta wave characteristics of brainwaves when imagining / attempting to move limbs, F α-image For the characteristics of alpha waves in brainwaves when imagining / attempting to move limbs, F β-image For the characteristics of beta waves in brainwaves when imagining / attempting to move limbs, F SMR-rest For the characteristics of SMR brain waves when imagining / attempting to move limbs, F δ-rest The characteristics of resting delta waves in brainwaves, F θ-rest The theta wave characteristics of resting brainwaves, F α-rest The resting EEG alpha wave characteristics, F β-rest The resting EEG beta wave characteristics, F SMR-rest The resting SMR wave characteristics are shown; m is the data length.

[0024] It should be noted that in this invention, when the EEG acquisition module collects EEG information, the signal acquisition of the imagined / attempted limb movement state or rest state will be continuously collected for a certain period of time. Therefore, there will be a certain data length, so as to obtain the feature information of different states more completely and avoid omissions.

[0025] S2, once training has officially begun, acquire real-time electromyography (EMG) and electroencephalography (EEG) data; S3, perform electromyography prediction: input the optimal electromyography threshold and real-time electromyography data into the electromyography direction / movement evaluation model, the electromyography direction / movement evaluation model calculates the score of the probability of movement in each direction, selects the direction with the highest score of the probability of movement in each direction as the direction of movement, or synthesizes the scores into a direction vector as the direction of movement. The electromyographic direction / movement assessment model of this invention is as follows: Among them, Score EMG This is a score matrix for electromyographic direction / movement assessment; EMG-up score EMG-down score EMG-left score EMG-right score EMG-front score EMG-back Scoring is based on muscle contraction in all directions (up, down, left, right, forward, backward); C EMG The preset auxiliary scores are: k1, k2, k3, k4, k5, and k6 are the characteristic coefficients of electromyography for each direction (up, down, left, right, front, and back); F EMG-up F EMG-down F EMG-left F EMG-right F EMG-front F EMG-back The characteristics of electromyography (EMG) in all directions (up, down, left, right, front, and back); T up T down T left T right T front T back The threshold values ​​for up, down, left, right, front, and back; k manual The therapist can manually adjust the parameters during use, increasing or decreasing the threshold to increase or decrease the difficulty; m is the data length.

[0026] h i Here, represents the coefficients for each different electromyographic channel, and ch represents the number of electromyographic channels. The parameters were the same during electromyography assessment; F EMG For the electromyographic characteristics of imagining / attempting to move a limb, T EMG is the optimal electromyography threshold, and m is the data length.

[0027] It should be noted that the preset auxiliary score can be set according to actual needs, and this auxiliary score can be used as a preset score to judge the imagination state.

[0028] S4, EEG prediction: Input the optimal EEG threshold and real-time EEG data into the EEG motor intention scoring model for calculation to obtain a real-time EEG score. Determine if the real-time EEG score is greater than a preset value. If not, and the termination condition is not met, continue to calculate the EMG and EEG scores in real time. If yes, drive the multi-degree-of-freedom rehabilitation training module to move in the direction of movement obtained in S3, and return to S2. Termination condition: The real-time EEG score is less than a preset value or the training time is reached within the set duration.

[0029] This invention combines brain-computer interfaces, facial electromyography (EMG), and multi-degree-of-freedom rehabilitation devices. By pre-collecting EEG and facial EMG information of the trainee's imagined limb movements, an EEG and facial EMG model is established. During training, the movement pattern or direction is determined by recognizing facial EMG, and the trainee's movement intention is recognized by EEG. When the trainee wants to move the corresponding limb in a specified pattern or direction, robots, electrical stimulation, magnetic stimulators, etc. are used to drive / stimulate the nerves and muscles of the relevant limb, thereby assisting the limb to perform the corresponding movement and conducting multi-degree-of-freedom active rehabilitation training.

[0030] The real-time EEG score of this invention is calculated using an EEG motor intention scoring model, which is as follows: Among them, Score EEG For brainwave motor intention scoring; C EEG The preset auxiliary score; h i Here, represents the coefficients for each different EEG channel, and ch represents the number of EEG channels. The parameters are the same during EEG assessment; k0 is the score conversion coefficient; k1 / k2 / k3 / k4 / k5 are the coefficients of the EEG characteristics of delta waves, theta waves, alpha waves, beta waves, and SMR waves in the EEG during imagination / attempt to move limbs and in the resting state, and the sum of the five is 1; k manual Therapists can manually adjust parameters during use, increasing or decreasing the threshold to adjust the difficulty level; F δ For the characteristics of delta waves in brainwaves when imagining / attempting to move limbs, F θ For the theta wave characteristics of brainwaves when imagining / attempting to move limbs, F α For the characteristics of alpha waves in brainwaves when imagining / attempting to move limbs, F β For the characteristics of beta waves in brainwaves when imagining / attempting to move limbs, F SMRThe image shows the SMR wave characteristics of EEG during imagined / attempted limb movement, with m representing the data length. A preset auxiliary score can be set according to actual needs and can be used as a preset score to assess the imagined state. The theta band, with a frequency range of 4-8 Hz, is found to be more prevalent in the brains of infants and adults and adolescents during drowsiness (or early sleep); it also appears when the brain is idle or during meditation; theta wave energy is positively correlated with the intensity of chronic pain. The alpha band, with a frequency range of 8-13 Hz, is the most prominent wave in rhythmic EEG, typically appearing at the back of the head, present on both sides, with a higher amplitude on the dominant side; alpha band signals can be detected in EEG scans of the occipital lobe when a conscious person is relaxed or has their eyes closed. Alpha energy is negatively correlated with pain intensity. In BCI applications, a specific beta band signal is called the SMR band (12-15 Hz), whose appearance is often associated with resting states, especially during sleep and wakefulness. In patients with neuropathic pain, there appears to be a decrease during the training phase. The trainee contracts corresponding muscles in the front, back, left, right, up, and down directions. The system calculates the desired direction or movement based on electromyographic direction / motor score. When one or more direction scores exceed a threshold, the direction with the highest score is selected as the user's movement direction. The system also scores motor intention using an EEG motor intention scoring model. The calculated imagined state score... EEG When the preset score is exceeded, the control module issues a control command, instructing the multi-degree-of-freedom rehabilitation training module to move according to the selected direction. The audiovisual feedback module displays the electromyography and electroencephalography characteristics of the trainee in real time, while also displaying animations of muscle contraction and relaxation, and providing auditory cues to guide the trainee to perform training better.

[0031] This embodiment also proposes an active rehabilitation training system based on brain-computer interface and electromyography, referencing... Figure 2The system includes an EEG acquisition module 1, which acquires EEG signals and inputs them to an EEG assessment module 2 and an EEG motion prediction module 3. During EEG assessment, the EEG assessment module 2 evaluates the characteristic signals related to brain movement, generates and stores an EEG model, and transmits the results to the EEG motion prediction module 3. During training, the EEG motion prediction module 3 predicts the user's movement intention based on the real-time EEG signals and the EEG model generated by the EEG assessment module, and sends the results to a control module 7. The system also includes an EMG acquisition module 4, which acquires EMG signals and inputs them to an EMG assessment module 5 and an EMG prediction module 6. During EMG assessment, the EMG assessment module 5 evaluates the characteristic signals of EMG, generates and stores an EMG model, and transmits the results to the EMG prediction module 7. During training, the EMG prediction module 7 predicts the user's movement direction or action based on the real-time EMG signals and the EMG model generated by the EMG assessment module 5, and sends the results to the control module 7. The EEG motion prediction module 3 predicts whether the user is ready to start exercising based on the real-time EEG signals and the EEG model generated by the EEG assessment module 2, and sends the results to the control module 7. The control module 7 controls the operation of the multi-degree-of-freedom rehabilitation training module 9 based on the results of the electromyography prediction module 6 and the electroencephalography motion prediction module 3.

[0032] The system works as follows: it collects and evaluates EEG and EMG signals; it evaluates the feature signals and stores the model, then transmits the results to the prediction module; it begins training by selecting different directions or movements based on EMG, and activates the device based on EEG to start the movement, providing real-time feedback to the user; the user switches directions or movements, and then training continues until the training ends.

[0033] The EEG acquisition module 1 includes an EEG cap (containing electrodes) and an EEG-to-analog converter module, which converts the acquired EEG signals into digital signals and transmits them to the system. The number of electrodes on the EEG cap can be 2, 4, 8, or 16 leads, but 16 leads are common. The electrodes can be dry, semi-dry, or wet electrodes, but wet electrodes are common. The EEG-to-analog converter module uses conventional technology from existing EEG machines to convert analog signals into digital signals for transmission, while reducing signal noise and improving the signal-to-noise ratio. Simultaneously, signal preprocessing is performed using common EEG signal analysis and processing methods such as Fast Fourier Transform, Butterworth filter, and Chebyshev filter.

[0034] The electromyography (EMG) acquisition and decoding module 5 includes EMG electrodes and an EMG-to-analog converter module. This module converts the acquired EMG signals into digital signals and sends them to the system. The number of EMG electrodes can be 6 or 8 leads, but 6 leads are common. The electrodes can be dry, semi-dry, or wet, but wet electrodes are common. The EMG-to-analog converter module uses conventional technology from existing EMG machines to convert analog signals into digital signals for transmission, while reducing signal noise and improving the signal-to-noise ratio. Simultaneously, signal preprocessing is performed using common EMG signal analysis and processing methods such as Fast Fourier Transform, Butterworth filter, and Chebyshev filter.

[0035] The control module 9 receives information from the EEG assessment module 2, EEG motion prediction module 3, EMG assessment module 5, and EMG prediction module 6. After processing by the control module 7, the collected information is output as control signals, activating the multi-degree-of-freedom rehabilitation training module 9. The control module can be a computer host, control motherboard, etc., and can run operating systems such as Windows, Linux, and Android.

[0036] Multi-degree-of-freedom rehabilitation training module 9 is a treatment method that uses rehabilitation equipment to act on the peripheral neuromuscular system; existing conventional lower limb exoskeleton robots, upper limb exoskeleton robots, functional electrical stimulation, etc. can be selected for motor rehabilitation training.

[0037] The electromyography assessment module 5 stores the electromyography threshold calculation model; it obtains the electromyography frequency band feature data of the required frequency band based on the collected electromyography signals; Based on the collected electromyography (EMG) data over a period of time, preprocessing was performed, including notch filtering at 50 Hz power frequency and bandpass filtering from 55 to 200 Hz, to obtain the preprocessed data. The characteristic data F of each EMG frequency band was then calculated. EMG Electromyographic frequency band characteristics include: mean, standard deviation, power spectral density, asymmetry coefficient, etc.

[0038] The training subjects' facial / neck muscles correspond to different movement directions or actions, such as: frontalis muscle – upper; mentalis muscle – lower; left masseter muscle – left; right masseter muscle – right; left platysma muscle – anterior; right platysma muscle – posterior. Electromyographic (EMG) signals from various muscle groups are extracted when the training subjects perform muscle contractions according to prompts. These signals are then substituted into an EMG threshold calculation model to calculate the optimal EMG threshold T. EMG .

[0039] The electromyography (EMG) assessment module decodes the optimal EMG threshold by collecting EMG information from the training subject when imagining / attempting to move a limb and when at rest, in order to distinguish whether the training subject is in an imagining limb movement state or a resting state, and establishes an imagination model for use by the EMG motion prediction module.

[0040] In this invention, when the electromyography acquisition module collects EEG information, the signals collected during muscle contraction or resting state in the up, down, left, right, front, and back are continuously collected for a certain period of time. Therefore, there will be a certain data length, which can obtain the characteristics of different states more completely and avoid omissions.

[0041] During the EEG assessment, the electromyography (EMG) assessment module calculates and updates EMG characteristics in real time, providing feedback to the trainee through the audiovisual feedback module's display in the form of line graphs, energy maps, and games. When prompted by the system to contract the muscles in all directions (up, down, left, right, forward, backward), the trainee contracts these muscles; when prompted to relax, the trainee relaxes their body and does not visualize movement. The assessment ends when the time is up or the trainee feels they have mastered the technique. After the assessment, the trainee's EMG characteristics and the optimal EMG threshold provided by the system can be viewed. The trainee can use the system's optimal threshold, manually adjust the threshold, or recommend a second assessment.

[0042] The electromyography (EMG) prediction module 6 stores an EMG direction / movement assessment model; it acquires EMG frequency band characteristic data for different frequency bands based on the collected EMG signals; based on the EMG data collected over a period of time, and distinguishing different types of EEG frequency bands, it performs preprocessing including notch filtering at 50Hz power frequency and bandpass filtering from 55-200Hz, obtains the preprocessed data, and calculates the EMG frequency band characteristic data F for each EMG signal. EMG The electromyographic frequency band characteristics include: mean, standard deviation, power spectral density, asymmetry coefficient, etc. These, along with the optimal electromyographic threshold obtained from the electromyographic assessment module, are input into the electromyographic direction / movement assessment model to calculate the training subject's scores in each direction. The direction with the highest score is taken as the movement direction.

[0043] The EEG assessment module 2 stores the EEG threshold calculation model; it acquires EEG frequency band characteristic data for different frequency bands based on the collected EEG signals; based on the EEG data collected over a period of time, and distinguishing them according to different types of EEG frequency bands, it performs preprocessing including notch filtering at 50Hz power frequency and bandpass filtering from 0.5-45Hz, obtains the preprocessed data, and calculates the EEG frequency band characteristic data F for each EEG frequency band. (δ / θ / α / β / SMR) EEG frequency band characteristics include: mean, standard deviation, power spectral density, asymmetry coefficient, etc. The imaginary state characteristics F of the training subject when imagining / attempting to move a limb are also considered. EEG-image and the state characteristics F during rest EEG-rest Substituting these values ​​into the EEG threshold calculation model, the optimal EEG threshold T for each EEG frequency band is calculated. EEG .

[0044] The EEG assessment module decodes the optimal EEG threshold by collecting EEG information from the training subject when imagining / attempting to move limbs and when at rest, in order to distinguish whether the training subject is in an imagining limb movement state or a resting state, and establishes an imagination model for use by the EEG motion prediction module.

[0045] The evaluation process will use the characteristic data of each frequency band F (δ / θ / α / β / SMR) The data is displayed in real-time as a line graph to both the trainee and the operator. When prompted by the system to imagine / attempt limb movement, the trainee performs the imagining / attempt; when prompted to rest, the trainee remains in a resting state. After the evaluation, the system calculates the optimal EEG threshold T based on the differences in EEG characteristics between the trainee's imagined / attempted movement state and relaxed state. EEG .

[0046] During the EEG assessment, the system calculates and updates the EEG characteristics of the imagined / attempted state in real time, providing feedback to the trainee through the audiovisual feedback module's display in the form of line graphs, energy maps, games, etc. When prompted by the system to imagine / attempt limb movements, the trainee imagines / attempts limb movements; when prompted by the system, the trainee relaxes their body and does not imagine movements. The trainee executes the movements through imagination or by using different mental strategies. The assessment ends when the time is up or the trainee believes they have mastered the skills. After the assessment, the trainee's EEG characteristics and the optimal EEG thresholds provided by the system can be viewed. The trainee can use the optimal EEG thresholds provided by the system or manually adjust the thresholds, or recommend that the trainee undergo another assessment.

[0047] The EEG motion prediction module 3 stores the EEG motion intention scoring model; it acquires EEG frequency band feature data of different frequency bands based on the collected EEG signals; based on the EEG data collected over a period of time, it distinguishes different types of EEG frequency bands, and then performs preprocessing including notch filtering at 50Hz power frequency and bandpass filtering from 0.5-45Hz to obtain the preprocessed data (data), and calculates the EEG frequency band feature data F for each EEG frequency band. (δ / θ / α / β / SMR) The EEG frequency band features include: mean, standard deviation, average, standard deviation, power spectral density, asymmetry coefficient, etc. These, along with the optimal EEG threshold obtained from the EEG assessment module, are input into the EEG motor intention scoring model to calculate the trainee's EEG motor intention score. The EEG motor prediction module decodes the EEG features of the trainee when imagining / attempting to move their limbs in real time, compares them with the optimal EEG threshold obtained from the EEG assessment module, and transmits the comparison result to the control module.

[0048] The specific implementation of the brain-computer interface and facial / neck electromyography rehabilitation training system of this invention is as follows: The therapist puts on the EEG cap for the patient and attaches the corresponding electromyography electrodes to the corresponding facial / neck muscles. The patient sits or lies comfortably in a chair or on a treatment bed. The therapist activates the electromyography assessment unit of the system. The system prompts the user to contract and rest the corresponding muscles. The patient tries accordingly. After the trial, the system generates an assessment report and recommends the optimal electromyography threshold. The therapist activates the EEG assessment unit of the system. The system prompts the user to imagine / attempt to move the limbs and rest. The patient tries accordingly. After the trial, the system generates an assessment report and recommends the optimal EEG threshold. The training mode is then activated. After training begins, the patient actively contracts the facial / neck muscles. The system calculates the scores for each direction of the training subject based on real-time electromyography and the electromyography direction / movement assessment model. The direction with the highest score is taken as the direction of movement. The patient then imagines / attempts to move the limbs. When the EEG movement intention score is... EEG When the preset score is exceeded, the multi-degree-of-freedom rehabilitation training module acts on the moving limbs, while receiving real-time feedback from the audiovisual feedback module's display and speakers. When the patient wants to move in other directions, they actively contract their facial / neck muscles to switch the direction of movement, and then imagine / attempt to move the limbs, performing active rehabilitation training until the training ends.

[0049] The specific embodiments described herein are merely illustrative examples illustrating the spirit of the invention. The above embodiments only express several implementation methods of this application, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art to which this application pertains can make various modifications or additions to the described specific embodiments or use similar methods to replace them, but without departing from the spirit of this application or exceeding the scope defined by the appended claims. For those skilled in the art, multiple variations and improvements can be made without departing from the concept of this application. Therefore, the scope of protection of this application should be determined by the appended claims.

[0050] Example 2: This embodiment improves upon Embodiment 1 by adding an audiovisual feedback module 8 and proposing an active rehabilitation training method based on brain-computer interface and electromyography (EMG). (Refer to...) Figure 1 This includes the following steps: S1. Before training, electromyography (EMG) and electroencephalography (EEG) assessments are performed to obtain the optimal EMG and EEG thresholds for reference calculation. EMG assessment: By collecting EMG information from the training subject when imagining / attempting to move a limb and at rest, the most suitable optimal EMG threshold is decoded to distinguish whether the training subject is in a state of imagining limb movement or at rest. EEG assessment: By collecting EEG information from the training subject when imagining / attempting to move a limb and at rest, the most suitable optimal EEG threshold is decoded to distinguish whether the training subject is in a state of imagining limb movement or at rest.

[0051] The present invention S1 further includes the following steps: determining whether the optimal EEG threshold and the optimal EMG threshold need to be adjusted; if adjustment is needed, manually adjusting the thresholds; the purpose of manually adjusting the thresholds is to control the intensity of training.

[0052] The optimal electromyography (EMG) threshold of this invention is calculated by an EMG threshold calculation model, which is a matrix, specifically: T EMG =[T up T down T left T right T front T back ] Among them, T EMG The optimal electromyographic threshold; T up T down T left T right T front T back The threshold values ​​for up, down, left, right, front, and back; a up a down a left a right a front a back a represents the characteristic coefficients of muscle contraction in all directions (up, down, left, right, forward, backward), and lateral. rest The characteristic coefficient during rest; F EMG-up F EMG-down F EMG-left F EMG-right F EMG-front F EMG-back F represents the electromyographic characteristics of muscles contracting in all directions (up, down, left, right, forward, backward). EMG-rest The values ​​represent the resting electromyographic (EMG) characteristics; m represents the data length. EMG frequency band characteristics include: mean, standard deviation, average, standard deviation, power spectral density, and asymmetry coefficient.

[0053] The optimal EEG threshold described in this invention is calculated using an EEG threshold calculation model, which is as follows: T EEG = (1+ε)*T std Among them, T EEG The optimal EEG threshold; ε is the characteristic coefficient, ε > -1; T std The preset standard threshold; h i Here, represents the coefficients for each different EEG channel, and ch represents the number of EEG channels. k1 / k2 / k3 / k4 / k5 are coefficients representing the EEG characteristics of delta waves, theta waves, alpha waves, beta waves, and SMR waves during imagined / attempted limb movements and at rest; the sum of these five coefficients is 1. F δ-image For the characteristics of delta waves in brainwaves when imagining / attempting to move limbs, F θ-image For the theta wave characteristics of brainwaves when imagining / attempting to move limbs, F α-image For the characteristics of alpha waves in brainwaves when imagining / attempting to move limbs, F β-image For the characteristics of beta waves in brainwaves when imagining / attempting to move limbs, F SMR-rest For the characteristics of SMR brain waves when imagining / attempting to move limbs, F δ-rest The characteristics of resting delta waves in brainwaves, F θ-rest The theta wave characteristics of resting brainwaves, F α-rest The resting EEG alpha wave characteristics, F β-rest The resting EEG beta wave characteristics, F SMR-rest The resting SMR wave characteristics are shown; m is the data length.

[0054] It should be noted that in this invention, when the EEG acquisition module collects EEG information, the signal acquisition of the imagined / attempted limb movement state or rest state will be continuously collected for a certain period of time. Therefore, there will be a certain data length, so as to obtain the feature information of different states more completely and avoid omissions.

[0055] S2, once training has officially begun, acquire real-time electromyography (EMG) and electroencephalography (EEG) data; S3, perform electromyography prediction: input the optimal electromyography threshold and real-time electromyography data into the electromyography direction / movement evaluation model, the electromyography direction / movement evaluation model calculates the score of the probability of movement in each direction, selects the direction with the highest score of the probability of movement in each direction as the direction of movement, or synthesizes the scores into a direction vector as the direction of movement. The electromyographic direction / movement assessment model of this invention is as follows: Among them, ScoreEMG This is a score matrix for electromyographic direction / movement assessment; EMG-up score EMG-down score EMG-left score EMG-right score EMG-front score EMG-back Scoring is based on muscle contraction in all directions (up, down, left, right, forward, backward); C EMG The preset auxiliary scores are: k1, k2, k3, k4, k5, and k6 are the characteristic coefficients of electromyography for each direction (up, down, left, right, front, and back); F EMG-up F EMG-down F EMG-left F EMG-right F EMG-front F EMG-back The characteristics of electromyography (EMG) in all directions (up, down, left, right, front, and back); T up T down T left T right T front T back The threshold values ​​for up, down, left, right, front, and back; k manual The therapist can manually adjust the parameters during use, increasing or decreasing the threshold to increase or decrease the difficulty; m is the data length.

[0056] h i Here, represents the coefficients for each different electromyographic channel, and ch represents the number of electromyographic channels. The parameters were the same during electromyography assessment; F EMG For the electromyographic characteristics of imagining / attempting to move a limb, T EMG is the optimal electromyography threshold, and m is the data length.

[0057] It should be noted that the preset auxiliary score can be set according to actual needs, and this auxiliary score can be used as a preset score to judge the imagination state.

[0058] S4, EEG prediction: Input the optimal EEG threshold and real-time EEG data into the EEG motor intention scoring model for calculation to obtain a real-time EEG score. Determine if the real-time EEG score is greater than a preset value. If not, and the termination condition is not met, continue to calculate the EMG and EEG scores in real time. If yes, drive the multi-degree-of-freedom rehabilitation training module to move in the direction of movement obtained in S3, and return to S2. Termination condition: The real-time EEG score is less than a preset value or the training time is reached within the set duration.

[0059] This invention combines brain-computer interfaces, facial electromyography (EMG), and multi-degree-of-freedom rehabilitation devices. By pre-collecting EEG and facial EMG information of the trainee's imagined limb movements, an EEG and facial EMG model is established. During training, the movement pattern or direction is determined by recognizing facial EMG, and the trainee's movement intention is recognized by EEG. When the trainee wants to move the corresponding limb in a specified pattern or direction, robots, electrical stimulation, magnetic stimulators, etc. are used to drive / stimulate the nerves and muscles of the relevant limb, thereby assisting the limb to perform the corresponding movement and conducting multi-degree-of-freedom active rehabilitation training.

[0060] The real-time EEG score of this invention is calculated using an EEG motor intention scoring model, which is as follows: Among them, Score EEG For brainwave motor intention scoring; C EEG The preset auxiliary score; h i Here, represents the coefficients for each different EEG channel, and ch represents the number of EEG channels. The parameters are the same during EEG assessment; k0 is the score conversion coefficient; k1 / k2 / k3 / k4 / k5 are the coefficients of the EEG characteristics of delta waves, theta waves, alpha waves, beta waves, and SMR waves in the EEG during imagination / attempt to move limbs and in the resting state, and the sum of the five is 1; k manual Therapists can manually adjust parameters during use, increasing or decreasing the threshold to adjust the difficulty level; F δ For the characteristics of delta waves in brainwaves when imagining / attempting to move limbs, F θ For the theta wave characteristics of brainwaves when imagining / attempting to move limbs, F α For the characteristics of alpha waves in brainwaves when imagining / attempting to move limbs, F β For the characteristics of beta waves in brainwaves when imagining / attempting to move limbs, F SMRThe image shows the SMR wave characteristics of EEG during imagined / attempted limb movement, with m representing the data length. A preset auxiliary score can be set according to actual needs and can be used as a preset score to assess the imagined state. The theta band, with a frequency range of 4-8 Hz, is found to be more prevalent in the brains of infants and adults and adolescents during drowsiness (or early sleep); it also appears when the brain is idle or during meditation; theta wave energy is positively correlated with the intensity of chronic pain. The alpha band, with a frequency range of 8-13 Hz, is the most prominent wave in rhythmic EEG, typically appearing at the back of the head, present on both sides, with a higher amplitude on the dominant side; alpha band signals can be detected in EEG scans of the occipital lobe when a conscious person is relaxed or has their eyes closed. Alpha energy is negatively correlated with pain intensity. In BCI applications, a specific beta band signal is called the SMR band (12-15 Hz), whose appearance is often associated with resting states, especially during sleep and wakefulness. In patients with neuropathic pain, there appears to be a decrease during the training phase. The trainee contracts corresponding muscles in the front, back, left, right, up, and down directions. The system calculates the desired direction or movement based on electromyographic direction / motor score. When one or more direction scores exceed a threshold, the direction with the highest score is selected as the user's movement direction. The system also scores motor intention using an EEG motor intention scoring model. The calculated imagined state score... EEG When the preset score is exceeded, the control module issues a control command, instructing the multi-degree-of-freedom rehabilitation training module to move according to the selected direction. The audiovisual feedback module displays the electromyography and electroencephalography characteristics of the trainee in real time, while also displaying animations of muscle contraction and relaxation, and providing auditory cues to guide the trainee to perform training better.

[0061] This embodiment also proposes an active rehabilitation training system based on brain-computer interface and electromyography, referencing... Figure 2The system includes an EEG acquisition module 1, which acquires EEG signals and inputs them to an EEG assessment module 2 and an EEG motion prediction module 3. During EEG assessment, the EEG assessment module 2 evaluates the characteristic signals related to brain movement, generates and stores an EEG model, and transmits the results to the EEG motion prediction module 3. During training, the EEG motion prediction module 3 predicts the user's movement intention based on the real-time EEG signals and the EEG model generated by the EEG assessment module, and sends the results to a control module 7. The system also includes an EMG acquisition module 4, which acquires EMG signals and inputs them to an EMG assessment module 5 and an EMG prediction module 6. During EMG assessment, the EMG assessment module 5 evaluates the characteristic signals of EMG, generates and stores an EMG model, and transmits the results to the EMG prediction module 7. During training, the EMG prediction module 7 predicts the user's movement direction or action based on the real-time EMG signals and the EMG model generated by the EMG assessment module 5, and sends the results to the control module 7. The EEG motion prediction module 3 predicts whether the user is ready to start exercising based on the real-time EEG signals and the EEG model generated by the EEG assessment module 2, and sends the results to the control module 7. The control module 7 controls the operation of the audiovisual feedback module 8 and the multi-degree-of-freedom rehabilitation training module 9 based on the results of the electromyography prediction module 6 and the electroencephalography prediction module 3.

[0062] The system works as follows: it collects and evaluates EEG and EMG signals; it evaluates the feature signals and stores the model, then transmits the results to the prediction module; it begins training by selecting different directions or movements based on EMG, and activates the device based on EEG to start the movement, providing real-time feedback to the user; the user switches directions or movements, and then training continues until the training ends.

[0063] The EEG acquisition module 1 includes an EEG cap (containing electrodes) and an EEG-to-analog converter module, which converts the acquired EEG signals into digital signals and transmits them to the system. The number of electrodes on the EEG cap can be 2, 4, 8, or 16 leads, but 16 leads are common. The electrodes can be dry, semi-dry, or wet electrodes, but wet electrodes are common. The EEG-to-analog converter module uses conventional technology from existing EEG machines to convert analog signals into digital signals for transmission, while reducing signal noise and improving the signal-to-noise ratio. Simultaneously, signal preprocessing is performed using common EEG signal analysis and processing methods such as Fast Fourier Transform, Butterworth filter, and Chebyshev filter.

[0064] The electromyography (EMG) acquisition and decoding module 5 includes EMG electrodes and an EMG-to-analog converter module. This module converts the acquired EMG signals into digital signals and sends them to the system. The number of EMG electrodes can be 6 or 8 leads, but 6 leads are common. The electrodes can be dry, semi-dry, or wet, but wet electrodes are common. The EMG-to-analog converter module uses conventional technology from existing EMG machines to convert analog signals into digital signals for transmission, while reducing signal noise and improving the signal-to-noise ratio. Simultaneously, signal preprocessing is performed using common EMG signal analysis and processing methods such as Fast Fourier Transform, Butterworth filter, and Chebyshev filter.

[0065] The control module 9 receives information from the EEG assessment module 2, EEG motion prediction module 3, EMG assessment module 5, and EMG prediction module 6. After processing by the control module 7, the collected information is output as control signals, activating the audiovisual feedback module 8 and the multi-degree-of-freedom rehabilitation training module 9. The control module can be a computer host, control motherboard, etc., and can run operating systems such as Windows, Linux, and Android.

[0066] The audiovisual feedback module 8 includes a monitor, projector, speakers (headphones), VR glasses, etc.; among which, visual feedback includes images of body movements provided by the computer, examples of models, expert guidance, etc., or movements presented using virtual reality technology, etc.; auditory feedback includes movement voice guidance provided by the system, start and stop prompts, success or failure feedback, etc.

[0067] Multi-degree-of-freedom rehabilitation training module 9 is a treatment method that uses rehabilitation equipment to act on the peripheral neuromuscular system; existing conventional lower limb exoskeleton robots, upper limb exoskeleton robots, functional electrical stimulation, etc. can be selected for motor rehabilitation training.

[0068] The electromyography assessment module 5 stores the electromyography threshold calculation model; it obtains the electromyography frequency band feature data of the required frequency band based on the collected electromyography signals; Based on the collected electromyography (EMG) data over a period of time, preprocessing was performed, including notch filtering at 50 Hz power frequency and bandpass filtering from 55 to 200 Hz, to obtain the preprocessed data. The characteristic data F of each EMG frequency band was then calculated. EMG Electromyographic frequency band characteristics include: mean, standard deviation, power spectral density, asymmetry coefficient, etc.

[0069] The training subjects' facial / neck muscles correspond to different movement directions or actions, such as: frontalis muscle – upper; mentalis muscle – lower; left masseter muscle – left; right masseter muscle – right; left platysma muscle – anterior; right platysma muscle – posterior. Electromyographic (EMG) signals from various muscle groups are extracted when the training subjects perform muscle contractions according to prompts. These signals are then substituted into an EMG threshold calculation model to calculate the optimal EMG threshold T.EMG .

[0070] The electromyography (EMG) assessment module decodes the optimal EMG threshold by collecting EMG information from the training subject when imagining / attempting to move a limb and when at rest, in order to distinguish whether the training subject is in an imagining limb movement state or a resting state, and establishes an imagination model for use by the EMG motion prediction module.

[0071] In this invention, when the electromyography acquisition module collects EEG information, the signals collected during muscle contraction or resting state in the up, down, left, right, front, and back are continuously collected for a certain period of time. Therefore, there will be a certain data length, which can obtain the characteristics of different states more completely and avoid omissions.

[0072] During the EEG assessment, the electromyography (EMG) assessment module calculates and updates EMG characteristics in real time, providing feedback to the trainee through the audiovisual feedback module's display in the form of line graphs, energy maps, and games. When prompted by the system to contract the muscles in all directions (up, down, left, right, forward, backward), the trainee contracts these muscles; when prompted to relax, the trainee relaxes their body and does not visualize movement. The assessment ends when the time is up or the trainee feels they have mastered the technique. After the assessment, the trainee's EMG characteristics and the optimal EMG threshold provided by the system can be viewed. The trainee can use the system's optimal threshold, manually adjust the threshold, or recommend a second assessment.

[0073] The electromyography (EMG) prediction module 6 stores an EMG direction / movement assessment model; it acquires EMG frequency band characteristic data for different frequency bands based on the collected EMG signals; based on the EMG data collected over a period of time, and distinguishing different types of EEG frequency bands, it performs preprocessing including notch filtering at 50Hz power frequency and bandpass filtering from 55-200Hz, obtains the preprocessed data, and calculates the EMG frequency band characteristic data F for each EMG signal. EMG The electromyographic frequency band characteristics include: mean, standard deviation, power spectral density, asymmetry coefficient, etc. These, along with the optimal electromyographic threshold obtained from the electromyographic assessment module, are input into the electromyographic direction / movement assessment model to calculate the training subject's scores in each direction. The direction with the highest score is taken as the movement direction.

[0074] The EEG assessment module 2 stores the EEG threshold calculation model; it acquires EEG frequency band characteristic data for different frequency bands based on the collected EEG signals; based on the EEG data collected over a period of time, and distinguishing them according to different types of EEG frequency bands, it performs preprocessing including notch filtering at 50Hz power frequency and bandpass filtering from 0.5-45Hz, obtains the preprocessed data, and calculates the EEG frequency band characteristic data F for each EEG frequency band. (δ / θ / α / β / SMR)EEG frequency band characteristics include: mean, standard deviation, power spectral density, asymmetry coefficient, etc. The imaginary state characteristics F of the training subject when imagining / attempting to move a limb are also considered. EEG-image and the state characteristics F during rest EEG-rest Substituting these values ​​into the EEG threshold calculation model, the optimal EEG threshold T for each EEG frequency band is calculated. EEG .

[0075] The EEG assessment module decodes the optimal EEG threshold by collecting EEG information from the training subject when imagining / attempting to move limbs and when at rest, in order to distinguish whether the training subject is in an imagining limb movement state or a resting state, and establishes an imagination model for use by the EEG motion prediction module.

[0076] The evaluation process will use the characteristic data of each frequency band F (δ / θ / α / β / SMR) The data is displayed in real-time as a line graph to both the trainee and the operator. When prompted by the system to imagine / attempt limb movement, the trainee performs the imagining / attempt; when prompted to rest, the trainee remains in a resting state. After the evaluation, the system calculates the optimal EEG threshold T based on the differences in EEG characteristics between the trainee's imagined / attempted movement state and relaxed state. EEG .

[0077] During the EEG assessment, the system calculates and updates the EEG characteristics of the imagined / attempted state in real time, providing feedback to the trainee through the audiovisual feedback module's display in the form of line graphs, energy maps, games, etc. When prompted by the system to imagine / attempt limb movements, the trainee imagines / attempts limb movements; when prompted by the system, the trainee relaxes their body and does not imagine movements. The trainee executes the movements through imagination or by using different mental strategies. The assessment ends when the time is up or the trainee believes they have mastered the skills. After the assessment, the trainee's EEG characteristics and the optimal EEG thresholds provided by the system can be viewed. The trainee can use the optimal EEG thresholds provided by the system or manually adjust the thresholds, or recommend that the trainee undergo another assessment.

[0078] The EEG motion prediction module 3 stores the EEG motion intention scoring model; it acquires EEG frequency band feature data of different frequency bands based on the collected EEG signals; based on the EEG data collected over a period of time, it distinguishes different types of EEG frequency bands, and then performs preprocessing including notch filtering at 50Hz power frequency and bandpass filtering from 0.5-45Hz to obtain the preprocessed data (data), and calculates the EEG frequency band feature data F for each EEG frequency band. (δ / θ / α / β / SMR)The EEG frequency band features include: mean, standard deviation, average, standard deviation, power spectral density, asymmetry coefficient, etc. These, along with the optimal EEG threshold obtained from the EEG assessment module, are input into the EEG motor intention scoring model to calculate the trainee's EEG motor intention score. The EEG motor prediction module decodes the EEG features of the trainee when imagining / attempting to move their limbs in real time, compares them with the optimal EEG threshold obtained from the EEG assessment module, and transmits the comparison result to the control module.

[0079] The specific implementation of the brain-computer interface and facial / neck electromyography rehabilitation training system of this invention is as follows: The therapist puts on the EEG cap for the patient and attaches the corresponding electromyography electrodes to the corresponding facial / neck muscles. The patient sits or lies comfortably in a chair or on a treatment bed. The therapist activates the electromyography assessment unit of the system. The system prompts the user to contract and rest the corresponding muscles. The patient tries accordingly. After the trial, the system generates an assessment report and recommends the optimal electromyography threshold. The therapist activates the EEG assessment unit of the system. The system prompts the user to imagine / attempt to move the limbs and rest. The patient tries accordingly. After the trial, the system generates an assessment report and recommends the optimal EEG threshold. The training mode is then activated. After training begins, the patient actively contracts the facial / neck muscles. The system calculates the scores for each direction of the training subject based on real-time electromyography and the electromyography direction / movement assessment model. The direction with the highest score is taken as the direction of movement. The patient then imagines / attempts to move the limbs. When the EEG movement intention score is... EEG When the preset score is exceeded, the multi-degree-of-freedom rehabilitation training module acts on the moving limbs, while receiving real-time feedback from the audiovisual feedback module's display and speakers. When the patient wants to move in other directions, they actively contract their facial / neck muscles to switch the direction of movement, and then imagine / attempt to move the limbs, performing active rehabilitation training until the training ends.

[0080] The specific embodiments described herein are merely illustrative examples illustrating the spirit of the invention. The above embodiments only express several implementation methods of this application, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art to which this application pertains can make various modifications or additions to the described specific embodiments or use similar methods to replace them, but without departing from the spirit of this application or exceeding the scope defined by the appended claims. For those skilled in the art, multiple variations and improvements can be made without departing from the concept of this application. Therefore, the scope of protection of this application should be determined by the appended claims.

Claims

1. An active rehabilitation training method based on brain-computer interface and electromyography, characterized in that, include: S1. Before training, EEG and EMG assessments are performed to obtain the optimal EEG and EMG thresholds. S2, after training begins, acquires real-time EEG data and real-time EMG data; S3: Input real-time electromyography data and the optimal electromyography threshold into the electromyography direction / movement assessment model to calculate the score for each direction, and take the direction with the highest score or synthesize the scores into a direction vector as the direction of movement. S4: The EEG motor intention scoring model calculates the real-time EEG score based on real-time EEG data and the optimal EEG threshold. It determines whether the real-time EEG score is greater than the preset value. If so, it drives the multi-degree-of-freedom rehabilitation training module to move in the direction of movement and returns to S2. If not, it returns to S2 if the termination condition has not been reached.

2. The active rehabilitation training method based on brain-computer interface and electromyography according to claim 1, characterized in that, The optimal EEG threshold and optimal EMG threshold are manually adjusted according to the training intensity.

3. The active rehabilitation training method based on brain-computer interface and electromyography according to claim 1 or 2, characterized in that, The optimal EEG threshold is obtained by decoding the EEG information collected from the training subject when imagining / attempting to move limbs and when at rest, thus distinguishing whether the training subject is in a state of imagining limb movement or a state of rest. The optimal electromyography (EMG) threshold is obtained by decoding the EMG information collected from the training subject when imagining / attempting to move the limbs and when at rest, thus distinguishing whether the training subject is in a state of imagining limb movement or a state of rest.

4. The active rehabilitation training method based on brain-computer interface and electromyography according to claim 2, characterized in that, The process for obtaining the optimal electromyographic threshold is as follows: Calculate the threshold values ​​for each direction: Set the characteristic coefficients for muscle contraction and rest in each direction; Acquire resting electromyographic characteristics and electromyographic characteristics of muscle contraction in various directions at a set data length; Calculate the mean value of electromyographic characteristics of muscle contraction in each direction under a set data length; Threshold for each direction = Characteristic coefficient of muscle contraction in each direction × Mean of electromyographic characteristics of muscle contraction in each direction under the set data length - Characteristic coefficient at rest × Electromyographic characteristics at rest; The optimal electromyography (EMG) threshold is obtained by extracting EMG frequency band features from the thresholds in each direction.

5. The active rehabilitation training method based on brain-computer interface and electromyography according to claim 2, characterized in that, The process for obtaining the optimal EEG threshold is as follows: Optimal EEG threshold = (1 + feature coefficient) × preset standard threshold The characteristic coefficient is greater than -1; Set the coefficients for each different EEG channel and the first coefficients for various EEG characteristics during imagination / attempt to move limbs and in the resting state; Acquire various EEG characteristics at a set data length during imagining / attempting to move limbs and at rest. Calculate the first cumulative value of the difference between the EEG characteristics of various types of EEG features when imagining / attempting to move limbs and the EEG characteristics at rest, under a set data length; and the second cumulative value of the sum of the EEG characteristics of various types of EEG features when imagining / attempting to move limbs and the EEG characteristics at rest, under a set data length. The ratio of the first cumulative value to the second cumulative value of each type of EEG feature is calculated and then multiplied by the first coefficient of each type of EEG feature to obtain the first reference value of each type of EEG feature; The sum of the first reference values ​​of various EEG characteristics is multiplied by the coefficients of each different EEG channel, and the cumulative value of all EEG channels is used as the characteristic coefficient.

6. The active rehabilitation training method based on brain-computer interface and electromyography according to claim 4, characterized in that, The specific process by which the electromyographic direction / movement assessment model calculates the score for each direction is as follows: Preset auxiliary scores and characteristic coefficients of electromyography in each direction; Calculate the mean value of the electromyographic characteristics of muscles in each direction during contraction, within a set data length; Set the manual adjustment parameters for each direction and obtain the threshold values ​​for each direction; Muscle contraction score in each direction = preset auxiliary score + characteristic coefficient of electromyography in each direction × (mean of electromyographic characteristics of muscle contraction in each direction under the set data length - manual adjustment parameter of each direction × score threshold of each direction).

7. The active rehabilitation training method based on brain-computer interface and electromyography according to claim 5, characterized in that, The S4 EEG motor intention scoring model calculates the real-time EEG score based on real-time EEG data and the optimal EEG threshold, specifically including the following steps: Preset auxiliary scores and characteristic coefficients of various EEG patterns; The first calculated value is the product of the cumulative value of the EEG characteristics of various types of EEG when imagining / attempting to move limbs over a set data length and the characteristic coefficients of various types of EEG. The first mean value is obtained by summing the first calculated values ​​of various EEG types and then dividing by the set data length. Obtain the coefficients and number of brainwave channels for each different brainwave channel; Obtain manually adjustable parameters and optimal EEG thresholds; The second calculated value is obtained by subtracting the product of the coefficients of each different EEG channel and the first mean from the product of the manually adjusted parameters and the optimal EEG threshold. The real-time EEG score is equal to the preset auxiliary score plus the second calculated value, which is the sum of the number of EEG channels.

8. An active rehabilitation training system based on brain-computer interface and electromyography, employing the active rehabilitation training method based on brain-computer interface and electromyography as described in any one of claims 1-7, characterized in that, The system includes a control module (7), which is electrically connected to an EEG assessment module (2) and an EEG motion prediction module (3), as well as an EMG assessment module (5) and an EMG prediction module (6). The EEG assessment module (2) and the EEG motion prediction module (3) are electrically connected to the EEG acquisition module (1), the EEG assessment module (2) and the EEG motion prediction module (3) are electrically connected, the EMG assessment module (5) and the EMG prediction module (6) are electrically connected, the EMG assessment module (5) and the EMG prediction module (6) are electrically connected to the EMG acquisition module (4), and the control module (7) is electrically connected to a multi-degree-of-freedom rehabilitation training module (9).

9. The active rehabilitation training system based on brain-computer interface and electromyography according to claim 8, characterized in that, The control module (7) is electrically connected to the audiovisual feedback module (8).

10. The active rehabilitation training system based on brain-computer interface and electromyography according to claim 8, characterized in that, The EEG assessment module (2) is equipped with an EEG threshold calculation model; the EEG motion prediction module (3) is equipped with an EEG motion intention scoring model; the EMG assessment module (5) is equipped with an EMG threshold calculation model; and the EMG prediction module (6) is equipped with an EMG direction / movement assessment model.

Images