Rehabilitation training plan processing method, brain-computer interface device and storage medium

Abstract

The invention provides a rehabilitation training plan processing method, a brain-computer interface device and a storage medium, and relates to the technical field of brain-computer interfaces. According to the method, the motor imagery ability evaluation result of the target object is obtained mainly based on the electroencephalogram signal of the target object, that is, the motor imagery ability of different patients can be quantitatively evaluated according to the scheme; then, aiming at the two indexes of the motor imagery ability evaluation result and the dyskinesia level of the patient, a rehabilitation training plan is generated for the patient in a personalized manner, namely, the rehabilitation plan has the characteristics of pertinence, scientificity, reasonability and the like, so that the rehabilitation efficiency of the patient is greatly improved, and the rehabilitation time of the patient is saved; meanwhile, the use efficiency of the rehabilitation robot is improved, that is, the rehabilitation effect of the motor imagery therapy is better than that of a traditional rehabilitation method, and the problem that in the prior art, rehabilitation training is unreasonable in formulation is solved.

Description

Technical Field

[0001] The present invention relates to the field of electroencephalogram signal technology, and in particular to a method for processing a rehabilitation training plan, a brain-computer interface device and a storage medium. Background Art

[0002] In the field of rehabilitation training, the human body often needs to perform specific movements to achieve the purpose of sports training and rehabilitation, such as hand and foot movements, torso twisting or bending, chest or abdominal contraction and expansion or pressure maintenance, upper and lower limb bending movements, etc. In order to achieve certain sports training and rehabilitation effects, it is necessary to formulate a targeted rehabilitation training plan based on the actual situation of the patient.

[0003] In the existing technology, a dedicated rehabilitation training plan is formulated for each patient based on the patient's motor disorder level. After each round of rehabilitation training, the rehabilitation training plan needs to be adjusted regularly according to the rehabilitation status of each patient to ensure the effectiveness of the rehabilitation training.

[0004] However, the rehabilitation training plan formulated in the prior art is only dependent on one indicator, the patient's motor disorder level, which may lead to unscientific problems in the formulation of the patient's rehabilitation training plan, thereby resulting in poor rehabilitation effect. Summary of the invention

[0005] The purpose of the present invention is to provide a method for processing a rehabilitation training plan, a brain-computer interface device and a storage medium in order to solve the technical problems existing in the prior art in view of the above-mentioned deficiencies in the prior art.

[0006] To achieve the above purpose, the technical solution adopted in the embodiment of the present application is as follows:

[0007] In a first aspect, an embodiment of the present application provides a method for processing a rehabilitation training plan, which is applied to a brain-computer interface device, and the method includes:

[0008] Acquiring EEG signals when the target object performs multiple motor imagery, wherein the EEG signals are acquired by an EEG acquisition unit on the brain-computer interface device;

[0009] Preprocessing the EEG signal to obtain a preprocessed EEG signal;

[0010] Performing feature extraction on the preprocessed EEG signal to obtain a feature matrix of the preprocessed EEG signal;

[0011] Inputting the feature matrix into a pre-trained target classification model to obtain a motor imagery ability evaluation result of the target object;

[0012] Generate a rehabilitation training plan for the target object according to the evaluation result of the target object's motor imagery ability and the motor disorder level.

[0013] Optionally, the step of inputting the feature matrix into a pre-trained target classification model to obtain the evaluation result of the target object's motor imagery ability includes:

[0014] Input the feature matrix into the target classification model to obtain the classification result output by the target classification model, and use the classification result as the evaluation result of the target object's motor imagery ability.

[0015] Optionally, the step of extracting features from the preprocessed EEG signals to obtain a feature matrix includes:

[0016] Use a preset window function to segment the preprocessed EEG signals to obtain at least one segment signal;

[0017] Extract features from each of the segment signals to obtain the feature matrix corresponding to each of the segment signals.

[0018] Optionally, the step of extracting features from each of the segment signals to obtain the feature matrix corresponding to each of the segment signals includes:

[0019] Perform wavelet packet transform on the segment signal to obtain multiple wavelet packet coefficients of the segment signal;

[0020] Select the first M wavelet packet coefficients from the multiple wavelet packet coefficients, and use the first M wavelet packet coefficients as the input parameters of the filter bank common spatial pattern algorithm to obtain the feature matrix of the segment signal based on the filter bank common spatial pattern algorithm.

[0021] Optionally, the step of performing wavelet packet transform on the segment signal to obtain multiple wavelet packet coefficients of the segment signal includes:

[0022] Perform N-layer wavelet packet decomposition on the segment signal to obtain 2^N wavelet packet coefficients of the segment signal.

[0023] Optionally, before inputting the feature matrix into a pre-trained target classification model to obtain the evaluation result of the target object's motor imagery ability, it further includes:

[0024] Use a plurality of pre-constructed sample data to obtain the sample feature matrix corresponding to each sample data, where each sample data includes EEG signals in a resting state or EEG signals of motor imagery, and the classification label corresponding to the EEG signal;

[0025] Combine the sample feature matrix corresponding to the sample data and the classification label corresponding to the sample data to form a training data set;

[0026] Train the target classification model based on the training data set.

[0027] Optionally, the training the target classification model based on the training data set includes:

[0028] Divide the training data set to obtain multiple sub-data sets;

[0029] Use one of the multiple sub-data sets as a validation sub-data set, and use the sub-data sets other than the validation sub-data set as training sub-data sets;

[0030] Iteratively train the constructed initial model based on the training sub-data set, iteratively validate the trained initial model based on the validation sub-data set, and use the initial model that meets the preset validation conditions as the target classification model.

[0031] Optionally, the target classification model includes: a support vector machine binary classification model, a decision tree classification model, a Bayesian classifier, or a neural network classifier.

[0032] Optionally, the preprocessing the electroencephalogram signal to obtain a preprocessed electroencephalogram signal includes:

[0033] Filter the electroencephalogram signal using a first non-recursive filter to obtain a filtered electroencephalogram signal;

[0034] Filter the filtered electroencephalogram signal using a second non-recursive filter to obtain a preprocessed electroencephalogram signal.

[0035] In a second aspect, an embodiment of the present application further provides a processing device for a rehabilitation training plan, which is applied to a brain-computer interface device. The device includes:

[0036] An acquisition module, configured to acquire electroencephalogram signals when a target object performs multiple motor imagery, and the electroencephalogram signals are acquired by an electroencephalogram acquisition unit on the brain-computer interface device;

[0037] A processing module, configured to preprocess the electroencephalogram signals to obtain preprocessed electroencephalogram signals;

[0038] An extraction module, configured to extract features from the preprocessed electroencephalogram signals to obtain a feature matrix of the preprocessed electroencephalogram signals;

[0039] The processing module is further configured to input the feature matrix into a pre-trained target classification model to obtain a motor imagery ability evaluation result of the target object;

[0040] A generation module, configured to generate a rehabilitation training plan for the target object according to the evaluation result of the motor imagery ability and the motor disorder level of the target object.

[0041] Optionally, the processing module is further configured to

[0042] Input the feature matrix into the target classification model, obtain the classification result output by the target classification model, and use the classification result as the evaluation result of the motor imagery ability of the target object.

[0043] Optionally, the extraction module is further configured to:

[0044] Use a preset window function to segment the preprocessed EEG signal to obtain at least one segment signal;

[0045] Extract features from each of the segment signals to obtain a feature matrix corresponding to each of the segment signals.

[0046] Optionally, the extraction module is further configured to:

[0047] Perform wavelet packet transform on the segment signal to obtain multiple wavelet packet coefficients of the segment signal;

[0048] Select the first M wavelet packet coefficients from the multiple wavelet packet coefficients, and use the first M wavelet packet coefficients as the input parameters of the filter bank common spatial pattern algorithm, and obtain the feature matrix of the segment signal based on the filter bank common spatial pattern algorithm.

[0049] Optionally, the extraction module is further configured to:

[0050] Perform N-layer wavelet packet decomposition on the segment signal to obtain 2^N wavelet packet coefficients of the segment signal.

[0051] Optionally, the extraction module is further configured to use a plurality of pre-constructed sample data to obtain a sample feature matrix corresponding to each sample data, where each sample data includes an EEG signal in a resting state or an EEG signal of motor imagery, and a classification label corresponding to the EEG signal;

[0052] The apparatus further includes:

[0053] A composition module, configured to form a training data set by combining the sample feature matrix corresponding to the sample data with the classification label corresponding to the sample data;

[0054] A training module, configured to train the target classification model based on the training data set.

[0055] Optionally, the training module is further configured to:

[0056] Partition the training data set to obtain multiple sub - data sets;

[0057] Use one of the multiple sub - data sets as the validation sub - data set, and use the sub - data sets other than the validation sub - data set as the training sub - data set;

[0058] Iteratively train the constructed initial model based on the training sub - data set, iteratively validate the trained initial model based on the validation sub - data set, and use the initial model that meets the preset validation conditions as the target classification model.

[0059] Optionally, the target classification model includes: a support vector machine binary classification model, a decision tree classification model, a Bayesian classifier, or a neural network classifier.

[0060] Optionally, the processing module is further configured to:

[0061] Filter the electroencephalogram (EEG) signal using a first non - recursive filter to obtain a filtered EEG signal;

[0062] Filter the filtered EEG signal using a second non - recursive filter to obtain a pre - processed EEG signal.

[0063] In a third aspect, an embodiment of the present application further provides a brain - computer interface device, including: a processor, a storage medium, and a bus. The storage medium stores machine - readable instructions executable by the processor. When the electronic device runs, the processor communicates with the storage medium through the bus, and the processor executes the machine - readable instructions to perform the steps of the method provided in the first aspect.

[0064] In a fourth aspect, an embodiment of the present application further provides a computer - readable storage medium. A computer program is stored on the storage medium, and when the computer program is run by a processor, it performs the steps of the method provided in the first aspect.

[0065] The beneficial effects of the present application are:

[0066] The embodiments of the present application provide a method for processing a rehabilitation training plan, a brain-computer interface device, and a storage medium. In this solution, mainly based on the electroencephalogram (EEG) signals of a target object, an evaluation result of the motor imagery ability of the target object is obtained, that is, this solution proposes to quantitatively evaluate the motor imagery ability of a patient; then, aiming at the two indicators of the evaluation result of the patient's motor imagery ability and the motor disorder level, a rehabilitation training plan is customized for the patient, that is, this rehabilitation plan is specifically tailored for the target object, and this rehabilitation plan has characteristics such as pertinence, scientificity, and rationality, so that the rehabilitation robot can help the patient carry out rehabilitation training targeted according to this rehabilitation plan, greatly improving the patient's rehabilitation efficiency and saving the patient's rehabilitation time, and at the same time improving the use efficiency of the rehabilitation robot, that is, the rehabilitation effect of the motor imagery therapy proposed in this application is better than that of traditional rehabilitation methods, solving the problem of unscientific formulation of rehabilitation training in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present invention, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings.

[0068] Figure 1 A schematic diagram of the architecture of a brain-computer interface device provided by the embodiments of the present application;

[0069] Figure 2 A schematic diagram of the structure of an electroencephalogram acquisition unit in a brain-computer interface device provided by the embodiments of the present application;

[0070] Figure 3 A schematic diagram of the structure of a brain-computer interface device provided by the embodiments of the present application;

[0071] Figure 4 A schematic flowchart of another method for processing a rehabilitation training plan provided by the embodiments of the present application;

[0072] Figure 5 A schematic flowchart of another method for processing a rehabilitation training plan provided by the embodiments of the present application;

[0073] Figure 6 A schematic flowchart of another method for processing a rehabilitation training plan provided by the embodiments of the present application;

[0074] Figure 7 A schematic diagram of wavelet packet transform of electroencephalogram signals in a method for processing a rehabilitation training plan provided by the embodiments of the present application;

[0075] Figure 8 Schematic flowchart of another method for processing a rehabilitation training plan provided by an embodiment of the present application;

[0076] Figure 9 Schematic flowchart of another method for processing a rehabilitation training plan provided by an embodiment of the present application;

[0077] Figure 10 Schematic flowchart of another method for processing a rehabilitation training plan provided by an embodiment of the present application;

[0078] Figure 11 Overall schematic flowchart of the method for processing a rehabilitation training plan provided by an embodiment of the present application;

[0079] Figure 12 Schematic structural diagram of a device for processing a rehabilitation training plan provided by an embodiment of the present application. Detailed implementation manners

[0080] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. It should be understood that the accompanying drawings in the present application are only for the purposes of illustration and description, and are not used to limit the protection scope of the present application. In addition, it should be understood that the schematic drawings are not drawn to actual scale. The flowcharts used in the present application illustrate the operations implemented according to some embodiments of the present application. It should be understood that the operations in the flowchart may not be implemented in sequence, and steps without logical context relationships may be reversed or implemented simultaneously. In addition, those skilled in the art may add one or more other operations to the flowchart or remove one or more operations from the flowchart under the guidance of the content of the present application.

[0081] In addition, the described embodiments are only some embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application described and illustrated in the accompanying drawings here may be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the present application claimed, but merely represents selected embodiments of the present application. All other embodiments obtained by those skilled in the art based on the embodiments of the present application without creative efforts fall within the protection scope of the present application.

[0082] It should be noted that the term "including" will be used in the embodiments of the present application to indicate the existence of the features stated hereinafter, but does not exclude the addition of other features.

[0083] First, before specifically describing the technical solutions provided by the present application, a brief description of the architecture of the brain-computer interface device provided by the present application will be given.

[0084] Figure 1 It is a schematic structural diagram of a brain-computer interface device provided by an embodiment of the present application; the brain-computer interface device includes: an electroencephalogram (EEG) acquisition unit 101 and an EEG analysis unit 102, wherein the EEG acquisition unit 101 is communicatively connected to the EEG analysis unit 102.

[0085] The EEG acquisition unit 101 is configured to acquire the EEG signals of a target object, amplify the EEG signals to obtain amplified EEG signals, and transmit the amplified EEG signals to the EEG analysis unit 102; the EEG analysis unit 102 analyzes and processes the amplified EEG signals to obtain an evaluation result of the motor imagery ability of the target object, and further generates a rehabilitation training plan for the target object in combination with the evaluation result of the motor imagery ability of the target object and the motor disorder level, and sends the rehabilitation training plan to a rehabilitation robot, so that the rehabilitation robot can assist the target object in performing rehabilitation training according to the rehabilitation training plan.

[0086] Optionally, the brain-computer interface device may be a processing module integrated on the rehabilitation robot, or the brain-computer interface device and the rehabilitation robot are two separate modules, that is, the brain-computer interface device can be communicatively connected to the rehabilitation robot through a control interface on the rehabilitation robot.

[0087] Among them, referring to Figure 2 as shown, the EEG acquisition unit 101 mainly consists of an EEG amplifier, acquisition electrodes, an EEG cap, a power supply, and wires.

[0088] The EEG cap is a fixing device mainly used to make the acquisition electrodes fully contact with the scalp of the target object to acquire EEG signals with good signal quality.

[0089] Exemplarily, for example, the EEG amplifier may be an ADS1299 chip, the acquisition electrodes are 3-channel dry electrodes, and the placement of the electrodes follows the international 10-20 system, and their positions are C3, Cz, and C4, that is, mainly near the motor imagery area of the target object's brain, and the EEG sampling frequency is 250 Hz.

[0090] Continuing to refer to Figure 2 as shown, the power supply is electrically connected to the EEG amplifier and each acquisition electrode respectively, and is used to supply electrical energy to the EEG amplifier and the acquisition electrodes. Exemplarily, for example, the power supply may be a 6V dry battery.

[0091] Each acquisition electrode is communicatively connected to the EEG amplifier. The EEG signals of the target object can be acquired through each acquisition electrode, and the acquired weak EEG signals are transmitted to the EEG amplifier through wires. The EEG amplifier amplifies the EEG signals to obtain amplified EEG signals, and then, the amplified EEG signals are output toFigure 1 The electroencephalogram analysis unit in

[0092] It can be understood that Figure 1 The structure described above is only schematic, and the brain-computer interface device may also include more or fewer components than those shown in Figure 1 or have a different configuration from that shown in Figure 1 The components shown in Figure 1 can be implemented by hardware, software, or a combination thereof.

[0093] Referring to Figure 3 shown, it is Figure 1 a schematic structural diagram of the brain-computer interface device in Figure 3 The brain-computer interface device may have a processing device with data analysis functions to implement the processing method of the rehabilitation training plan provided in this application. As

[0094] shown, the brain-computer interface device includes: a processor 301 and a memory 302.

[0095] The processor 301 and the memory 302 are directly or indirectly electrically connected to achieve data transmission or interaction. For example, they can be electrically connected through one or more communication buses or signal lines.

[0096] Among them, the processor 301 can be an integrated circuit chip with signal processing capabilities. The above-mentioned processor 301 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc. It can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0097] The memory 302 can be, but is not limited to, a random access memory (RAM), a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), etc.

[0098] The following will illustrate the processing method of the rehabilitation training plan provided by this application and the corresponding beneficial effects through multiple embodiments.

[0099] Figure 4 As shown in the flowchart of a processing method for a rehabilitation training plan provided by an embodiment of this application. Optionally, the execution subject of this method can be Figure 1 the brain-computer interface device shown in, which has data processing capabilities.

[0100] It should be understood that in other embodiments, the order of some steps in the processing method of the rehabilitation training plan can be exchanged according to actual needs, or some of the steps can also be omitted or deleted. As Figure 4 shown, this method includes:

[0101] S401. Obtain the electroencephalogram (EEG) signals of the target object when performing multiple motor imagery tasks.

[0102] Among them, the EEG signals are collected by the EEG acquisition unit on the brain-computer interface device.

[0103] In this embodiment, in order to improve the accuracy of the evaluation result of the target object's motor imagery ability, it is necessary to obtain the EEG signals of the target object collected by the EEG acquisition unit on the brain-computer interface device when performing multiple motor imagery tasks (such as imagining lifting the leg) according to a certain process.

[0104] Optionally, when evaluating the motor imagery ability of the target object, this application uses 3-channel dry electrodes to collect the EEG signals of the target object. Compared with wet electrodes and multi-channels, the acquisition method of this solution is more convenient and fast.

[0105] S402. Preprocess the EEG signals to obtain the preprocessed EEG signals.

[0106] Optionally, in order to remove the low-frequency noise interference in the EEG signals, a filter can be used to preprocess the EEG signals to obtain the preprocessed EEG signals.

[0107] S403. Extract features from the preprocessed EEG signals to obtain the feature matrix of the preprocessed EEG signals.

[0108] It should be understood that there are mainly three types of features of EEG signals: time-domain features, frequency-domain features, and spatial-domain features. Different features require different feature extraction methods.

[0109] In this embodiment, for example, Fourier transform, wavelet transform, or autoregressive (AR) model can be used to extract features from the preprocessed EEG signals to obtain the feature matrix of the preprocessed EEG signals, that is, the feature matrix of the preprocessed EEG signals is a frequency-domain feature.

[0110] S404. Input the feature matrix into the pre-trained target classification model to obtain the evaluation result of the target object's motor imagery ability.

[0111] Exemplarily, for example, the classification model can be a CART decision tree classification model, a neural network classifier (NeuralNetwork Nlassifier, abbreviated as NNC), a Bayesian classifier (Bayesian classifier, abbreviated as BC), etc.

[0112] Among them, the evaluation results of the motor imagery ability include: poor, average, and good.

[0113] In this embodiment, for example, the feature matrix can be input into the pre-trained CART decision tree classification model to obtain the classification accuracy output by the CART decision tree classification model, that is, the current motor imagery ability score of the target object = the classification accuracy of the CART decision tree classification model * 100. Then, according to the interval section where the motor imagery ability score is located, the evaluation result of the motor imagery ability is determined. Among them, the motor imagery ability score is 0 - 60, and the motor imagery ability evaluation result is poor; the motor imagery ability score is 60 - 80, and the motor imagery ability is average; the motor imagery ability score is 80 - 100, and the motor imagery ability is good.

[0114] For example, calculate the motor imagery ability score of patient A = the classification accuracy of the CART decision tree classification model * 100 = 86, then the evaluation result of patient A's motor imagery ability is good.

[0115] Optionally, in order to improve the accuracy of the evaluation result of the target object's motor imagery ability, the feature matrix can be input into the pre-trained target classification model multiple times, and the evaluation results of each time are summarized, and the multiple classification accuracies of the classification model are averaged to obtain the average classification accuracy.

[0116] S405. Generate a rehabilitation training plan for the target object according to the evaluation result of the target object's motor imagery ability and the motor disorder level.

[0117] Among them, the motor disorder levels include: level 0, level 1, and level 2. Among them, level 0 means that the patient's muscles are completely paralyzed, without any contraction and no movement; level 1 means that the patient's muscles have slight contractions, but cannot drive joint activities, and slight muscle contractions can be seen; level 2 means that the patient can drive joint horizontal activities, but cannot resist gravity, and the limb can move parallel on the bed.

[0118] The rehabilitation training plan refers to the number of training days, the number of training times per day, and the single training time during the rehabilitation training period of the target object.

[0119] In this embodiment, in order to improve the rehabilitation efficiency of patients and save their rehabilitation time, a rehabilitation training plan for the target object is generated by combining the evaluation results of the target object's motor imagery ability and the motor disorder level. For example, if the evaluation result of patient A's motor imagery ability is average and the motor disorder level is level 1, then patient A's rehabilitation training plan is that the training duration is 20 days, the number of training sessions per day is 6 times, and the duration of each training session is 9 minutes.

[0120] Table 1 shows the rehabilitation training plan for patients

[0121]

[0122] In this application, based on the evaluation results of the target object's motor imagery ability and the motor disorder level, a personalized one-cycle motor rehabilitation training plan for the patient is customized. For example, the higher the motor disorder level, the longer the training duration of a single training cycle; the better the motor imagery ability, the more the number of training sessions per day and the duration of each training session can be appropriately reduced. After completing the training plan for one rehabilitation cycle, the rehabilitation effect of the patient is evaluated. If the patient still needs to carry out the next rehabilitation plan, the motor imagery ability of the patient is evaluated again. After one cycle of rehabilitation training, the patient's motor imagery ability is improved to a certain extent. Then, a new cycle training plan is customized and the plan is executed, and this process is repeated until the rehabilitation effect reaches satisfaction.

[0123] Compared with the traditional rehabilitation training method, this solution proposes to generate a rehabilitation plan based on the evaluation results of the patient's motor imagery ability and the motor disorder level, that is, the generated rehabilitation plan is specifically tailored for the patient. That is, the rehabilitation plan has the characteristics of pertinence, scientificity, and rationality, so that the rehabilitation robot can help the patient carry out rehabilitation training targeted according to this rehabilitation plan, greatly improving the rehabilitation efficiency of the patient and saving the patient's rehabilitation time. At the same time, the use efficiency of the rehabilitation robot is improved, that is, the rehabilitation effect of the motor imagery therapy proposed in this application is better than the traditional rehabilitation method.

[0124] In summary, the embodiment of the present application provides a method for processing a rehabilitation training plan. Based on the EEG signals of the target object, an evaluation result of the motor imagery ability of the target object is obtained, that is, this solution proposes to quantitatively evaluate the motor imagery ability of different patients. Then, aiming at the two indicators of the evaluation result of the patient's motor imagery ability and the motor disorder level, a rehabilitation training plan is personalized for the patient. That is, the rehabilitation plan is tailored specifically for the target object, and the rehabilitation plan has the characteristics of pertinence, scientificity and rationality, so that the rehabilitation robot can help the patient carry out rehabilitation training targeted according to the rehabilitation plan, greatly improving the patient's rehabilitation efficiency and saving the patient's rehabilitation time. At the same time, the use efficiency of the rehabilitation robot is improved. That is, the rehabilitation effect of the motor imagery therapy proposed in the present application is better than the traditional rehabilitation method, and the problem of unreasonable formulation in existing rehabilitation training is solved.

[0125] Optionally, input the feature matrix into a pre-trained target classification model to obtain the evaluation result of the motor imagery ability of the target object, including:

[0126] Input the feature matrix into the target classification model, obtain the classification result output by the target classification model, and use the classification result as the evaluation result of the motor imagery ability of the target object.

[0127] Among them, the classification result is the classification accuracy of the target classification model. For example, the classification result is 68%, that is, the evaluation result of the motor imagery ability of the target object is 68 points.

[0128] In an implementable manner, in order to improve the accuracy of the quantitative evaluation of the patient's motor imagery ability, the feature matrix of the patient's EEG signals can also be directly input into the target classification model to obtain the classification accuracy output by the target classification model. Then, use this classification accuracy as the evaluation result of the motor imagery ability of the target object, realizing the numerical quantitative evaluation of the patient's motor imagery ability, so as to more scientifically customize the patient's motor rehabilitation training plan based on the numerical quantitative result of the patient's motor imagery ability and improve the rehabilitation training effect.

[0129] Optionally, as shown in Figure 5 the above step S403 includes:

[0130] S501. Use a preset window function to segment the preprocessed EEG signals to obtain at least one segment signal.

[0131] Exemplarily, for example, the preset window function can be a Hamming window, that is, the preprocessed EEG signals can be truncated by the Hamming window to obtain at least one segment signal.

[0132] S502. Feature extraction is performed on each fragment signal to obtain a feature matrix corresponding to each fragment signal.

[0133] Optionally, in order to obtain more potential useful information in the EEG signal and improve the calculation speed of subsequent algorithms at the same time, the present application proposes to segment the preprocessed EEG signal to obtain at least one fragment signal. In this way, the preprocessed EEG signal can be made smooth in the time domain and the spectral leakage phenomenon can be reduced to a certain extent.

[0134] In this embodiment, for example, the preprocessed EEG signal can be multiplied point by point with a rectangular window, that is, the 2s EEG signal is taken as one segment, and the next segment is taken every 0.2s to obtain at least one fragment signal; then, feature extraction is performed on each fragment signal respectively to obtain a feature matrix corresponding to each fragment signal.

[0135] Optionally, referring to Figure 6 As shown, the above step S502 includes:

[0136] S601. Perform wavelet packet transform on the fragment signal to obtain multiple wavelet packet coefficients of the fragment signal.

[0137] S602. Select the first M wavelet packet coefficients among the multiple wavelet packet coefficients, and use the first M wavelet packet coefficients as the input parameters of the filter bank common spatial pattern algorithm, and obtain the feature matrix of the fragment signal based on the filter bank common spatial pattern algorithm.

[0138] In this embodiment, when performing feature extraction on the EEG signal, the main method used is the combination of wavelet packet transform and the filter bank common spatial pattern algorithm.

[0139] Among them, wavelet packet transform (WPT) can perform hierarchical approximation analysis on the signal according to the frequency level. First, the original signal is decomposed by wavelet packet to obtain a high-frequency detail part and a low-frequency approximation part, and then the high-frequency part and the low-frequency part are respectively decomposed again, and so on; finally, the wavelet packet decomposition results of each layer are obtained.

[0140] The main principle of the common spatial pattern algorithm (CSP) is to project the EEG signal between different categories, so that the variance of the EEG signal of one category is the largest and the variance of the EEG signal of the other category is the smallest under certain projections, so as to achieve the purpose of classification. Through such projections, the CSP filter can make the EEG signals of different categories have the greatest difference on specific channels. The filter bank common spatial pattern algorithm (FBCSP) uses multiple band-pass filters to select distinguishable CSP features.

[0141] Specifically, for example, referring to Figure 7As shown, wavelet packet transform is performed on each segment signal to obtain multiple wavelet packet coefficients of each segment signal (such as f11, f12, …, f58); then, the first 8 sub-coefficients are selected as 8 sub-bands of the filter bank common spatial pattern algorithm, namely CSP1 to CSP8, where the frequency band ranges of the first 8 sub-bands are different, equivalent to 8 different band-pass filters; then, the first 8 wavelet packet coefficients are used as the input parameters of the filter bank common spatial pattern algorithm, and the dimension m of the filter in the filter bank common spatial pattern algorithm is set to 6, that is, 12 CSP feature matrices of each segment signal can be extracted based on the filter bank common spatial pattern algorithm. Finally, the feature matrices of each segment signal and the classification labels of each segment signal are combined into a data set.

[0142] Optionally, the above step S601 includes:

[0143] Perform N-layer wavelet packet decomposition on the segment signal to obtain 2^N wavelet packet coefficients of the segment signal.

[0144] Exemplarily, for example, N-layer is 5 layers.

[0145] In this embodiment, for example, reference can be continued to Figure 7 As shown, 5-layer wavelet packet decomposition can be performed on each segment signal respectively, that is, 2^5 = 32 wavelet packet coefficients, namely f51, f52, …, f532, can be obtained at the 5th layer.

[0146] The following embodiments will specifically explain how to train a target classification model based on a large amount of sample data.

[0147] Optionally, referring to Figure 8 As shown, before the above step S404, it further includes:

[0148] S801. Use a plurality of pre-constructed sample data to obtain the sample feature matrix corresponding to each sample data.

[0149] Among them, each sample data includes electroencephalogram signals in the resting state or electroencephalogram signals of motor imagery, and the classification label corresponding to the electroencephalogram signal.

[0150] In this embodiment, a large number of stroke patients with different degrees are selected. When each patient performs multiple motor imagery (such as imagining lifting the leg) according to a certain process, the electroencephalogram signals in the resting state and the electroencephalogram signals during motor imagery are collected through the electroencephalogram acquisition unit in the brain-computer interface device, that is, a plurality of sample data are obtained; then, the obtained electroencephalogram signals are marked as the resting state (clam) and the motor imagery state (MI), that is, the classification label corresponding to each electroencephalogram signal is obtained; finally, wavelet packet transform combined with the filter bank common spatial pattern algorithm is used to extract features from each sample data respectively to obtain the sample feature matrix corresponding to each sample data.

[0151] S802. Combine the sample feature matrix corresponding to the sample data with the classification label corresponding to the sample data to form a training data set.

[0152] S803. Train a target classification model based on the training data set.

[0153] In this embodiment, for example, combine the sample feature matrix of each sample data with the classification label corresponding to each sample data to form a training data set. Then, use the training data set to iteratively train the initial classification model until the error rate of the finally trained classification model is less than the preset error, then end the training, and use the classification model obtained in the last training as the target classification model.

[0154] Optionally, as shown in Figure 9 Step S803 includes:

[0155] S901. Divide the training data set to obtain multiple sub-data sets.

[0156] S902. Use one of the multiple sub-data sets as the validation sub-data set, and use the sub-data sets other than the validation sub-data set as the training sub-data set.

[0157] In this embodiment, in order to ensure the robustness of the finally trained target classification model and make the generalization ability of the target classification model stronger, it is proposed that the training data set can be divided into sub-data sets according to random sampling. For example, divide the training data set into 5 sub-data sets, and use 4 of the sub-data sets as the training sub-data sets, and the remaining one sub-data set as the validation sub-data set.

[0158] S903. Iteratively train the constructed initial model based on the training sub-data set, and iteratively validate the trained initial model based on the validation sub-data set, and use the initial model that meets the preset validation conditions as the target classification model.

[0159] Exemplarily, for example, the selected initial model can be a CART decision tree classification model.

[0160] In this embodiment, use the 4 training sub-data sets obtained by the above division to iteratively train the constructed initial model, and after each round of training, use the validation sub-data set to verify the classification accuracy of the initial model, that is, evaluate the model performance, until after multiple rounds of iterative training, if the accuracy of the trained initial model meets the preset validation conditions, then use the initial model obtained after the last round of iterative training as the target classification model.

[0161] In another implementable manner, for example, after five rounds of iteration are completed, the evaluation results of each round are aggregated, and the classification accuracies of the five iterations are averaged to obtain an average classification accuracy. If the average classification accuracy meets the preset verification condition, the initial model obtained after the last round of iterative training is used as the target classification model.

[0162] Optionally, the target classification model includes: a support vector machine binary classification model, a decision tree classification model, a Bayesian classifier, or a neural network classifier.

[0163] Optionally, referring to Figure 10 as shown, the above step S402 includes:

[0164] S1001. Filter the electroencephalogram (EEG) signal using a first non-recursive filter to obtain a filtered EEG signal.

[0165] Exemplarily, for example, the high-pass cut-off frequency of the first non-recursive filter is 5 Hz and the low-pass cut-off frequency is 30 Hz.

[0166] S1002. Filter the filtered EEG signal using a second non-recursive filter to obtain a preprocessed EEG signal.

[0167] Exemplarily, for example, the cut-off frequency of the second non-recursive filter is 50 Hz.

[0168] In this embodiment, a first non-recursive filter with a high-pass cut-off frequency of 5 Hz and a low-pass cut-off frequency of 30 Hz can be used to filter the EEG signal to obtain an EEG signal in the frequency band related to motor imagery; then, a second non-recursive filter with a cut-off frequency of 50 Hz is used to filter the EEG signal in the frequency band related to motor imagery to remove power frequency interference and obtain a preprocessed EEG signal. In this way, it can be ensured that the preprocessed EEG signal is an EEG signal in the frequency band related to motor imagery and is free from low-frequency noise interference such as electrooculogram and electromyogram, so as to facilitate improving the accuracy of the processing result of the EEG signal in the subsequent process.

[0169] Optionally, referring to Figure 11 as shown, it is a schematic diagram of the overall process of the rehabilitation training plan processing method provided by the embodiment of the present application; as Figure 11 shown, the method includes:

[0170] S1101. Obtain the EEG signal when the target object performs multiple motor imagery, and preprocess the EEG signal to obtain a preprocessed EEG signal.

[0171] S1102. Use a preset window function to segment the preprocessed EEG signal to obtain at least one segment signal.

[0172] S1103. Perform wavelet packet transform on each segment signal for N layers to obtain 2^N wavelet packet coefficients of each segment signal.

[0173] S1104. Select the first M wavelet packet coefficients from the 2^N wavelet packet coefficients, and use the first M wavelet packet coefficients as the input parameters of the filter bank common spatial pattern algorithm, and obtain the feature matrix of each segment signal based on the filter bank common spatial pattern algorithm.

[0174] S1105. Input the feature matrix of each segment signal into the pre-trained target classification model to obtain the evaluation result of the target object's motor imagery ability.

[0175] S1106. Generate a rehabilitation training plan for the target object according to the evaluation result of the target object's motor imagery ability and the level of motor impairment.

[0176] Optionally, the specific implementation steps and beneficial effects of this rehabilitation training plan processing method have been described in detail in the previous specific embodiments, and will not be elaborated here one by one.

[0177] The following describes a processing device for executing the rehabilitation training plan provided in this application, etc. The specific implementation process and technical effects are as described above, and will not be elaborated below.

[0178] Figure 12 It is a schematic structural diagram of a processing device for a rehabilitation training plan provided by an embodiment of this application; it is applied to a brain-computer interface device, such as Figure 12 shown, the device includes:

[0179] An acquisition module 1201, configured to acquire the electroencephalogram signals when the target object performs multiple motor imagery, and the electroencephalogram signals are acquired by an electroencephalogram acquisition unit on the brain-computer interface device;

[0180] A processing module 1202, configured to preprocess the electroencephalogram signals to obtain preprocessed electroencephalogram signals;

[0181] An extraction module 1203, configured to extract features from the preprocessed electroencephalogram signals to obtain the feature matrix of the preprocessed electroencephalogram signals;

[0182] The processing module 1202 is further configured to input the feature matrix into the pre-trained target classification model to obtain the evaluation result of the target object's motor imagery ability;

[0183] A generation module 1204, configured to generate a rehabilitation training plan for the target object according to the evaluation result of the target object's motor imagery ability and the level of motor impairment.

[0184] Optionally, the extraction module 1203 is further configured to:

[0185] Using a preset window function, segment the preprocessed EEG signals to obtain at least one segment signal;

[0186] Extract features from each segment signal to obtain a feature matrix corresponding to each segment signal.

[0187] Optionally, the extraction module 1203 is further configured to:

[0188] Perform wavelet packet transform on the segment signal to obtain multiple wavelet packet coefficients of the segment signal;

[0189] Select the first M wavelet packet coefficients from the multiple wavelet packet coefficients, and use the first M wavelet packet coefficients as input parameters of the filter bank common spatial pattern algorithm, and obtain the feature matrix of the segment signal based on the filter bank common spatial pattern algorithm.

[0190] Optionally, the extraction module 1203 is further configured to:

[0191] Perform N - layer wavelet packet decomposition on the segment signal to obtain 2^N wavelet packet coefficients of the segment signal.

[0192] Optionally, the extraction module 1203 is further configured to use a plurality of pre - constructed sample data to obtain a sample feature matrix corresponding to each sample data, where each sample data includes EEG signals in a resting state or EEG signals of motor imagery, and a classification label corresponding to the EEG signal;

[0193] The apparatus further includes:

[0194] A composition module, configured to compose the sample feature matrix corresponding to the sample data and the classification label corresponding to the sample data into a training data set;

[0195] A training module, configured to train a target classification model based on the training data set.

[0196] Optionally, the training module is further configured to:

[0197] Divide the training data set to obtain a plurality of sub - data sets;

[0198] Use one of the plurality of sub - data sets as a validation sub - data set, and use the sub - data sets other than the validation sub - data set as training sub - data sets;

[0199] Iteratively train the constructed initial model based on the training sub - data set, and iteratively validate the trained initial model based on the validation sub - data set, and use the initial model that meets the preset validation conditions as the target classification model.

[0200] Optionally, the processing module 1202 is further configured to:

[0201] Filter the electroencephalogram (EEG) signal using a first non-recursive filter to obtain the filtered EEG signal;

[0202] Filter the filtered EEG signal using a second non-recursive filter to obtain the preprocessed EEG signal.

[0203] The above device is used to execute the method provided in the foregoing embodiment, and its implementation principle and technical effects are similar, so details are not described herein again.

[0204] The above modules can be one or more integrated circuits configured to implement the above method. For example: one or more application specific integrated circuits (ASICs), or, one or more digital signal processors (DSPs), or, one or more field programmable gate arrays (FPGAs), etc. Again, when a certain above module is implemented in the form of a processing element scheduling program code, the processing element can be a general-purpose processor, such as a central processing unit (CPU) or other processors that can call program code. Again, these modules can be integrated together and implemented in the form of a system-on-a-chip (SOC).

[0205] Optionally, the present invention further provides a program product, such as a computer-readable storage medium, including a program that is used to execute the above method embodiment when executed by a processor.

[0206] In several embodiments provided by the present invention, it should be understood that the disclosed device and method can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection to each other can be through some interfaces, and the indirect coupling or communication connection of the device or unit can be in an electrical, mechanical or other form.

[0207] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place, or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0208] In addition, each functional unit in various embodiments of the present invention may be integrated into one processing unit, may exist separately as individual physical units, or two or more units may be integrated into one unit. The above-mentioned integrated units may be implemented in the form of hardware, or in the form of a combination of hardware and software functional units.

[0209] The above-mentioned integrated units implemented in the form of software functional units may be stored in a computer-readable storage medium. The above-mentioned software functional units stored in a storage medium include several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor (English: processor) to execute some steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (English: Read-Only Memory, abbreviated as: ROM), random access memories (English: Random Access Memory, abbreviated as: RAM), magnetic disks, or optical discs that can store program codes.

Claims

1. A processing method for a rehabilitation training plan, characterized in that, Applied to a brain-computer interface device, the method includes: Obtaining electroencephalogram (EEG) signals when a target object performs multiple motor imagery tasks, where the EEG signals are collected by an EEG acquisition unit on the brain-computer interface device; Preprocessing the EEG signals to obtain preprocessed EEG signals; Extracting features from the preprocessed EEG signals to obtain a feature matrix of the preprocessed EEG signals; Inputting the feature matrix into a pre-trained target classification model to obtain an evaluation result of the motor imagery ability of the target object; Generating a rehabilitation training plan for the target object according to the evaluation result of the motor imagery ability of the target object and the motor disorder level.

2. The method according to claim 1, wherein The step of inputting the feature matrix into a pre-trained target classification model to obtain an evaluation result of the motor imagery ability of the target object includes: Inputting the feature matrix into the target classification model to obtain a classification result output by the target classification model, and using the classification result as the evaluation result of the motor imagery ability of the target object.

3. The method according to claim 1, wherein The step of extracting features from the preprocessed EEG signals to obtain a feature matrix includes: Using a preset window function to segment the preprocessed EEG signals to obtain at least one segment signal; Extracting features from each of the segment signals respectively to obtain a feature matrix corresponding to each of the segment signals.

4. The method according to claim 3, wherein The step of extracting features from each of the segment signals respectively to obtain a feature matrix corresponding to each of the segment signals includes: Performing wavelet packet transform on the segment signal to obtain multiple wavelet packet coefficients of the segment signal; Selecting the first M wavelet packet coefficients from the multiple wavelet packet coefficients, and using the first M wavelet packet coefficients as input parameters of the filter bank common spatial pattern algorithm, and obtaining a feature matrix of the segment signal based on the filter bank common spatial pattern algorithm.

5. The method according to claim 4, wherein The step of performing wavelet packet transform on the segment signal to obtain multiple wavelet packet coefficients of the segment signal includes: Performing N-layer wavelet packet decomposition on the segment signal to obtain 2^N wavelet packet coefficients of the segment signal.

6. The method according to claim 1, wherein Before inputting the feature matrix into a pre-trained target classification model to obtain an evaluation result of the motor imagery ability of the target object, it further includes: Using a plurality of pre-constructed sample data to obtain a sample feature matrix corresponding to each sample data, where each sample data includes EEG signals in a resting state or EEG signals of motor imagery, and a classification label corresponding to the EEG signals; Combining the sample feature matrix corresponding to the sample data with the classification label corresponding to the sample data to form a training data set; Training the target classification model based on the training data set.

7. The method according to claim 6, wherein The step of training the target classification model based on the training data set includes: Dividing the training data set to obtain a plurality of sub-data sets; Using one of the plurality of sub-data sets as a validation sub-data set, and using the sub-data sets other than the validation sub-data set as training sub-data sets; Iteratively train the constructed initial model based on the training sub-dataset, and iteratively validate the trained initial model based on the validation sub-dataset, and use the initial model that meets the preset validation conditions as the target classification model.

8. The method according to claim 7, wherein The target classification model includes: a support vector machine binary classification model, a decision tree classification model, a Bayesian classifier, or a neural network classifier.

9. The method according to claim 1, characterized in that, The preprocessing of the electroencephalogram signal to obtain the preprocessed electroencephalogram signal includes: Filter the electroencephalogram signal using a first non-recursive filter to obtain a filtered electroencephalogram signal; Filter the filtered electroencephalogram signal using a second non-recursive filter to obtain the preprocessed electroencephalogram signal.

10. A brain-computer interface device, characterized in that, It includes: A processor, a storage medium, and a bus. The storage medium stores machine-readable instructions executable by the processor. When the electronic device runs, the processor communicates with the storage medium through the bus, and the processor executes the machine-readable instructions to perform the steps of the method according to any one of claims 1-9.

11. A computer-readable storage medium, characterized in that, A computer program is stored on the storage medium, and when the computer program is run by the processor, it executes the method according to any one of claims 1-9.

Images