Orthopedic rehabilitation training system and method

Abstract

The invention provides an orthopedic rehabilitation training system and method. The method comprises the following steps: acquiring original electromyographic signals and three-dimensional motion data of limbs when a target patient executes orthopedic standard rehabilitation actions; determining an activation time sequence mode of neuromuscles according to the original electromyographic signals, and determining multi-modal motion features based on the activation time sequence mode and the three-dimensional kinematics data; determining a motion function index of the skeletal muscle system in the current rehabilitation period based on the multi-modal motion characteristics, and determining a function recovery track of the skeletal muscle system of the target patient in different training load scenes in the next rehabilitation period; and generating a personalized rehabilitation instruction set according to the motion function index and all the function recovery tracks, and adjusting auxiliary torque output of the rehabilitation training equipment based on the personalized rehabilitation instruction set. By adopting the scheme of the invention, the recovery tracks under different training loads can be predicted through the multi-modal motion characteristics of the patient so as to dynamically generate the personalized instruction set to assist the patient in rehabilitation training.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of motion measurement, more particularly, the present application relates to an orthopedic rehabilitation training system and method. BACKGROUND

[0002] With the development of motion measurement technology and intelligent rehabilitation equipment, in the field of orthopedic rehabilitation, the core demand of postoperative muscle-bone system function recovery of patients is to realize the closed-loop adaptation of evaluation-intervention-adjustment, and this process relies on the cooperative capture of neuromuscular control state and joint motion function to realize the precise quantification of the current function state of the patient, and to provide scientific support for subsequent training scheme formulation and equipment intervention.

[0003] However, the implementation of the existing orthopedic rehabilitation training scheme has obvious limitations: first, the functional evaluation relies on the subjective experience of doctors or single-dimensional data (such as only measuring joint range of motion), without integrating the neuromuscular activation timing information in the electromyographic signal, resulting in the inability to accurately identify the different functional short boards of patients (such as muscle strength deficiency or neuromuscular control disorder), and the evaluation results are one-sided; second, the training load selection adopts a fixed gradient mode, without dynamically predicting the recovery trajectory under different loads based on the individual function of the patient, which is prone to problems such as poor fracture healing due to overloading or low rehabilitation efficiency due to low loading; third, the rehabilitation device auxiliary torque adjustment is preset parameters, which cannot dynamically adapt to the real-time training state of the patient (such as insufficient force and motion deviation) and individual needs, and the intervention precision is insufficient, therefore, how to predict the recovery trajectory under different training loads through the multi-modal motion features of the patient to dynamically generate individualized instruction sets to assist the patient in rehabilitation training has become a difficult problem in the industry. SUMMARY

[0004] The present application provides an orthopedic rehabilitation training system and method, which can predict the recovery trajectory under different training loads through the multi-modal motion features of the patient to dynamically generate individualized instruction sets to assist the patient in rehabilitation training.

[0005] In a first aspect, the present application provides an orthopedic rehabilitation training method based on individual function dynamic evaluation and prediction, comprising: Collecting the original electromyographic signal and three-dimensional motion data of the limb of the target patient when performing the orthopedic standard rehabilitation action; Extracting the discharge sequence of the motor unit in the original electromyographic signal for time-frequency domain fusion to obtain the activation timing pattern of the neuromuscular system of the target patient when performing the orthopedic rehabilitation action, and then determining the multi-modal motion features of the target patient when performing the orthopedic rehabilitation training based on the activation timing pattern and the joint activity characteristics of the target patient in the three-dimensional kinematic data; perform individualized function prediction based on the multi-modal motion feature to obtain a motion function index of a skeletal muscle system of the target patient in a current rehabilitation cycle when the target patient performs orthopedic rehabilitation training, and determine a function recovery trajectory of the skeletal muscle system of the target patient in different training load scenarios in a next rehabilitation cycle; generate a personalized rehabilitation instruction set matched with a current rehabilitation state of the target patient according to the motion function index and all function recovery trajectories, and then adjust an auxiliary torque output of a rehabilitation training device when the target patient performs orthopedic rehabilitation training based on the personalized rehabilitation instruction set.

[0006] In some embodiments, the discharge sequence of the motor unit is extracted from the original electromyography signal for time-frequency domain fusion to obtain an activation timing pattern of a neuromuscular system when the target patient performs orthopedic rehabilitation actions, and the activation timing pattern specifically includes: The discharge sequence of each motor unit is separated from the original electromyography signal; The discharge sequence of all motor units is subjected to continuous wavelet transform, and then the energy distribution features of each discharge sequence in the time-frequency domain are extracted; All energy distribution features are synergistically fused to obtain the activation timing pattern of the neuromuscular system when the target patient performs orthopedic rehabilitation actions.

[0007] In some embodiments, the multi-modal motion feature of the target patient when performing orthopedic rehabilitation training is determined based on the activation timing pattern and joint activity features of the target patient in the three-dimensional kinematic data, and the multi-modal motion feature specifically includes: The joint activity features of the target patient when performing orthopedic rehabilitation training are extracted from the three-dimensional kinematic data; The spatiotemporal coupling relationship between neuromuscular activation and joint movement of the target patient when performing orthopedic rehabilitation training is determined; The activation timing pattern and the joint activity features are feature-cascaded fused based on the spatiotemporal coupling relationship to obtain the multi-modal motion feature of the target patient when performing orthopedic rehabilitation training.

[0008] In some embodiments, the multi-modal motion feature is used to perform individualized function prediction to obtain a motion function index of a skeletal muscle system of the target patient in a current rehabilitation cycle when the target patient performs orthopedic rehabilitation training, and the motion function index specifically includes: The historical clinical data of the target patient is obtained; The clinical baseline features of the target patient are extracted from the historical clinical data; A muscle-bone system function prediction network of the target patient when performing orthopedic rehabilitation training is constructed based on the clinical baseline features and the multi-modal motion feature; The motion function index of the skeletal muscle system of the target patient in the current rehabilitation cycle when performing orthopedic rehabilitation training is determined according to the muscle-bone system function prediction network.

[0009] In some embodiments, determining the functional recovery trajectory of the skeletal muscle system of the target patient under different training load scenarios in the next rehabilitation cycle specifically comprises: determining different training load scenarios for the target patient in the next rehabilitation cycle; predicting the recovery curves of the skeletal muscle system of the target patient under different training load scenarios based on the musculoskeletal system function prediction network; performing curve smoothing on all the recovery curves to obtain the functional recovery trajectory of the skeletal muscle system of the target patient under different training load scenarios in the next rehabilitation cycle.

[0010] In some embodiments, generating a personalized rehabilitation instruction set matching the current rehabilitation state of the target patient according to the motor function index and all the functional recovery trajectories specifically comprises: performing performance quantization on all the functional recovery trajectories to generate a performance ranking table of the training load scenarios; constructing a load adaptation matrix of the motor function of the target patient; filtering out an adapted training load scenario from the performance ranking table of the training load scenarios through the motor function index and the load adaptation matrix; generating a personalized rehabilitation instruction set matching the current rehabilitation state of the target patient according to the adapted training load scenario.

[0011] In some embodiments, the rehabilitation training device comprises a torque adjustment component and a monitoring feedback component.

[0012] In a second aspect, the present application provides an orthopedic rehabilitation training system, comprising: a collection module configured to collect original electromyographic signals and three-dimensional kinematic data of a limb of a target patient performing orthopedic standard rehabilitation actions; a processing module configured to extract discharge sequences of motor units from the original electromyographic signals for time-frequency domain fusion, to obtain an activation timing pattern of neuromuscular of the target patient performing orthopedic rehabilitation actions, and to determine a multi-modal motion feature of the target patient performing orthopedic rehabilitation training based on the activation timing pattern and joint activity characteristics of the target patient in the three-dimensional kinematic data; the processing module is configured to perform individualized function prediction based on the multi-modal motion feature, to obtain a motor function index of the skeletal muscle system of the target patient performing orthopedic rehabilitation training in the current rehabilitation cycle, and to determine a functional recovery trajectory of the skeletal muscle system of the target patient under different training load scenarios in the next rehabilitation cycle; The execution module is configured to generate a personalized rehabilitation instruction set matched with the current rehabilitation state of the target patient according to the motor function index and all the function recovery trajectories, and to adjust the auxiliary torque output of the rehabilitation training device when the target patient performs the orthopedic rehabilitation training based on the personalized rehabilitation instruction set.

[0013] In a third aspect, the present application provides a computer device, which comprises a memory and a processor, the memory stores a code, and the processor is configured to acquire the code and execute the orthopedic rehabilitation training method based on dynamic evaluation and prediction of individual function.

[0014] In a fourth aspect, the present application provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the orthopedic rehabilitation training method based on dynamic evaluation and prediction of individual function.

[0015] The technical scheme provided by the embodiments of the present application has the following beneficial effects: In the orthopedic rehabilitation training system and method provided by the present application, the original electromyographic signal and the three-dimensional kinematic data of the limb of the target patient performing the orthopedic standard rehabilitation action are first collected; the discharge sequence of the motor unit in the original electromyographic signal is extracted for time-frequency domain fusion to obtain the activation timing pattern of the neuromuscular system of the target patient performing the orthopedic rehabilitation action, and then the multi-modal motion feature of the target patient performing the orthopedic rehabilitation training is determined based on the activation timing pattern and the joint activity feature of the target patient in the three-dimensional kinematic data; the individualized function prediction is performed based on the multi-modal motion feature to obtain the motor function index of the skeletal muscle system of the target patient in the current rehabilitation cycle when performing the orthopedic rehabilitation training, and the function recovery trajectory of the skeletal muscle system of the target patient in different training load scenarios in the next rehabilitation cycle is determined; a personalized rehabilitation instruction set matched with the current rehabilitation state of the target patient is generated according to the motor function index and all the function recovery trajectories, and then the auxiliary torque output of the rehabilitation training device when the target patient performs the orthopedic rehabilitation training is adjusted based on the personalized rehabilitation instruction set.

[0016] It can be seen that the application adjusts the assistive torque output of the rehabilitation training equipment when the target patient performs orthopedic rehabilitation training based on the personalized rehabilitation instruction set; first, the multi-modal motion feature is determined to obtain a feature vector reflecting the neuromuscular activation law and joint motion characteristics of the target patient during rehabilitation training; the determination of the multi-modal motion feature can break through the limitation of single data dimension by fusing the neuromuscular activation timing pattern and the joint activity characteristics, accurately identify the dysfunction subtype (such as abnormal neuromuscular activation or muscle insufficiency), and provide objective feature input in the neuromotor dimension for subsequent individualized function prediction, avoiding the poor adaptability of the scheme caused by the lack of pathophysiological relevance in the background; then, the motor function index is determined to obtain a score vector quantifying the musculoskeletal function status of the target patient in the current rehabilitation training; the determination of the motor function index replaces the subjective experience or single data evaluation method in the prior art, which can not only intuitively reflect the overall function level of the patient's musculoskeletal system, but also accurately locate the functional short board, such as low neuromuscular coordination ratio and muscle insufficiency, to provide a targeted basis for subsequent rehabilitation intervention, avoiding the poor adaptability of the scheme caused by ambiguous evaluation; finally, the function recovery trajectory is determined to obtain a continuous curve representing the coordinated change of the target patient's daily muscle strength and joint range of motion under different training load scenarios; the determination of the function recovery trajectory can intuitively quantify the recovery efficiency of different training loads, and combined with the load safety margin, it can provide objective evidence for screening and adapting the training load scenario from the efficiency-safety dimension, avoiding the defect that the load does not match the patient's fracture healing stage and current function level, solving the problem of lack of individualized prediction support for the general load gradient in the background, and compared with the static mode in the prior art that once the scheme is determined, it is continuously executed without considering the recovery trend changes, the function recovery trajectory can predict potential problems in advance, so that the subsequent personalized rehabilitation instruction set can set a phased target based on the trajectory trend, ensuring that the scheme dynamically adapts to the patient's recovery, avoiding rehabilitation stagnation or risks caused by static intervention, and breaking through the limitation of static intervention in the prior art that cannot dynamically adapt to recovery changes; in summary, based on the above scheme, the recovery trajectory under different training loads can be predicted based on the multi-modal motion features of the patient to dynamically generate a personalized instruction set to assist the patient in rehabilitation training. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 is an exemplary flowchart of an orthopedic rehabilitation training method based on individual function dynamic evaluation and prediction according to some embodiments of the application; Figure 2 is an operation flowchart for determining the activation timing pattern according to some embodiments of the application; Figure 3 is an exemplary flowchart of determining the motor function index according to some embodiments of the application; Figure 4is a structural schematic diagram of an orthopedic rehabilitation training system according to some embodiments of the present application; Figure 5 is an internal structure diagram of a computer device implementing an orthopedic rehabilitation training method based on individual function dynamic assessment and prediction according to some embodiments of the present application. DETAILED DESCRIPTION

[0018] In order to better understand the technical solutions of the present application, the technical solutions of the present application will be described in detail below in combination with the drawings in the specification and specific embodiments.

[0019] Reference Figure 1 The figure is an exemplary flowchart of an orthopedic rehabilitation training method based on individual function dynamic assessment and prediction according to some embodiments of the present application, which mainly includes the following steps: In step 101, the original electromyographic signal and three-dimensional motion data of the target patient performing orthopedic standard rehabilitation actions are collected.

[0020] It should be noted that in the present application, the original electromyographic signal is an electrophysiological signal reflecting the neuromuscular activation state of the target patient performing orthopedic standard rehabilitation actions, which can provide basic data for extracting motor unit discharge sequences and constructing neuromuscular activation timing patterns, thereby accurately assessing the control function of nerves to muscles; the three-dimensional motion data is time series data reflecting the spatial position and motion posture changes of the target patient performing orthopedic standard rehabilitation actions, which can be used to extract joint activity features and electromyographic signals to form multi-modal motion features, thereby realizing comprehensive assessment of joint activity and motion coordination.

[0021] When specifically implemented, the original electromyographic signal and three-dimensional motion data of the target patient performing orthopedic standard rehabilitation actions can be implemented in the following manner, that is: first, a surface electromyographic sensor closely attached to the skin surface of the main force muscle group (such as biceps brachii, triceps brachii, etc.) of the target patient's affected limb can be used to capture electromyographic signals at a sampling frequency of 200Hz, and the collection of all captured electromyographic signals is used as the original electromyographic signal; then, a wearable inertial measurement unit containing an accelerometer, a gyroscope and a magnetometer fixed on the key skeletal nodes (such as the lateral epicondyle of the humerus, the radial styloid process, etc.) of the target patient's affected limb can be used to collect motion data at a sampling frequency of 200Hz, and the collection of all collected motion data is used as three-dimensional motion data.

[0022] In step 102, the discharge sequence of the motor unit is extracted from the original electromyography signal for time-frequency domain fusion, so as to obtain the activation timing mode of the neuromuscular system of the target patient when performing the orthopedic rehabilitation action, and then the multi-modal motion feature of the target patient when performing the orthopedic rehabilitation training is determined based on the activation timing mode and the joint activity feature of the target patient in the three-dimensional kinematics data.

[0023] In some embodiments, the reference Figure 2 The figure is an operation flow chart for determining the activation timing mode according to some embodiments of the present application. The extraction of the discharge sequence of the motor unit from the original electromyography signal for time-frequency domain fusion to obtain the activation timing mode of the neuromuscular system of the target patient when performing the orthopedic rehabilitation action can be implemented by the following steps: The discharge sequence of each motor unit is separated from the original electromyography signal; The continuous wavelet transform is performed on the discharge sequence of all motor units, and then the energy distribution feature of each discharge sequence in the time-frequency domain is extracted; The energy distribution features are cooperatively fused to obtain the activation timing mode of the neuromuscular system of the target patient when performing the orthopedic rehabilitation action.

[0024] It should be noted that, in the present application, the motor unit is a basic unit of muscle contraction function composed of a single motor neuron of the target patient and all muscle fibers dominated by the motor neuron. The motor unit, as the smallest functional unit of neuromuscular activation, can provide a basic anatomical functional unit reference for analyzing the neural control mechanism of muscle contraction.

[0025] When specifically implemented, the separation of the discharge sequence of each motor unit from the original electromyography signal can be implemented by the following method, that is, the original electromyography signal can be band-pass filtered and denoised, and the original electromyography signal after denoising can be processed by using an improved blind source separation algorithm (such as a motor unit decomposition algorithm based on convolution sparse coding). That is, the original electromyography signal is decomposed into a plurality of sparse components by using a constructed complete dictionary, each sparse component corresponds to an action potential waveform of a single motor unit, for each sparse component, an adaptive threshold method can be used to identify the firing time of the action potential of the sparse component, the threshold can be set to 3 times the standard deviation of the sparse component, and all identified time points are sorted by time to form a discharge time sequence of the motor unit corresponding to the sparse component. Through the above steps, the discharge sequence of each motor unit can be obtained. The discharge sequence is a sequence of time points at which the motor unit fires an action potential during the action execution process. The discharge sequence can provide original timing data for analyzing the activation rhythm and cooperative mode of the neuromuscular system by recording the timing information of the neural impulse triggering muscle contraction.

[0026] In a specific implementation, the discharge sequence of all movement units is subjected to continuous wavelet transform, and then the energy distribution features of each discharge sequence in the time-frequency domain can be extracted in the following manner: for the discharge sequence of each movement unit, the Morlet wavelet can be selected as the base function, and the scale range is set to 1 to 32. The discharge sequence is subjected to continuous wavelet transform to obtain a time-frequency energy matrix of the discharge sequence. Then, the energy peak value, dominant frequency and frequency band energy proportion in each time window of the time-frequency energy matrix are calculated. The sequence composed of all the energy peak values, dominant frequencies and frequency band energy proportions in time sequence is taken as the energy distribution features of the discharge sequence in the time-frequency domain. Through the above steps, the energy distribution features of each discharge sequence in the time-frequency domain can be obtained. The energy distribution features are a set of quantitative parameters representing the distribution law of the energy of the discharge sequence on the time axis and the frequency axis. The energy distribution features can enhance the description ability of the fine degree of neural control by quantifying the time-frequency domain characteristics of neuromuscular activation. The length of the time window can be set to 100 milliseconds, and the step length is 50 milliseconds. The dominant frequency is the frequency with the highest energy proportion. The frequency band energy proportion is the proportion of the energy of the 20-50 Hz, 50-100 Hz and 100-200 Hz frequency bands in the total energy, respectively.

[0027] In a specific implementation, the energy distribution features of all movement units are cooperatively fused to obtain the activation timing pattern of the neuromuscular system of the target patient during the execution of the orthopedic rehabilitation action in the following manner: first, the Pearson correlation coefficients between the energy distribution features of different movement units are calculated, and a cooperative activation matrix is constructed based on all the Pearson correlation coefficients. The matrix elements of the cooperative activation matrix are the correlation coefficients between the energy distribution features of two movement units. Then, the attention mechanism is used to dynamically assign weights based on the cooperative activation matrix, i.e., the higher the Pearson correlation coefficient, the higher the weight given to the energy distribution feature. Finally, all the energy distribution features are fused into a comprehensive feature matrix by weighted summation, and the feature matrix obtained after time axis alignment of the comprehensive feature matrix based on the starting time of the action is taken as the activation timing pattern of the neuromuscular system of the target patient during the execution of the orthopedic rehabilitation action.

[0028] It should be noted that in this application, the activation timing pattern is a comprehensive matrix reflecting the dynamic activation law of the neuromuscular system of the target patient during the execution of the rehabilitation training action. It contains the activation intensity and time-frequency feature evolution sequence of each movement unit at different time points. The activation timing pattern can be used as a core representation of the functional state of the neuromuscular system, and provides key neural control features for subsequent multi-modal motion feature construction and functional evaluation.

[0029] In some embodiments, the multi-modal motion features of the target patient during the orthopedic rehabilitation training can be determined based on the activation timing pattern and the joint activity features of the target patient in the three-dimensional kinematic data in the following steps: extract a joint activity feature of the target patient during orthopedic rehabilitation training from the three-dimensional kinematics data; determine a spatiotemporal coupling relationship between neuromuscular activation and joint movement of the target patient during orthopedic rehabilitation training; perform feature cascade fusion on the activation timing pattern and the joint activity feature based on the spatiotemporal coupling relationship, to obtain a multi-modal movement feature of the target patient during orthopedic rehabilitation training.

[0030] It should be noted that in the present application, the joint activity feature is a quantitative parameter vector representing the movement state of the key joints of the affected limb of the target patient, which can provide basic data at the kinematics level for evaluating the movement function of the skeletal muscle system by representing the movement ability and coordination of the joints.

[0031] In a specific implementation, the joint activity feature of the target patient during orthopedic rehabilitation training can be extracted from the three-dimensional kinematics data in the following manner: first, the Kalman filtering algorithm can be used to remove noise in the three-dimensional kinematics data, the state equation can be set as a uniform motion model, and the observation equation can be set as a combination of acceleration and angular velocity error compensation; then, the denoised three-dimensional kinematics data is input into an existing skeletal kinematics model (such as an improved extended Kalman filter fusion skeletal kinematics model) to calculate the angle (such as flexion and rotation angle) timing, angular velocity timing and angular acceleration timing of the key joints (such as elbow joint, wrist joint) of the affected limb; finally, based on the angle timing, angular velocity timing and angular acceleration timing, the range of motion peak value, movement smoothness and joint movement symmetry of each key joint are calculated, and a multi-dimensional vector composed of the range of motion peak value, movement smoothness and joint movement symmetry of all joints is taken as the joint activity feature of the target patient during orthopedic rehabilitation training; wherein, the range of motion peak value is the maximum activity angle; the movement smoothness is the integral value of the velocity curve divided by time; and the joint movement symmetry is the deviation rate of the movement parameters of the same named joints on the left and right sides.

[0032] In a specific implementation, the spatiotemporal coupling relationship between neuromuscular activation and joint movement of the target patient during orthopedic rehabilitation training can be determined in the following manner: first, a sliding window method can be used to extract data segments from the joint activity characteristics, and the time offset between the activation time of the motor unit and the change time of the joint angle in each time window and the correlation strength between the feature values in the activation timing pattern feature matrix and the peak values of the activity of each key joint can be calculated through cross-correlation analysis according to the activation timing pattern feature matrix and the extracted data segments in the joint activity characteristics, and the average of the normalized time offset and correlation strength can be used as the spatiotemporal coupling degree; then, the spatiotemporal coupling degrees in all time windows are integrated into a matrix, with rows representing motor units and columns representing key joints, and the matrix elements being spatiotemporal coupling degrees, and the obtained matrix is used as the spatiotemporal coupling relationship between neuromuscular activation and joint movement; wherein the spatiotemporal coupling relationship is a data structure reflecting the coupling relationship between the timing of neuromuscular activation and the parameters of joint movement in terms of time synchronization and spatial correlation, which can provide scientific basis for multi-dimensional feature fusion by revealing the internal relationship between neural control and limb movement, and improve feature correlation; the length of the time window can be set to 100 milliseconds, and the step size is 50 milliseconds; the time offset is the time difference corresponding to the maximum correlation coefficient; the correlation strength is the Pearson correlation coefficient.

[0033] In a specific implementation, the feature-level fusion of the activation timing pattern and the joint activity characteristics based on the spatiotemporal coupling relationship can be realized in the following manner: first, the spatiotemporal coupling degrees in the spatiotemporal coupling relationship can be used to assign weights to the activation timing pattern and the joint activity characteristics, i.e., the higher the coupling strength, the higher the weight assigned to the feature; then, the activation timing pattern and the joint activity characteristics are weighted and spliced according to the assigned weights to form a high-dimensional feature set, and principal component analysis algorithm is used to reduce the dimension of the high-dimensional feature set to remove redundant information, thereby obtaining the multi-modal movement feature of the target patient during orthopedic rehabilitation training.

[0034] It should be noted that in this application, the multi-modal movement feature is a feature vector reflecting the neuromuscular activation pattern and joint movement characteristics of the target patient during rehabilitation training, which can provide comprehensive and accurate feature support for individualized function prediction and rehabilitation program development by integrating information from the neural control and movement execution levels, and improve the relevance of evaluation and intervention.

[0035] In step 103, individualized function prediction is performed based on the multi-modal movement feature to obtain the motor function index of the skeletal muscle system of the target patient during the current rehabilitation period, and to determine the functional recovery trajectory of the skeletal muscle system of the target patient under different training load scenarios in the next rehabilitation period.

[0036] In some embodiments, reference is made to Figure 3 FIG. 12 is an exemplary flowchart for determining the motor function index according to some embodiments of the present application, in which individualized function prediction is made based on the multi-modal movement features, and the motor function index of the skeletal muscle system of the target patient during the current rehabilitation period when the target patient undergoes orthopedic rehabilitation training is obtained according to the following steps: In step 1031, the historical clinical data of the target patient is obtained. In step 1032, the clinical baseline features of the target patient are extracted from the historical clinical data. In step 1033, a muscle-bone system function prediction network for the target patient when the target patient undergoes orthopedic rehabilitation training is constructed based on the clinical baseline features and the multi-modal movement features. In step 1034, the motor function index of the skeletal muscle system of the target patient during the current rehabilitation period when the target patient undergoes orthopedic rehabilitation training is determined according to the muscle-bone system function prediction network.

[0037] It should be noted that in the present application, the historical clinical data is a set of clinical information showing the basic information, orthopedic condition, disease course and previous rehabilitation records of the target patient, which can provide complete data support for extracting clinical baseline features and constructing a muscle-bone system function prediction network, and ensure the individualized data basis for function prediction.

[0038] In specific implementation, the historical clinical data of the target patient can be obtained in the following manner, i.e., the clinical data of the target patient can be called through the application programming interface of the hospital rehabilitation department management platform, including: basic information such as age, gender, body mass index; orthopedic condition data such as fracture site, fracture type, injury severity, surgical method and internal fixation type; disease course data such as the interval time from injury to rehabilitation training, current rehabilitation stage; previous rehabilitation data such as previous training program, functional assessment score after each training, adverse reaction record, and the collected clinical data is standardized and cleaned, i.e., the missing values of continuous data are filled with the median of the same type of patients, the missing values of categorical data are marked as unknown, and the outliers in the data are identified and removed by the quartile range method, and finally the structured historical clinical data is output.

[0039] In a specific implementation, the extraction of the clinical baseline features of the target patient from the historical clinical data can be achieved in the following manner. First, the clinical baseline features can be extracted from the historical clinical data by dimension according to data types. For categorical data, the one-hot encoding technique can be used to convert the data into a binary vector, and the binary vector can be used as a categorical feature. For example, a distal radius type A fracture can be encoded as [1, 0, 0], and a distal radius type B fracture can be encoded as [0, 1, 0]. For continuous data, the maximum-minimum normalization technique can be used to map the continuous data to the 0-1 interval, eliminate the dimension effect, and use the mapped data as a continuous feature. Then, all the categorical features and continuous features are integrated into a multi-dimensional vector, and the multi-dimensional vector is used as the clinical baseline features of the target patient. The clinical baseline features are the core feature vector reflecting the individual disease differences of the target patient and the historical rehabilitation basis.

[0040] In a specific implementation, the construction of the musculoskeletal system function prediction network of the target patient during orthopedic rehabilitation training based on the clinical baseline features and the multi-modal motion features can be achieved in the following manner. An existing prediction network model (such as a prediction network based on a random forest algorithm) can be loaded, and a clinical baseline feature-multi-modal motion feature double-input integrated learning framework can be used to construct the musculoskeletal system function prediction network of the target patient during orthopedic rehabilitation training based on the prediction network model as a base model. The clinical baseline features and the multi-modal motion features can be spliced by dimension to form a fusion feature input in the input layer. An attention module can be set in the hidden layer to dynamically allocate feature weights by calculating the Pearson correlation coefficient of each feature and the musculoskeletal function score. A 5-fold cross-validation method can be used to optimize the model hyperparameters, and the average absolute error is less than 3 points and the determination coefficient is greater than 0.85 as the model acceptance standard. The output layer can be set to output the current state score of the musculoskeletal system of the target patient and the constituent component score proportion of the state score, such as muscle strength level accounting for 30%, joint range of motion accounting for 30%, and neural control coordination accounting for 40%. Then, a feature sample set including the clinical baseline features, the multi-modal motion features, and the corresponding musculoskeletal function score labels can be input to train the prediction network model, and a musculoskeletal system function prediction network that is trained and meets the performance standards can be obtained. The musculoskeletal system function prediction network is a double-input integrated learning model that integrates the clinical relevant information of the target patient and multi-dimensional motion features to output a quantitative musculoskeletal system current function state score. The musculoskeletal system function prediction network can provide a core quantitative basis for subsequent determination of function recovery trajectories under different training loads, screening of adaptive training load scenarios, and generation of personalized rehabilitation instruction sets, ensuring that the rehabilitation intervention can accurately match the current function state of the patient and improving the scientificity and effectiveness of the rehabilitation scheme. The musculoskeletal function score is the score of the patient's muscle strength and joint range of motion made by the doctor, and the value range is 0-100.

[0041] It should be noted that in the present application, the motor function index is a score vector quantifying the musculoskeletal function state of the target patient in the current rehabilitation training, including a state score with a value range of 0-100 and component score proportions of muscle strength level, joint range of motion and neural coordination of the target patient, the higher the state score, the better the recovery of the musculoskeletal system function, the motor function index can provide intuitive quantitative basis for generating personalized rehabilitation instruction set, and the functional short board of the target patient is clear to optimize the rehabilitation intervention direction; in specific implementation, the motor function index of the skeletal muscle system of the target patient in the current rehabilitation cycle during orthopedic rehabilitation training can be realized in the following manner, i.e., the clinical baseline features and multi-modal motion features of the target patient can be input into the trained musculoskeletal system function prediction network, the musculoskeletal system function prediction network predicts the feature values at the current time from the clinical baseline features and multi-modal motion features, outputs the current state score of the skeletal muscle system of the target patient, and outputs the component score proportions of the state score, and the vector composed of the state score and the component score proportions is taken as the motor function index of the skeletal muscle system of the target patient in the current rehabilitation cycle during orthopedic rehabilitation training.

[0042] In some embodiments, determining the functional recovery trajectory of the skeletal muscle system of the target patient under different training load scenarios in the next rehabilitation cycle can be realized in the following steps: determining different training load scenarios of the target patient in the next rehabilitation cycle; predicting the recovery curves of the skeletal muscle system of the target patient under different training load scenarios based on the musculoskeletal system function prediction network; curve smoothing is performed on all the recovery curves to obtain the functional recovery trajectory of the skeletal muscle system of the target patient under different training load scenarios in the next rehabilitation cycle.

[0043] It should be noted that in the present application, the training load scenario is a diversified training scheme containing different load intensities and supporting training parameters in orthopedic rehabilitation training, which can provide a scenario basis for simulating the recovery under different intervention conditions and help to screen the optimal rehabilitation load subsequently.

[0044] In a specific implementation, the determination of different training load scenarios of the target patient in the next rehabilitation cycle can be achieved by the following method: the current maximum voluntary contraction test value of the target patient is extracted from the historical clinical data, and the load progression principle in clinical rehabilitation medicine is referred to, and the maximum voluntary contraction test value is taken as the benchmark to divide 5 training load scenarios and determine the training parameters of each training load scenario, i.e., 20%-30% of the maximum voluntary contraction test value is low load, 30%-40% of the maximum voluntary contraction test value is low-medium load, 40%-50% of the maximum voluntary contraction test value is medium load, 50%-60% of the maximum voluntary contraction test value is high-medium load, and 60%-70% of the maximum voluntary contraction test value is high load. Each load corresponds to 3-5 standard rehabilitation actions, for example, wrist flexion and extension, pronation and supination for distal radius fracture, 8-15 repetitions per group, 60-120 seconds rest between groups, 1-2 groups per day, thereby obtaining different training load scenarios of the target patient in the next rehabilitation cycle.

[0045] In a specific implementation, the prediction of the recovery curve of the skeletal muscle system of the target patient under different training load scenarios based on the musculoskeletal system function prediction network can be achieved by the following method: based on the musculoskeletal system function prediction network, a load-time prediction module (such as a linear mixed effect model) is added, the target patient's each training load scenario in the next rehabilitation cycle is input, the recovery process is simulated through the load-time prediction module, and the recovery curve of the skeletal muscle system of the target patient under different training load scenarios is output, i.e., time and load intensity are taken as fixed effect variables, and individual differences (such as muscle fiber type ratio) of the target patient are taken as random effect variables. The muscle strength prediction value and joint range of motion prediction value of each training load scenario per day are output through the formula "predicted function value = intercept + time coefficient x days + load coefficient x intensity + individual random effect term" to form two discrete function value time series, i.e., muscle strength recovery curve and joint range of motion recovery curve. The set of muscle strength recovery curve and joint range of motion recovery curve is taken as the recovery curve of the skeletal muscle system of the target patient under each training load scenario. The recovery curve is a discrete prediction sequence reflecting the daily muscle strength and joint range of motion changes of the target patient under different training load scenarios. The recovery curve can intuitively quantify the single-dimensional functional recovery trend of the target patient, and provide raw data support for subsequent generation of continuous functional recovery trajectory. Time is the number of rehabilitation training days. Load intensity is the maximum voluntary contraction test value.

[0046] In a specific implementation, the functional recovery trajectory of the skeletal muscle system of the target patient under different training load scenarios in the next rehabilitation period can be obtained by performing curve smoothing on all the recovery curves, that is, the discrete recovery curves in each recovery curve can be processed by using the local weighted scatter smoothing method in the prior art to obtain the functional recovery trajectory of the skeletal muscle system of the target patient under different training load scenarios in the next rehabilitation period, that is, the size of the smoothing window can be set to 7 days, covering one rehabilitation period, and a continuous curve can be fitted by using the weighted average of the functional values of the recovery curves in the smoothing window, so as to reduce the random fluctuations of the daily predicted values, and at the same time, abnormal functional values in the recovery curve are identified by using the 3σ criterion, and the mean value of the adjacent three normal functional values is used to replace the abnormal functional values, so as to avoid distortion of the trajectory, and the functional recovery trajectory of the skeletal muscle system including the muscle strength recovery curve and the joint range of motion recovery curve under each training load scenario is output; wherein, the abnormal functional value is a functional value deviating from the curve trend by 3 times the standard deviation.

[0047] It should be noted that in this application, the functional recovery trajectory is a continuous curve representing the coordinated change of muscle strength and joint range of motion of the target patient under different training load scenarios, which can clearly show the overall recovery law under different loads, and provide a key decision basis for selecting the training load scenario with the highest weekly improvement rate and generating a personalized rehabilitation instruction set.

[0048] In step 104, a personalized rehabilitation instruction set matching the current rehabilitation state of the target patient is generated according to the motor function index and all the functional recovery trajectories, and then the assistive torque output of the rehabilitation training equipment is adjusted when the target patient performs orthopedic rehabilitation training based on the personalized rehabilitation instruction set.

[0049] In some embodiments, the personalized rehabilitation instruction set matching the current rehabilitation state of the target patient can be generated according to the motor function index and all the functional recovery trajectories by using the following steps: Quantifying the effectiveness of all the functional recovery trajectories to generate an effectiveness ranking table of the training load scenarios; Constructing a load adaptation matrix of the motor function of the target patient; Selecting an adaptive training load scenario from the effectiveness ranking table of the training load scenarios by using the motor function index and the load adaptation matrix; Generating a personalized rehabilitation instruction set matching the current rehabilitation state of the target patient according to the adaptive training load scenario.

[0050] In a specific implementation, the performance quantification of all functional recovery trajectories can be achieved by generating a performance ranking table of the training load scenarios in the following manner: first, for each functional recovery trajectory, the weekly average functional improvement rate and the recovery fluctuation entropy of the muscle strength recovery curve and the joint range of motion recovery curve in the functional recovery trajectory are calculated respectively, and the load safety margin of the training load scenario corresponding to the functional recovery trajectory, i.e., the percentage difference between the actual training load and the maximum tolerable load of the patient in the fracture healing stage, is calculated; then, the weekly average functional improvement rate, the recovery fluctuation entropy, and the load safety margin are weighted and summed according to the preset weights to obtain the comprehensive performance value of the training load scenario corresponding to the functional recovery trajectory, for example, a weekly average functional improvement rate of 45%, a recovery fluctuation entropy of 30%, and a load safety margin of 25%. Through the above steps, the comprehensive performance value of each functional recovery trajectory corresponding to the training load scenario can be obtained; finally, the performance ranking table of the training load scenarios is obtained by ranking all the training load scenarios in descending order of the comprehensive performance value; wherein, the performance ranking table is a training load scenario ranking table formed by ranking all the training load scenarios in descending order of the comprehensive performance value. The performance ranking table can intuitively present the pros and cons of each training load scenario in training performance, providing a quantitative basis for subsequent selection of the optimal load and avoiding subjective selection bias; the weekly average functional improvement rate is the average growth rate of the functional value per week; the recovery fluctuation entropy is the degree of dispersion of the functional value per day.

[0051] In a specific implementation, the load adaptation matrix of the target patient's motor function can be constructed in the following manner: a 3x3-dimensional load adaptation matrix can be constructed with the motor function index as the core combined with the clinical orthopedic rehabilitation guidelines. The row dimension of the load adaptation matrix is the motor function level, including: a low function level with a motor function index less than 40 points, which can correspond to the early stage of fracture healing with weak muscle strength; a medium function level with a motor function index between 40 and 70 points, which can correspond to the medium stage of fracture healing with medium muscle strength; and a high function level with a motor function index greater than 70 points, which can correspond to the late stage of fracture healing with strong muscle strength. The column dimension is the type of training load scenario, including: low-medium low load, medium-medium high load, and medium-high high load. Each function level is matched with an adaptive load type, for example: the low function level is only adapted to low-medium low load to avoid injury from excessive load, the medium function level is adapted to medium-medium high load to balance recovery efficiency and safety, and the high function level is adapted to medium-high high load to promote functional breakthrough. At the same time, the safety constraints of each adaptive combination are marked in the load adaptation matrix, for example: the upper limit of the low function level load is not more than 40% of the maximum voluntary contraction force. Thus, the load adaptation matrix of the target patient's motor function is obtained; wherein, the load adaptation matrix is an adaptive rule matrix based on the division of function levels according to the motor function index and the matching of corresponding load types and safety constraints. The load adaptation matrix can establish scientific adaptation rules for function and load, avoid injury from excessive load or low efficiency from insufficient load, and ensure the safety and adaptability of load selection.

[0052] In specific implementation, the selection of suitable training load scenarios from the training load scenario efficacy ranking table using the motor function index and the load adaptation matrix can be achieved in the following way: First, the functional level of the target patient can be determined based on the motor function index, and the corresponding suitable load type can be found based on the load adaptation matrix; then, in the training load scenario efficacy ranking table, all training load scenarios belonging to the suitable load type are extracted as candidate subsets, and the candidate subset is subjected to safety verification to remove training load scenarios with a load safety margin of less than 10% to avoid the load exceeding the patient's tolerance range; then, the training load scenario with the highest comprehensive efficacy value is selected from the remaining training load scenarios as the suitable training load scenario; wherein, the suitable training load scenario is a training load scenario that combines efficient recovery and safety, selected based on the target patient's current motor function state and pathological state. This suitable training load scenario can determine the most suitable training conditions for the target patient's current state, providing core parameters for generating personalized instruction sets.

[0053] It should be noted that in this application, the personalized rehabilitation instruction set is an orthopedic rehabilitation training instruction set that matches the current motor function status and pathological status of the target patient and has both high efficiency and safety. This personalized rehabilitation instruction set can directly guide the target patient to carry out orthopedic rehabilitation training, making rehabilitation intervention more precise and executable, and ensuring the standardization and effectiveness of training.

[0054] In specific implementation, the generation of a personalized rehabilitation instruction set matching the current rehabilitation status of the target patient based on the adapted training load scenario can be achieved in the following way: The training parameters in the adapted training load scenario can be integrated with existing clinical rehabilitation guidelines to generate a personalized rehabilitation instruction set matching the current rehabilitation status of the target patient. Specifically: First, determine the daily training action list and training frequency based on the training parameters in the adapted training load scenario; second, set and add safety and monitoring clauses, namely: set the force safety threshold to 70% of the target patient's maximum voluntary contractile force, and combine it with the real-time force value monitoring function of the rehabilitation equipment. If the limit is exceeded, an audible and visual warning will be triggered. Daily functional monitoring indicators are also specified, such as wrist flexion and extension range of motion must be greater than or equal to 45° and grip strength must be greater than or equal to 15 Newtons. Finally, a structured rehabilitation instruction set containing action names, execution standards, training frequency, safety thresholds, and monitoring requirements is formed. Among these, clinical rehabilitation guidelines refer to clinical operational guidelines in the field of orthopedic rehabilitation that are formulated based on the patient's pathological state and functional level, and include training safety thresholds, action standards, progressive load principles, and functional monitoring indicators.

[0055] In a specific implementation, the adjustment of the assistive torque output of the rehabilitation training device when the target patient performs orthopedic rehabilitation training based on the personalized rehabilitation instruction set can be implemented in the following manner, that is, the rehabilitation training device takes the force exertion safety threshold, the standard trajectory of the target training action, and the corresponding target load intensity in the personalized rehabilitation instruction set as the basis for torque adjustment, calculates the motion trajectory deviation, that is, the spatial distance between the actual trajectory of the target patient and the standard trajectory and the force exertion deviation value between the actual force exertion value of the target patient and the target load intensity, and then uses the proportional-integral-derivative control algorithm in the prior art to dynamically adjust the assistive torque output according to the motion trajectory deviation and the force exertion deviation value, that is, if the force exertion deviation value is negative and the absolute value is greater than 15%, that is, the force exertion is insufficient, the proportional-integral-derivative control algorithm outputs a positive torque increment in proportion, and the increment is not more than 30% of the target load intensity, helping the patient to reach the target load; if the force exertion deviation value is positive and exceeds the force exertion safety threshold, that is, the force exertion is out of limit, a negative torque decrement is output, and the decrement is not more than 20% of the target load intensity, avoiding excessive force exertion; if the motion trajectory deviation is greater than 5°, that is, the action deviates from the standard, the device first prompts to correct the action through voice, and then continues to adjust the torque after the trajectory deviation is less than or equal to 5°, and the data and torque are updated every 50-100 milliseconds to ensure that the assistive torque always matches the current force exertion capacity of the patient and the requirements of the instruction set, both ensuring the training effect and avoiding injury; wherein the rehabilitation training device comprises a torque adjustment component and a monitoring feedback component, the torque adjustment component is used to output the assistive torque according to the personalized rehabilitation instruction set to drive the affected limb to complete the standard rehabilitation action, and realize the accurate landing of instruction execution, and the monitoring feedback component is used to monitor the safety and effect indicators in the training process in real time, correct the deviation in time and feedback the state.

[0056] In addition, another aspect of the present application, in some embodiments, the present application provides an orthopedic rehabilitation training system, referring to Figure 4 The figure is a structural schematic diagram of an orthopedic rehabilitation training system according to some embodiments of the present application, which comprises a collection module 401, a processing module 402 and an execution module 403, which are described as follows: The collection module 401 is mainly used for collecting the original electromyographic signal and the three-dimensional motion data of the limb when the target patient performs the orthopedic standard rehabilitation action in the present application; The processing module 402 is mainly used for extracting the discharge sequence of the motion unit in the original electromyographic signal for time-frequency domain fusion to obtain the activation timing pattern of the neuromuscular system when the target patient performs the orthopedic rehabilitation action, and then determine the multi-modal motion characteristics of the target patient when performing orthopedic rehabilitation training based on the activation timing pattern and the joint activity characteristics of the target patient in the three-dimensional kinematic data; It should be noted that the processing module 402 is also used to perform individual function prediction based on the multi-modal motion features, obtain a motor function index of a skeletal muscle system of the target patient in a current rehabilitation cycle when the target patient performs orthopedic rehabilitation training, and determine a function recovery trajectory of the skeletal muscle system of the target patient in a next rehabilitation cycle under different training load scenarios. The execution module 403 is mainly used to generate a personalized rehabilitation instruction set matched with a current rehabilitation state of the target patient according to the motor function index and all function recovery trajectories, and then adjust an auxiliary torque output of a rehabilitation training device when the target patient performs orthopedic rehabilitation training based on the personalized rehabilitation instruction set.

[0057] The above modules in the orthopedic rehabilitation training system can be all or partially implemented by software, hardware, and combinations thereof. The above modules can be embedded in or independent of a processor in a computer device in a hardware form, or can be stored in a memory in the computer device in a software form, so as to be called and executed by the processor to perform operations corresponding to the above modules.

[0058] In addition, in an embodiment, the present application provides a computer device, which can be a server, and an internal structure diagram thereof can be as shown in Figure 5 The computer device includes a processor, a memory, and a network interface connected through a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The database of the computer device is used to store orthopedic rehabilitation training method data based on individual function dynamic assessment and prediction. The network interface of the computer device is used to communicate with an external terminal through a network connection. The computer program, when executed by the processor, can implement an orthopedic rehabilitation training method based on individual function dynamic assessment and prediction.

[0059] Those skilled in the art can understand that Figure 5 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. A specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0060] In an embodiment, a computer device is also provided, including a memory and a processor, the memory storing a computer program, and the processor implementing the steps in the above embodiments of the orthopedic rehabilitation training method based on individual function dynamic assessment and prediction when executing the computer program.

[0061] In one embodiment, a computer readable storage medium storing a computer program is provided, the computer program, when executed by a processor, implements the steps in the above-described orthopedic rehabilitation training method embodiments based on individual function dynamic assessment and prediction.

[0062] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program comprising computer instructions stored in a computer readable storage medium. A processor of a computer device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions to cause the computer device to perform the steps in the above-described orthopedic rehabilitation training method embodiments based on individual function dynamic assessment and prediction.

[0063] A person of ordinary skill in the art can understand that all or part of the processes in the above-described embodiments can be completed by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, it can include the processes of the above-described embodiments. Any reference to memory, storage, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM can be in various forms such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0064] Any combination of the technical features of the above embodiments can be made. In order to make the description concise, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combination of the technical features does not exist, it should be considered as the scope of the present application.

[0065] The above-described embodiments only express several implementation manners of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are all within the scope of the present application. Therefore, the scope of the patent of the present application should be subject to the appended claims.

Claims

1. An orthopedic rehabilitation training method based on individual functional dynamic assessment and prediction, characterized in that, Includes the following steps: Collect raw electromyographic signals and three-dimensional motion data of the limbs of the target patients when they perform standard orthopedic rehabilitation movements; The discharge sequence of motor units is extracted from the original electromyography signal and fused in the time and frequency domains to obtain the neuromuscular activation time sequence pattern when the target patient performs orthopedic rehabilitation movements. Then, based on the activation time sequence pattern and the joint activity characteristics of the target patient in the three-dimensional kinematic data, the multimodal motion characteristics of the target patient during orthopedic rehabilitation training are determined. Based on the multimodal motion characteristics, individualized functional prediction is performed to obtain the motor function index of the musculoskeletal system in the current rehabilitation cycle when the target patient is undergoing orthopedic rehabilitation training, and to determine the functional recovery trajectory of the musculoskeletal system of the target patient under different training load scenarios in the next rehabilitation cycle. Based on the motor function index and all functional recovery trajectories, a personalized rehabilitation instruction set is generated to match the current rehabilitation status of the target patient. Then, based on the personalized rehabilitation instruction set, the auxiliary torque output of the rehabilitation training equipment is adjusted when the target patient undergoes orthopedic rehabilitation training.

2. The method as described in claim 1, characterized in that, The discharge sequence of motor units extracted from the original electromyography signal is fused in the time and frequency domains to obtain the neuromuscular activation timing pattern when the target patient performs orthopedic rehabilitation movements. Specifically, this includes: The discharge sequences of each motor unit are isolated from the raw electromyographic signals; Continuous wavelet transform is performed on the discharge sequences of all motion units to extract the energy distribution characteristics of each discharge sequence in the time-frequency domain; By synergistically fusing all energy distribution characteristics, the neuromuscular activation timing pattern during orthopedic rehabilitation movements performed by the target patient is obtained.

3. The method as described in claim 1, characterized in that, Based on the activation timing pattern and the joint activity characteristics of the target patient in the three-dimensional kinematic data, the multimodal motion characteristics of the target patient during orthopedic rehabilitation training are determined, specifically including: Extract joint movement characteristics of the target patient during orthopedic rehabilitation training from the three-dimensional kinematic data; To determine the spatiotemporal coupling relationship between neuromuscular activation and joint movement during orthopedic rehabilitation training in target patients; Based on the spatiotemporal coupling relationship, the activation timing pattern and the joint activity features are cascaded and fused to obtain the multimodal motion features of the target patient during orthopedic rehabilitation training.

4. The method as described in claim 1, characterized in that, Based on the aforementioned multimodal motion characteristics, individualized functional prediction is performed to obtain the musculoskeletal system motor function index of the target patient during the current rehabilitation cycle of orthopedic rehabilitation training. This index specifically includes: Obtain historical clinical data from the target patients; Extract the target patient's clinical baseline characteristics from the historical clinical data; Based on the clinical baseline features and the multimodal motion features, a musculoskeletal system function prediction network is constructed for target patients undergoing orthopedic rehabilitation training. The musculoskeletal system function prediction network is used to determine the motor function index of the musculoskeletal system in the current rehabilitation cycle when the target patient is undergoing orthopedic rehabilitation training.

5. The method as described in claim 4, characterized in that, Determining the functional recovery trajectory of the musculoskeletal system in the target patient under different training load scenarios during the next rehabilitation cycle specifically includes: Identify different training load scenarios for the target patient in the next rehabilitation cycle; Based on the musculoskeletal system function prediction network, the recovery curves of the target patient's musculoskeletal system under different training load scenarios are predicted. All recovery curves were smoothed to obtain the functional recovery trajectory of the musculoskeletal system of the target patient under different training load scenarios in the next rehabilitation cycle.

6. The method as described in claim 1, characterized in that, Based on the motor function index and all functional recovery trajectories, a personalized rehabilitation instruction set matching the target patient's current rehabilitation status is generated, specifically including: The effectiveness of all functional recovery trajectories is quantified to generate an effectiveness ranking table for training load scenarios; Construct a load adaptation matrix for the target patient's motor function; The appropriate training load scenarios are selected from the training load scenario effectiveness ranking table by using the motor function index and the load adaptation matrix. Based on the adaptive training load scenario, a personalized set of rehabilitation instructions is generated that matches the current rehabilitation status of the target patient.

7. The method as described in claim 1, characterized in that, The rehabilitation training equipment includes a torque adjustment component and a monitoring and feedback component.

8. An orthopedic rehabilitation training system, characterized in that, include: The acquisition module is used to acquire raw electromyographic signals and three-dimensional motion data of the limbs when the target patient performs standard orthopedic rehabilitation movements; The processing module is used to extract the discharge sequence of motor units from the original electromyography signal and perform time-frequency domain fusion to obtain the neuromuscular activation timing pattern when the target patient performs orthopedic rehabilitation movements. Then, based on the activation timing pattern and the joint activity characteristics of the target patient in the three-dimensional kinematic data, the multimodal motion characteristics of the target patient during orthopedic rehabilitation training are determined. The processing module is used to perform individualized functional prediction based on the multimodal motion features, obtain the motor function index of the musculoskeletal system in the current rehabilitation cycle when the target patient is undergoing orthopedic rehabilitation training, and determine the functional recovery trajectory of the musculoskeletal system of the target patient under different training load scenarios in the next rehabilitation cycle. The execution module is used to generate a personalized rehabilitation instruction set that matches the current rehabilitation status of the target patient based on the motor function index and all functional recovery trajectories, and then adjust the auxiliary torque output of the rehabilitation training equipment when the target patient is undergoing orthopedic rehabilitation training based on the personalized rehabilitation instruction set.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the orthopedic rehabilitation training method based on individual functional dynamic assessment and prediction as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the orthopedic rehabilitation training method based on individual functional dynamic assessment and prediction as described in any one of claims 1 to 7.

Images