CN-121983239-A - Intelligent orthopaedics joint rehabilitation monitoring method and system

CN121983239ACN 121983239 ACN121983239 ACN 121983239ACN-121983239-A

Abstract

The invention discloses a method and a system for intelligent orthopaedics joint rehabilitation monitoring, wherein the method comprises the steps of obtaining three-dimensional space-time characteristic vectors of joint motions according to joint motion data, muscle electric signals and pressure distribution information acquired by a multi-mode sensor; the method comprises the steps of extracting key parameters of joint function states based on three-dimensional space-time feature vectors by utilizing a space-time diagram convolutional neural network to obtain joint stability quantification indexes, obtaining rehabilitation stage quantification indexes according to the joint stability quantification indexes and pain feedback data of a patient, generating a safety training instruction set which accords with the current rehabilitation stage based on the rehabilitation stage quantification indexes, driving a wearable exoskeleton to execute rehabilitation training according to the safety training instruction set, and synchronously monitoring neuromuscular activation signals through a brain-computer interface to verify consistency of training actions and nerve control. By utilizing the embodiment of the invention, accurate, continuous and self-adaptive joint rehabilitation overall process monitoring and management can be realized.

Inventors

YANG PEI
ZHANG JIEWEN
LIU CHENGYAN
LIU ZEYU
CHEN YANG
TIAN RUN
WANG KUNZHENG
KONG NING
LI YIYANG
QIAO TIAN
LV JINGYI
ZHAO YIWEI
Cao Ruomu

Assignees

西安交通大学医学院第二附属医院

Dates

Publication Date: 20260505
Application Date: 20260123

Claims (10)

1. A method for intelligent orthopedic joint rehabilitation monitoring, the method comprising: According to joint motion data, muscle electric signals and pressure distribution information acquired by the multi-mode sensor, performing dynamic biomechanical modeling through a space-time feature fusion algorithm to obtain a three-dimensional space-time feature vector of joint motion; Based on the three-dimensional space-time feature vector, extracting key parameters of the joint function state by utilizing a space-time diagram convolutional neural network, and capturing a cooperative motion mode of joint ligaments and bones through a self-adaptive topological structure to obtain a joint stability quantization index; Calculating a dynamic evolution coefficient of a rehabilitation stage through a rehabilitation entropy model according to the joint stability quantization index and pain feedback data of a patient, and carrying out nonlinear weighting through fusion of biomechanical parameters and physiological response parameters to obtain a quantization index of the rehabilitation stage; Based on the quantitative index of the rehabilitation stage, constructing a personalized rehabilitation action sequence by adopting an countermeasure generation network, verifying the medical compliance of the action sequence by a discriminator, and generating a safety training instruction set conforming to the current rehabilitation stage; And driving the wearable exoskeleton to perform rehabilitation training according to the safety training instruction set, and synchronously monitoring neuromuscular activation signals through a brain-computer interface to verify the consistency of training actions and nerve control.
2. The method according to claim 1, wherein the dynamic biomechanical modeling by a spatio-temporal feature fusion algorithm based on the joint motion data, the muscle electrical signals, and the pressure distribution information acquired by the multi-modal sensor, to obtain a three-dimensional spatio-temporal feature vector of the joint motion, comprises: According to the joint motion data, performing six-degree-of-freedom trajectory Kalman filtering processing to obtain a noise-reduced joint motion trajectory; Based on electromyographic signals and pressure distribution information, a time domain phase alignment algorithm is adopted for processing, and a biomechanical coupling feature matrix is generated; and fusing the joint motion track and the biomechanical coupling feature matrix, and outputting a three-dimensional space-time feature vector through tensor cross decomposition processing.
3. The method according to claim 2, wherein the capturing the cooperative motion pattern of the joint ligament and the bone through the adaptive topology structure based on the three-dimensional space-time feature vector by using a space-time convolutional neural network to extract key parameters of the joint functional state, and obtaining the joint stability quantization index comprises: according to the three-dimensional space-time feature vector, carrying out dynamic skeleton topology construction processing to generate a self-adaptive biomechanical diagram structure; based on the self-adaptive biomechanical diagram structure, adopting layered space-time diagram convolution processing to extract ligament cooperative motion characteristics; Fusing ligament cooperative motion characteristics, and obtaining a key parameter set of joint stability through attention weighted pooling treatment; Based on the joint stability key parameter set, combining clinical evaluation standards to perform radial basis function mapping processing, and outputting joint stability quantization indexes.
4. The method according to claim 3, wherein calculating the rehabilitation phase dynamic evolution coefficient according to the joint stability quantization index and the patient pain feedback data through a rehabilitation entropy model, and obtaining the rehabilitation phase quantization index through non-linear weighting by fusing the biomechanical parameter and the physiological response parameter comprises: Performing time sequence differentiation treatment according to the joint stability quantization index to obtain the biomechanical parameter change rate; based on pain feedback data of a patient, adopting BioBERT characteristic extraction processing to generate a standardized pain characteristic vector; Fusing the biomechanical parameter change rate and the pain characteristic vector, and outputting a dynamic evolution coefficient through dual entropy coupling calculation processing; based on the dynamic evolution coefficient, performing parameter self-adaptive weighting processing to generate biomechanical-physiological response fusion weights; And according to the fusion weight combined with the clinical threshold, outputting a quantitative index of the rehabilitation stage through S-shaped function conversion processing.
5. The method of claim 4, wherein constructing a personalized rehabilitation motion sequence based on the rehabilitation phase quantization index using an countermeasure generation network, verifying medical compliance of the motion sequence by a arbiter, generating a safety training instruction set conforming to a current rehabilitation phase, comprises: according to the quantitative index of the rehabilitation stage, retrieving medical knowledge base processing to generate a safety boundary matrix of the joint activity degree; Based on the safety boundary matrix, adopting LSTM-transducer mixed network processing to generate an initial action sequence; according to the initial action sequence, obtaining joint angle space constraint actions through inverse kinematics mapping processing; Based on joint angle space constraint actions, adopting a three-dimensional biomechanical discriminator to verify and process, and screening out medical compliance actions; and integrating medical compliance actions and a diversity rewarding mechanism, and outputting a safety training instruction set through instruction coding processing.
6. The method of claim 5, wherein driving the wearable exoskeleton according to the safety training instruction set to perform rehabilitation training, and synchronizing monitoring neuromuscular activation signals through a brain-computer interface to verify consistency of training actions with neural control, comprises: according to the safety training instruction set, motor signal decoding processing is carried out, and the exoskeleton is driven to execute training actions; extracting neuromuscular activation characteristics by blind source separation processing based on brain-computer interface signals; fusing the actual motion trail and neuromuscular activation characteristics, and generating a nerve control matching degree through mutual information entropy calculation processing; and triggering reinforcement learning dynamic adjustment processing according to the nerve control matching degree, and outputting a nerve-mechanical synchronization verification report.
7. An intelligent orthopedic joint rehabilitation monitoring system, characterized in that the system comprises: The fusion module is used for carrying out dynamic biomechanical modeling through a space-time feature fusion algorithm according to joint motion data, muscle electrical signals and pressure distribution information acquired by the multi-mode sensor to obtain a three-dimensional space-time feature vector of joint motion; The extraction module is used for extracting key parameters of the joint function state based on the three-dimensional space-time feature vector by utilizing a space-time diagram convolutional neural network, capturing a cooperative motion mode of joint ligaments and bones through a self-adaptive topological structure, and obtaining a joint stability quantification index; The calculation module is used for calculating a dynamic evolution coefficient of a rehabilitation stage through a rehabilitation entropy model according to the joint stability quantization index and pain feedback data of a patient, and obtaining a quantization index of the rehabilitation stage through nonlinear weighting by fusing biomechanical parameters and physiological response parameters; the construction module is used for constructing a personalized rehabilitation action sequence by adopting an countermeasure generation network based on the quantitative index of the rehabilitation stage, verifying the medical compliance of the action sequence by a discriminator and generating a safety training instruction set conforming to the current rehabilitation stage; And the monitoring module is used for driving the wearable exoskeleton to perform rehabilitation training according to the safety training instruction set, and synchronously monitoring neuromuscular activation signals through the brain-computer interface so as to verify the consistency of training actions and nerve control.
8. The system according to claim 7, wherein the fusion module is specifically configured to: According to the joint motion data, performing six-degree-of-freedom trajectory Kalman filtering processing to obtain a noise-reduced joint motion trajectory; Based on electromyographic signals and pressure distribution information, a time domain phase alignment algorithm is adopted for processing, and a biomechanical coupling feature matrix is generated; and fusing the joint motion track and the biomechanical coupling feature matrix, and outputting a three-dimensional space-time feature vector through tensor cross decomposition processing.
9. A storage medium having a computer program stored therein, wherein the computer program is arranged to perform the method of any of claims 1-6 when run.
10. An electronic device comprising a memory and a processor, wherein the memory has stored therein a computer program, the processor being arranged to run the computer program to perform the method of any of claims 1-6.

Description

Intelligent orthopaedics joint rehabilitation monitoring method and system Technical Field The invention belongs to the technical field of medical treatment, and particularly relates to a method and a system for intelligent orthopedic joint rehabilitation monitoring. Background In the rehabilitation process after the orthopaedics joint operation, the traditional rehabilitation monitoring method mainly relies on regular examination of doctors and subjective feedback of patients, and has the problems of untimely monitoring, discontinuous data, inaccurate assessment and the like. Existing wearable devices can collect partial motion data, but generally lack comprehensive monitoring capability on multidimensional mechanical parameters (such as angles, pressures and torques) of joints, and biomechanical differences of active motion and passive motion are difficult to distinguish. In addition, the multi-modal fusion analysis of rehabilitation data is insufficient, resulting in an inability to accurately assess rehabilitation progress and predict complications risk. Disclosure of Invention The invention aims to provide an intelligent orthopaedics joint rehabilitation monitoring method and system, which are used for solving the defects in the prior art and realizing accurate, continuous and self-adaptive joint rehabilitation overall process monitoring and management. One embodiment of the application provides a smart orthopedic joint rehabilitation monitoring method, which comprises the following steps: According to joint motion data, muscle electric signals and pressure distribution information acquired by the multi-mode sensor, performing dynamic biomechanical modeling through a space-time feature fusion algorithm to obtain a three-dimensional space-time feature vector of joint motion; Based on the three-dimensional space-time feature vector, extracting key parameters of the joint function state by utilizing a space-time diagram convolutional neural network, and capturing a cooperative motion mode of joint ligaments and bones through a self-adaptive topological structure to obtain a joint stability quantization index; Calculating a dynamic evolution coefficient of a rehabilitation stage through a rehabilitation entropy model according to the joint stability quantization index and pain feedback data of a patient, and carrying out nonlinear weighting through fusion of biomechanical parameters and physiological response parameters to obtain a quantization index of the rehabilitation stage; Based on the quantitative index of the rehabilitation stage, constructing a personalized rehabilitation action sequence by adopting an countermeasure generation network, verifying the medical compliance of the action sequence by a discriminator, and generating a safety training instruction set conforming to the current rehabilitation stage; And driving the wearable exoskeleton to perform rehabilitation training according to the safety training instruction set, and synchronously monitoring neuromuscular activation signals through a brain-computer interface to verify the consistency of training actions and nerve control. Optionally, the dynamic biomechanical modeling is performed by a space-time feature fusion algorithm according to the joint motion data, the muscle electrical signals and the pressure distribution information acquired by the multi-mode sensor, so as to obtain a three-dimensional space-time feature vector of the joint motion, which comprises: According to the joint motion data, performing six-degree-of-freedom trajectory Kalman filtering processing to obtain a noise-reduced joint motion trajectory; Based on electromyographic signals and pressure distribution information, a time domain phase alignment algorithm is adopted for processing, and a biomechanical coupling feature matrix is generated; and fusing the joint motion track and the biomechanical coupling feature matrix, and outputting a three-dimensional space-time feature vector through tensor cross decomposition processing. Optionally, based on the three-dimensional space-time feature vector, extracting key parameters of a joint function state by using a space-time convolutional neural network, capturing a cooperative motion mode of a joint ligament and a bone through a self-adaptive topological structure, and obtaining a joint stability quantization index, including: according to the three-dimensional space-time feature vector, carrying out dynamic skeleton topology construction processing to generate a self-adaptive biomechanical diagram structure; based on the self-adaptive biomechanical diagram structure, adopting layered space-time diagram convolution processing to extract ligament cooperative motion characteristics; Fusing ligament cooperative motion characteristics, and obtaining a key parameter set of joint stability through attention weighted pooling treatment; Based on the joint stability key parameter set, combining clinical evaluation standards to per