Bone joint disease rehabilitation training optimization system

Abstract

The invention discloses an osteoarticular disease rehabilitation training optimization system, and relates to the field of medical training, and the system comprises an authority control module which is used for connecting an electronic medical system, obtaining the reading authority of historical and current medical data of a patient, and obtaining the instruction control and data reading authority of each functional module; the multimode acquisition module is used for acquiring pathological bone joints, associated musculoskeletal data and personal basic data through the authority control module, and converting original data into feature vectors of muscle activation degree, bone joint activity degree and skeleton stress distribution of a corresponding area of a patient body; by integrating medical data, musculoskeletal features and economic condition information of a patient, a personalized model is constructed to output an optimal adaptive training item and specific data thereof, the compensatory load of a non-ill-conditioned joint is evaluated, and timely intervention is performed through a fusing mechanism when abnormity is monitored in real time, so that the safety is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of medical training, and in particular to a bone and joint disease rehabilitation training optimization system. Background Art

[0002] Bone and joint diseases, such as arthritis and bone and joint injuries, are common health problems worldwide, especially among the elderly. These diseases not only affect the quality of life of patients, but also increase the medical and economic burden. Therefore, rehabilitation training is essential for recovery. However, existing training methods often focus on a single diseased joint, while ignoring the coordination of other joints and muscles throughout the body, resulting in unsatisfactory rehabilitation results. Traditional rehabilitation training lacks personalization, leading to poor results or secondary injuries. For example, individual differences, such as muscle condition and joint load capacity, are not taken into account. Many patients are constrained by economic conditions and cannot accept high-cost, highly specialized or difficult training programs, resulting in the loss of the meaning of training optimization. In addition, the patient's rehabilitation process is dynamic. As the training progresses, the physical condition and rehabilitation process will change. Therefore, there is a need for a flexible and optimized system for rehabilitation training of bone and joint diseases that can monitor and evaluate the patient's training progress in real time and adjust the training parameters in a timely manner to optimize the patient's rehabilitation effect. Summary of the invention

[0003] 1. Technical issues to be resolved In view of the above-mentioned shortcomings of the prior art, the present invention provides a bone and joint disease rehabilitation training optimization system, which can effectively solve the problems of the prior art.

[0004] (II) Technical solution To achieve the above objectives, the present invention is implemented through the following technical solutions: The present invention discloses a bone and joint disease rehabilitation training optimization system, comprising: The authority control module is used to connect to the electronic medical system, obtain the reading authority of the patient's historical and current medical data, and obtain the instruction control and data reading authority of each functional module; The multi-mode acquisition module is used to obtain pathological bone and joint, related muscle and bone and personal basic data through the authority control module, and convert the original data into the characteristic vectors of muscle activation, bone and joint mobility and bone stress distribution in the corresponding area of ​​the patient's body; The state assessment module is used to build a personalized musculoskeletal model, input the current patient feature vector, and generate a joint chain risk list; The phase generation module is used to automatically divide the current patient's training cycle into the acute phase, recovery phase, and intensive phase, and to match candidate training items in the preset standard training action item library; The adaptation analysis model is used to build a multi-task learning model using the deep forest algorithm and generate the adaptation coefficients of the training intensity index and frequency index of the corresponding training items; Configuration module, used to obtain the adaptation coefficient obtained by the adaptation analysis model and adjust the training items in the current training cycle; The reality-assisted module is used to build a holographic projection model of the current patient's skeletal muscle system, render the changes in force distribution of pathological joints in real time, mark when compensatory movements of associated bones and joints caused by pathological joints are detected, provide corrective guidance through spatial audio prompts, and immediately activate the fuse mechanism when the patient's physical condition is abnormal.

[0005] Furthermore, the basic personal data collected by the multi-modal collection module includes: comprehensive physical fitness score, economic strength score and willingness score for each training item.

[0006] Furthermore, the stage generation module is interactively connected to a storage module via a wireless network, and the storage module is used to classify the training items according to economic costs and training bone and joint properties, synchronously record the pressure index of the training items on associated non-pathological joints under different training intensities and frequencies, read the training item update data of the electronic medical system in real time, and synchronously upload historical patient training data.

[0007] Furthermore, the stage generation module is used to automatically divide the training cycle of the current patient into the acute phase, the recovery phase and the intensive phase, and the specific steps of matching the candidate training items in the preset standard training action project library are as follows: presetting a number of standard training action project libraries, and annotating each action with an economic cost index, a space requirement level and an instrument dependency, obtaining the patient's personal basic data through the multi-mode acquisition module, automatically dividing the training cycle of the acute phase, the recovery phase and the intensive phase according to the preset training standards, matching the candidate training items in the standard training action project library according to different training cycles through the constraint satisfaction algorithm, and generating a candidate training set; The automatic division logic of the acute phase, recovery phase, and intensive phase is as follows: Acute phase: When the patient's inflammatory index or pain score is higher than the preset threshold, it is triggered and the low-load training mode is locked; Recovery period: When the joint range of motion is restored and muscle strength reaches the preset threshold, the training intensity begins to be gradually increased; Strengthening phase: Based on isokinetic strength testing, the system is activated when the ratio of muscle strength between the affected side and the healthy side reaches the preset threshold to further improve the muscle strength of the affected part.

[0008] Furthermore, the adaptation analysis model is used to construct a multi-task learning model using a deep forest algorithm, and the specific steps of generating the adaptation coefficient of the training intensity index and the frequency index of the corresponding training project are as follows: the multi-task learning model is constructed using the deep forest algorithm, the input layer receives the patient's body index and the joint chain risk list obtained by the state assessment module and the candidate training set of the stage generation module, and the association weights of each physiological index and the training parameter are calculated through the attention mechanism, and the output layer generates the adaptation coefficient of the training intensity index and the frequency index of the corresponding training project in parallel, and the adaptation coefficient calculation is formalized into the following multi-stage hybrid calculation framework: ; In the formula, represents the training intensity index of project j in cycle k, represents the training frequency index of project j in cycle k, Represents the total number of projects. represents the attention weight of the intensity task, represents the attention weight of the frequency task, represents the deep forest feature extraction function, represents the characteristic evolution function driven by the stochastic differential equation, represents the ith physiological feature vector, represents the feature vector of training item j, represents the weight of the rehabilitation phase of cycle k, represents the revenue ratio of project j, represents the physiological tolerance of cycle k, represents the device fit of item j, represents the dynamic attenuation coefficient, represents a numerical stability constant, and is the activation function; The working logic of the hybrid computing framework is as follows: extract the feature combinations of items that are sensitive to intensity and frequency respectively through the attention weights α and β, use deep forests for multi-scale feature fusion, capture nonlinear interaction effects, introduce period-related adjustment factors for dynamic range scaling, and finally constrain the output range through the activation function.

[0009] Furthermore, the reality assistance module is interactively connected to a dynamic adjustment module via a wireless network, and the dynamic adjustment module is electrically connected to the configuration module for reading the patient stage test data of the reality assistance module, and based on the marked compensatory movements and abnormal patient physical condition data, comparing the deviation between the actual physiological indicator change rate and the expected rehabilitation trajectory, using a fuzzy logic algorithm to map the degree of completion into a number of gear adjustment intensities, searching for the optimal gear adjustment intensity plan in the parameter space through a genetic algorithm, generating an adjustment instruction set including alternative movement suggestions and auxiliary equipment configuration plan elements, and submitting it to the configuration module to participate in the next stage of periodic training items and the corresponding intensity and frequency adjustments.

[0010] Furthermore, the reality assistance module is interactively connected to a progress display module via a wireless network. The progress display module is used to construct an interactive rehabilitation dashboard, design a dual timeline comparison view to display the changing trends of various physical indicators of patients, convert the rehabilitation progress into the growth status of the virtual ecosystem, generate a PDF report that complies with medical document specifications, and automatically mark the international standard scores of key indicators.

[0011] Furthermore, the items of the physical fitness index change trend in the progress display module include: joint range of motion improvement rate, ultrasonic detection of tendon thickness change and pain diary score curve.

[0012] Furthermore, the reality auxiliary module is deployed with submodules at the lower level, including: a monitoring feedback module, a compensation calculation module and a fuse protection module, the monitoring feedback module is interactively connected with the compensation calculation module through a wireless network, and the compensation calculation module is interactively connected with the fuse protection module through a wireless network, wherein: Monitoring and feedback module, used to deploy wearable detection equipment and image acquisition equipment to extract the joint motion trajectory and three-dimensional mechanical parameters of patients during training; The compensation calculation module is used to calculate the load distribution of each joint in the patient's training movements based on the data obtained by the monitoring feedback module. When it is detected that the pressure on the healthy joint exceeds the safety threshold, the voice correction prompt is automatically triggered; The fuse protection module is used to monitor the heart rate variability and local blood oxygen saturation in real time during the patient's training. When a sudden pain signal or abnormal load pattern is detected, the fuse mechanism is immediately activated, and based on the abnormal content, it selects to perform reduced-order action replacement, intensity attenuation or emergency braking protection measures.

[0013] Furthermore, the authority control module is interactively connected to the multi-mode acquisition module through a wireless network, the multi-mode acquisition module is interactively connected to the status assessment module through a wireless network, the adaptation analysis model is interactively connected to the status assessment module, the stage generation module and the configuration module through a wireless network, and the reality assistance module is interactively connected to the configuration module and the progress display module through a wireless network.

[0014] (III) Beneficial effects Compared with the known prior art, the technical solution provided by the present invention has the following beneficial effects: By integrating multi-dimensional information such as the patient's medical data, musculoskeletal characteristics, and economic conditions, and building a personalized model through data cleaning and feature extraction, finite element analysis is used to evaluate the compensatory load of non-pathological joints, and a joint chain risk list is generated. The cascade propagation of the compensatory effect is quantified by the joint chain risk list, and a dynamic equilibrium equation of the muscle co-contraction index and ligament load distribution is established. Biomechanical simulation is used to predict the long-term injury risk of non-pathological joints. Compared with traditional static models, it can more accurately predict the potential risks in the rehabilitation process.

[0015] By combining deep forest multi-task learning and attention mechanism, the system can dynamically adjust training parameters, calculate the association weights between physiological indicators and training items through the attention mechanism, quantify the implicit association between physiological indicators and training parameters, and the multi-task model simultaneously optimizes intensity and frequency and searches for the optimal adjustment plan. The dual timeline comparison view drives the fitness function design of the genetic algorithm, and introduces a fuse mechanism as an evolutionary constraint. The fuse mechanism can be used to intervene in a timely manner when abnormalities are detected in real-time, ensuring that the search space meets the clinical safety boundary and improves safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0017] Figure 1 It is a schematic diagram of the framework of the present invention; Figure 2 It is a schematic diagram of the framework of the reality assistance module in the present invention.

[0018] The numbers in the figure represent, respectively, 1. authority control module; 2. multi-mode acquisition module; 3. state assessment module; 4. stage generation module; 5. adaptation analysis model; 6. configuration module; 7. reality assistance module; 71. monitoring feedback module; 72. compensation calculation module; 73. fuse protection module; 8. dynamic adjustment module; 9. progress display module; 10. storage module. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0020] The present invention will be further described below in conjunction with the embodiments.

[0021] Embodiment 1: A bone and joint disease rehabilitation training optimization system of this embodiment, such as Figure 1 As shown, including: The authority control module 1 is used to connect to the electronic medical system, obtain the reading authority of the patient's historical and current medical data, and obtain the instruction control and data reading authority of each functional module; obtain the patient's historical and current medical data in real time, realize the continuity and integrity of information, and ensure the accuracy and scientificity when formulating training plans; The multi-mode acquisition module 2 is used to obtain the basic data of pathological bones and joints, related muscles and bones, and individuals through the authority control module 1, eliminate motion artifacts through the data cleaning algorithm, and use the feature extraction algorithm to convert the original data into the characteristic vectors of muscle activation, bone and joint mobility, and bone stress distribution in the corresponding area of ​​the patient's body; the basic personal data includes: comprehensive physical fitness score, economic strength score, and willingness score for each training project; through data cleaning and feature extraction, artifacts can be eliminated, data quality can be improved, and accurate basic data can be provided for the formulation of personalized rehabilitation programs, directly reflecting the patient's physical condition; State assessment module 3 is used to build a personalized skeletal muscle model. After inputting the current patient's characteristic vector, it calculates the torque distribution ratio of each joint, the ligament load distribution and the body index of muscle co-contraction. It evaluates the compensatory load of non-pathological associated bone joints through finite element analysis and generates a joint chain risk list. It calculates the torque distribution and muscle co-contraction of each joint based on the specific characteristics of the patient, providing a scientific basis for formulating a personalized rehabilitation plan and effectively improving the rehabilitation effect. The stage generation module 4 is used to preset several standard training action project libraries, and annotate each action with an economic cost index, space requirement level and device dependence. The patient's personal basic data obtained by the multi-mode acquisition module 2 is automatically divided into the acute phase, recovery phase and intensive phase training cycles according to the preset training standards. The constraint satisfaction algorithm is used to match the candidate training projects in the standard training action project library according to different training cycles to generate a candidate training set; by automatically dividing the training cycle and matching the corresponding training projects, it is ensured that the patient receives the most appropriate training at different recovery stages to minimize the risk of injury; The stage generation module 4 is interactively connected to the storage module 10 via a wireless network. The storage module 10 is used to classify the training items according to the economic cost and the training bone and joint properties, synchronously record the pressure index of the training items on the associated non-pathological joints at different training intensities and frequencies, read the training item update data of the electronic medical system in real time, and synchronously upload the historical patient training data; The automatic division logic of the acute phase, recovery phase, and intensive phase in phase generation module 4 is as follows: Acute phase: When the patient's inflammatory index or pain score is higher than the preset threshold, it is triggered and the low-load training mode is locked, specifically: inflammatory index: CRP>20mg / L, ESR>30mm / h, pain score: VAS≥4 points; Recovery period: activated when the joint range of motion recovery and muscle strength reach the preset thresholds, specifically: joint range of motion recovery: affected side ROM ≥ 80% of the healthy side, muscle strength: affected side muscle strength ≥ 70% of the healthy side; Strengthening phase: Based on isokinetic strength test, the system is activated when the ratio of the affected side to the healthy side reaches the preset threshold, specifically: affected side / healthy side ≥ 80%, and the thresholds are manually preset; The adaptation analysis model 5 is used to construct a multi-task learning model using a deep forest algorithm. The input layer receives the patient's body index and joint chain risk list obtained by the state assessment module 3 and the candidate training set of the stage generation module 4, and calculates the association weights of each physiological index and the training parameter through the attention mechanism. The output layer generates the adaptation coefficients of the training intensity index and the frequency index of the corresponding training project in parallel. Through the multi-task learning model, the system can adjust the training intensity and frequency in real time to ensure the adaptability and scientificity of the training and improve the training effect. Configuration module 6 is used to obtain the adaptation coefficient obtained by the adaptation analysis model 5, and adjust the training items in each current training cycle; real-time monitoring of the patient's physical condition can timely discover and correct improper movements, maximize joint protection, reduce the probability of injury, and enhance user experience; Reality assistance module 7 is used to build a holographic projection model of the current patient's skeletal muscle system, monitor the patient's physical condition, render the force distribution changes of the pathological joints in real time, mark the associated bone and joint compensatory movements caused by the pathological joints, provide correction guidance through spatial audio prompts, and immediately activate the fuse mechanism when the patient's physical condition is abnormal; real-time monitoring of joint point dynamics, feedback of accurate motion parameters, to ensure the safety and effectiveness of training; like Figure 2 As shown, the reality auxiliary module 7 is deployed with submodules at the lower level, including: a monitoring feedback module 71, a compensation calculation module 72 and a fuse protection module 73. The monitoring feedback module 71 is interactively connected with the compensation calculation module 72 through a wireless network, and the compensation calculation module 72 is interactively connected with the fuse protection module 73 through a wireless network, wherein: Monitoring and feedback module 71, used to deploy wearable detection equipment and image acquisition equipment to extract joint motion trajectories and three-dimensional mechanical parameters of patients during training; The compensation calculation module 72 is used to calculate the load distribution of each joint in the patient's training action based on the data obtained by the monitoring feedback module 71, and automatically trigger the voice correction prompt when it is detected that the pressure on the healthy joint exceeds the safety threshold; The fuse protection module 73 is used to monitor the heart rate variability and local blood oxygen saturation in real time during the patient's training process. When a sudden pain signal or an abnormal load mode is detected, the fuse mechanism is immediately activated, and the downgraded action replacement, intensity attenuation or emergency braking protection measures are selected based on the abnormal content; the real-time monitoring of heart rate and blood oxygen saturation can timely judge the patient's physical condition, ensure the patient's safety, and adjust the training plan in time; The dynamic adjustment module 8 is used to read the patient stage test data of the reality assistance module 7, compare the deviation between the actual physiological index change rate and the expected rehabilitation trajectory based on the marked compensatory movements and abnormal patient physical condition data, use fuzzy logic algorithm to map the completion degree into several gear adjustment intensities, search for the optimal gear adjustment intensity plan in the parameter space through genetic algorithm, generate an adjustment instruction set containing alternative movement suggestions and auxiliary equipment configuration plan elements, and submit it to the configuration module 6 to participate in the next stage of periodic training items and the corresponding intensity and frequency adjustment; by comparing the deviation between the actual physiological indicators and the expected rehabilitation trajectory, it can intelligently adjust the training intensity, provide scientific rehabilitation feedback, and promote the patient's continuous improvement and recovery.

[0022] Compared with the existing technology, this system establishes a training optimization system based on individualized evaluation, which can better adapt to the patient's physical condition and economic situation. Through the deep forest algorithm and multi-task learning model, it can update and optimize the training intensity and frequency in real time during the training process to ensure adaptability and effectiveness. Many existing rehabilitation training programs fail to be adjusted in time according to the actual situation of the patient, which easily leads to inappropriate training intensity or excessive economic burden. This system can comprehensively monitor motion trajectories and physiological parameters, quickly identify and feedback improper movements or abnormal conditions, significantly reduce the risk of injury, and take into account the status of healthy bones and joints to avoid compensatory impairment during training. However, many existing rehabilitation training programs often lack real-time monitoring of the training process, which makes it difficult to take timely measures when patients have abnormalities.

[0023] Embodiment 2: This embodiment is further optimized at other levels to provide a multi-stage hybrid computing framework for formalizing the calculation of the adaptation coefficient of a multi-task learning model, specifically: ; In the formula, represents the training intensity index of project j in cycle k, represents the training frequency index of project j in cycle k, Represents the total number of projects. Represents the attention weight of the intensity task, calculated by bilinear transformation, represents the attention weight of the frequency task, represents the deep forest feature extraction function, which generates high-order interactive features through multi-granularity scanning. represents the characteristic evolution function driven by the stochastic differential equation, represents the i-th physiological feature vector, and its dimensions include joint torque distribution ratio, ligament load distribution entropy, muscle co-contraction index, compensatory joint stress concentration coefficient, joint chain risk level, inflammatory factor attenuation rate, activity recovery gradient and muscle strength balance ratio. The feature vector representing training item j, whose dimensions include economic cost index, space requirement level, equipment dependence, action complexity, historical success rate and multi-joint coupling degree, represents the weight of the rehabilitation phase of cycle k, represents the revenue ratio of project j, represents the physiological tolerance of cycle k, represents the device fit of item j, represents the dynamic attenuation coefficient, represents a numerical stability constant, and is the activation function.

[0024] As a preferred implementation method in this embodiment, the working logic of the above-mentioned hybrid computing framework is: extracting the feature combinations of items that are sensitive to intensity and frequency respectively through attention weights α and β, using deep forests for multi-scale feature fusion, capturing nonlinear interaction effects, introducing period-related adjustment factors for dynamic range scaling, and finally constraining the output range through activation functions, establishing a dynamic association between physiological indicators and training parameters through the attention mechanism, using deep forests to process the nonlinear relationship of high-dimensional heterogeneous features, introducing adjustment factors such as rehabilitation stage and risk-benefit ratio to enhance decision-making robustness, and the dual-task output layer design to ensure the coordinated optimization of intensity and frequency parameters.

[0025] Embodiment 3: In this embodiment, as Figure 1As shown, the reality assistance module 7 is interactively connected to the progress display module 9 through a wireless network. The progress display module 9 is used to build an interactive rehabilitation dashboard, design a dual time axis comparison view to display the change trend of each physical indicator of the patient, convert the rehabilitation progress into the growth state of the virtual ecosystem, generate a PDF report that complies with the medical document specifications, and automatically annotate the international standard scores of key indicators; the project content of the physical indicator change trend of the progress display module 9 includes: joint range of motion improvement rate, ultrasonic detection of tendon thickness changes and pain diary score curve.

[0026] This embodiment adopts dual-time axis dynamic comparison technology to achieve spatiotemporal alignment of biomechanical indicators and histological data, such as joint range of motion and ultrasonic tendon thickness, and establishes causal relationship mapping between pain score curve and physical indicators, which improves user experience, data interpretation efficiency and accuracy of medical assessment, and is more comprehensive and user-friendly than existing technologies.

[0027] Working principle: In the specific implementation of the present invention, the authority control module 1 obtains the patient's historical and current medical data authority through the electronic medical system, the multi-mode acquisition module 2 integrates the pathological bone joint and related muscle and skeleton data and performs feature extraction to generate feature vectors, the state evaluation module 3 constructs a personalized skeletal muscle model to calculate the joint torque distribution ratio and generate a joint chain risk list, the stage generation module 4 divides the training cycle according to the patient's basic data and matches the candidate training items through the constraint satisfaction algorithm, the adaptation analysis model 5 uses the deep forest multi-task model to calculate the training intensity and frequency adaptation coefficient, the configuration module 6 adjusts the training items of each cycle according to the adaptation coefficient, the reality auxiliary module 7 monitors the training process in real time through the holographic projection model and marks the compensatory actions, the dynamic adjustment module 8 reads the stage test data and compares the deviation to generate an adjustment instruction set and feeds it back to the configuration module 6, forming a closed-loop optimization process, wherein the stage generation module 4 realizes the classification of training items and data synchronization through the storage module 10, the reality auxiliary module 7 generates a visual report through the progress display module 9 and implements real-time monitoring through the monitoring feedback module 71, the compensation calculation module 72 performs compensation calculation, and the fuse protection module 73 provides fuse protection; In summary, the present invention realizes precise individual modeling through multi-source medical data fusion and feature engineering, constructs a dynamic risk assessment system using finite element analysis and deep forest algorithm, realizes adaptive adjustment of training parameters based on a two-level optimization mechanism of constraint satisfaction and genetic algorithm, ensures training safety by relying on real-time biofeedback strategy, and improves the feasibility of the scheme by combining a multi-objective decision-making model with economic cost and physiological indicators, thus forming a closed-loop intelligent rehabilitation ecosystem covering data collection, model calculation, action execution, and effect evaluation.

[0028] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A bone and joint disease rehabilitation training optimization system, characterized in that: include: The authority control module (1) is used to connect to the electronic medical system, obtain the reading authority of the patient's historical and current medical data, and obtain the instruction control and data reading authority of each functional module; The multi-mode acquisition module (2) is used to obtain the pathological bone and joint, related muscle and bone and personal basic data through the authority control module (1), and convert the original data into the characteristic vector of the muscle activation degree, bone and joint mobility and bone stress distribution of the corresponding area of ​​the patient's body; The state assessment module (3) is used to build a personalized musculoskeletal model, input the current patient feature vector, and generate a joint chain risk list; The phase generation module (4) is used to automatically divide the training period of the current patient into the acute phase, the recovery phase and the strengthening phase, and to match the candidate training items in the preset standard training action item library; An adaptation analysis model (5) is used to construct a multi-task learning model using a deep forest algorithm and generate adaptation coefficients of training intensity indicators and frequency indicators corresponding to training items; A configuration module (6) is used to obtain the adaptation coefficient obtained by the adaptation analysis model (5) and adjust the training items in each current training cycle; The reality assistance module (7) is used to construct a holographic projection model of the current patient's musculoskeletal system, render the force distribution changes of the pathological joints in real time, mark the associated bone and joint compensatory movements caused by the pathological joints when they are detected, provide correction guidance through spatial audio prompts, and immediately activate the fuse mechanism when the patient's physical condition is abnormal.

2. The bone and joint disease rehabilitation training optimization system according to claim 1 is characterized in that: The basic personal data collected by the multi-mode collection module (2) includes: a comprehensive physical fitness score, an economic strength score, and a willingness score for each training item.

3. The bone and joint disease rehabilitation training optimization system according to claim 1 is characterized in that: The stage generation module (4) is interactively connected to a storage module (10) via a wireless network. The storage module (10) is used to classify training items according to economic costs and training bone and joint properties, synchronously record the pressure index of the training items on the associated non-pathological joints at different training intensities and frequencies, read the training item update data of the electronic medical system in real time, and synchronously upload historical patient training data.

4. The bone and joint disease rehabilitation training optimization system according to claim 1, characterized in that: The stage generation module (4) is used to automatically divide the training cycle of the current patient into the acute phase, the recovery phase and the intensive phase. The specific steps of matching candidate training projects in the preset standard training action project library are as follows: presetting a number of standard training action project libraries, and annotating each action with an economic cost index, a space requirement level and a device dependency, using the patient's personal basic data collected by the multi-mode acquisition module (2), automatically dividing the training cycle of the acute phase, the recovery phase and the intensive phase according to the preset training standards, matching candidate training projects in the standard training action project library according to different training cycles through a constraint satisfaction algorithm, and generating a candidate training set; The automatic division logic of the acute phase, recovery phase, and intensive phase is as follows: Acute phase: When the patient's inflammatory index or pain score is higher than the preset threshold, it is triggered and the low-load training mode is locked; Recovery period: When the joint range of motion is restored and muscle strength reaches the preset threshold, the training intensity begins to be gradually increased; Strengthening phase: Based on isokinetic strength testing, the system is activated when the ratio of muscle strength between the affected side and the healthy side reaches the preset threshold to further improve the muscle strength of the affected part.

5. The bone and joint disease rehabilitation training optimization system according to claim 1, characterized in that: The adaptation analysis model (5) is used to construct a multi-task learning model using a deep forest algorithm to generate the adaptation coefficient of the training intensity index and frequency index of the corresponding training project. The specific steps are: constructing the multi-task learning model using a deep forest algorithm, the input layer receives the patient's body index and joint chain risk list obtained by the state evaluation module (3) and the candidate training set of the stage generation module (4), calculates the association weight of each physiological index and the training parameter through the attention mechanism, and the output layer generates the adaptation coefficient of the training intensity index and frequency index of the corresponding training project in parallel. The adaptation coefficient calculation is formalized into the following multi-stage hybrid calculation framework: ； In the formula, represents the training intensity index of project j in cycle k, represents the training frequency index of project j in cycle k, Represents the total number of projects. represents the attention weight of the intensity task, represents the attention weight of the frequency task, represents the deep forest feature extraction function, represents the characteristic evolution function driven by the stochastic differential equation, represents the ith physiological feature vector, represents the feature vector of training item j, represents the weight of the rehabilitation phase of cycle k, represents the revenue ratio of project j, represents the physiological tolerance of cycle k, represents the device fit of item j, represents the dynamic attenuation coefficient, represents a numerical stability constant, and is the activation function; The working logic of the hybrid computing framework is as follows: extract the feature combinations of items that are sensitive to intensity and frequency respectively through the attention weights α and β, use deep forests for multi-scale feature fusion, capture nonlinear interaction effects, introduce period-related adjustment factors for dynamic range scaling, and finally constrain the output range through the activation function.

6. The bone and joint disease rehabilitation training optimization system according to claim 1, characterized in that: The reality assistance module (7) is interactively connected to a dynamic adjustment module (8) via a wireless network. The dynamic adjustment module (8) is electrically connected to the configuration module (6) and is used to read the patient stage test data of the reality assistance module (7). Based on the marked compensatory movements and abnormal patient physical condition data, the actual physiological index change rate is compared with the deviation of the expected rehabilitation trajectory. The fuzzy logic algorithm is used to map the completion degree into a number of gear adjustment intensities. The genetic algorithm is used to search for the optimal gear adjustment intensity plan in the parameter space, and an adjustment instruction set containing alternative movement suggestions and auxiliary equipment configuration plan elements is generated and submitted to the configuration module (6) to participate in the adjustment of the next stage cycle training items and the corresponding intensity and frequency.

7. The bone and joint disease rehabilitation training optimization system according to claim 1, characterized in that: The reality assistance module (7) is interactively connected to a progress display module (9) via a wireless network. The progress display module (9) is used to construct an interactive rehabilitation dashboard, design a dual time axis comparison view to display the changing trend of each physical index of the patient, convert the rehabilitation progress into the growth state of the virtual ecosystem, generate a PDF report that complies with medical document specifications, and automatically annotate the international standard scores of key indicators.

8. The bone and joint disease rehabilitation training optimization system according to claim 7, characterized in that: The items of the physical index change trend in the progress display module (9) include: joint range of motion improvement rate, ultrasonic detection of tendon thickness change and pain diary score curve.

9. The bone and joint disease rehabilitation training optimization system according to claim 1, characterized in that: The reality auxiliary module (7) is provided with submodules at a lower level, including: a monitoring feedback module (71), a compensation calculation module (72) and a fuse protection module (73), wherein the monitoring feedback module (71) and the compensation calculation module (72) are interactively connected via a wireless network, and the compensation calculation module (72) and the fuse protection module (73) are interactively connected via a wireless network, wherein: A monitoring feedback module (71) is used to deploy wearable detection equipment and image acquisition equipment to extract the joint motion trajectory and three-dimensional mechanical parameters of the patient during training; A compensation calculation module (72) is used to calculate the load distribution of each joint in the patient's training action based on the data obtained by the monitoring feedback module (71), and automatically trigger a voice correction prompt when it is detected that the pressure on the healthy joint exceeds the safety threshold; The fuse protection module (73) is used to monitor the heart rate variability and local blood oxygen saturation in real time during the patient's training process. When a sudden pain signal or an abnormal load mode is detected, the fuse mechanism is immediately activated, and based on the abnormal content, a reduced-order action replacement, intensity reduction or emergency braking protection measure is selected.

10. The bone and joint disease rehabilitation training optimization system according to claim 1, characterized in that: The authority control module (1) is interactively connected to the multi-mode acquisition module (2) via a wireless network, the multi-mode acquisition module (2) is interactively connected to the state evaluation module (3) via a wireless network, the adaptation analysis model (5) is interactively connected to the state evaluation module (3), the stage generation module (4) and the configuration module (6) via a wireless network, and the reality assistance module (7) is interactively connected to the configuration module (6) and the progress display module (9) via a wireless network.

Images