Wearable joint rehabilitation training device and pressure self-adaptive control method

Abstract

The present application belongs to the technical field of rehabilitation medical apparatus and discloses a wearable joint rehabilitation training device and a pressure self-adaptive control method. The training device comprises a flexible wearable main body, which is prepared by using medical-grade silica gel and knitted flexible composite material. Modular replaceable structural components are used to realize cross-joint and cross-injury-level adaptation and replacement. A multi-sensor feature acquisition module is used to acquire multi-dimensional feature data of human-machine contact in real time. A pressure self-adaptive driving adjustment module realizes single-region and multi-region differential pressure adjustment. A main control processing module is used for intelligent control, scheme generation and effect evaluation whole-process management and control. A man-machine interaction terminal is used to receive user manual adjustment instructions and display rehabilitation evaluation results. The wearable joint rehabilitation training device and the pressure self-adaptive control method are used to realize comfortable and adaptive equipment wearing, personalized and customized rehabilitation parameter adjustment and pressure self-adaptive precise output, thereby effectively improving the pertinence, safety and efficiency of joint rehabilitation training.

Description

Technical Field

[0001] This invention relates to the field of rehabilitation medical device technology, and in particular to a wearable joint rehabilitation trainer and a pressure adaptive control method. Background Technology

[0002] Joint injuries, postoperative immobilization, and degenerative joint diseases leading to limited joint mobility and muscle weakness are common clinical rehabilitation problems. Wearable joint rehabilitation devices, with their portability and home-based training advantages, have become one of the core devices in rehabilitation medicine. Currently, traditional wearable joint rehabilitation training devices on the market suffer from two major technical defects: First, poor wearability. Most use rigid fixed structures or integrated flexible structures, which cannot adapt to the different limb sizes and joint morphologies of different users. Prolonged wear can easily lead to problems such as local pressure, slippage, and insufficient fit, seriously affecting training comfort and rehabilitation accuracy. Second, fixed control methods. Most use standardized control modes with fixed pressure, fixed movement trajectories, and fixed training frequencies, failing to provide customized dynamic adjustments based on the user's real-time joint status, tolerance, and rehabilitation progress, resulting in extremely poor adaptability.

[0003] Meanwhile, existing equipment lacks accurate multi-dimensional data collection and intelligent reasoning capabilities, and can only achieve simple mechanical movement assistance. It cannot quantify joint movement status and human-computer interaction pressure, cannot predict user rehabilitation adaptation trends, and is difficult to generate personalized rehabilitation training plans. Furthermore, there is no standardized rehabilitation assessment and calculation system, and it relies solely on the subjective judgment of medical staff on the recovery status. The accuracy and objectivity of the assessment results are insufficient, resulting in uneven rehabilitation training effects. It is very easy to cause secondary injuries due to overtraining or delayed rehabilitation due to undertraining.

[0004] Based on the shortcomings of the existing technologies, there is an urgent need to design a wearable joint rehabilitation trainer and control method that is highly adaptable, customizable, pressure-adaptive and precise, and has intelligent rehabilitation planning and quantitative assessment capabilities. Summary of the Invention

[0005] The purpose of this invention is to provide a wearable joint rehabilitation trainer and a pressure adaptive control method to achieve precise fitting for multiple groups and multiple joints, real-time dynamic adaptive pressure adjustment, intelligent generation of personalized rehabilitation training plans and quantitative evaluation of rehabilitation effects, and comprehensively improve the intelligence, precision and personalization of joint rehabilitation training.

[0006] To achieve the above objectives, the present invention provides a wearable joint rehabilitation trainer, comprising a flexible wearable body, modular and replaceable structural components, a multi-sensor feature acquisition module, a pressure adaptive drive adjustment module, a main control processing module, a human-computer interaction terminal, and a rehabilitation assessment module. The flexible wearable body is made of medical-grade silicone and highly elastic knitted flexible composite material, with no rigid protrusions, and conforms to the curvature of the human limbs; The modular replaceable structural components are available in three specifications: mild, moderate, and severe damage, to fit human joints and enable replacement across joints and damage levels. The multi-sensor feature acquisition module is used to collect multi-dimensional feature data such as human-machine contact pressure distribution, joint flexion and extension angle, joint movement displacement, muscle activation, and limb surface temperature in real time. The pressure adaptive drive adjustment module adopts an array of miniature pneumatic drive units, with independent control of corresponding sensor zones to achieve differentiated pressure adjustment in single or multiple areas, matching the force requirements of different joint positions; The main control processing module has built-in data preprocessing algorithm, pressure adaptive control algorithm, time series prediction inference model and rehabilitation assessment algorithm, which are electrically connected to the multi-sensor feature acquisition module, pressure adaptive drive adjustment module, human-computer interaction terminal and rehabilitation assessment module respectively, to realize full-process management and control of data processing, intelligent control, plan generation and effect evaluation; The human-computer interaction terminal is used to receive manual adjustment commands from users, display real-time training data, push personalized rehabilitation plans, and display rehabilitation assessment results.

[0007] Preferably, the modular replaceable structural components include: a joint movement limiting module, a size adaptation and adjustment module, and a pressure zone support module. Each module adopts a snap-on detachable structure, and the thickness, support strength, and movement limiting angle of each module are set differently, so that they can be freely replaced and combined to fit according to the user's limb size, type of rehabilitation joint, and degree of injury.

[0008] Preferably, the multi-sensor feature acquisition module includes: a flexible pressure sensor array, an angle sensor, a displacement sensor, an electromyography sensor, and a temperature sensor, which are distributed and integrated on the inner side of the flexible wearable body and the joint movement area; The flexible pressure sensor array adopts a high-density distributed layout, with 8-12 independent pressure acquisition points set in a single joint area. The sampling frequency is greater than or equal to 100Hz, and the pressure acquisition accuracy is ±0.5kPa, capturing minute local pressure changes in the joint.

[0009] A pressure adaptive control method for a wearable joint rehabilitation trainer, used to execute the wearable joint rehabilitation trainer, includes the following steps: Step S1, Personalized Wearing and Adaptation of the Device: Based on the user's rehabilitation joint type, limb size, and injury level, replace the modular structural components to complete the wearing and fixation, and achieve a close fit between the device and the limb; Step S2, Accurate Acquisition and Preprocessing of Multi-Dimensional Feature Data: Through the multi-sensor feature acquisition module, joint motion parameters, human-machine pressure parameters, and muscle state parameters are acquired in real time, and data preprocessing is completed through filtering and noise reduction and outlier removal algorithms; Step S3, Pressure Adaptive Closed-Loop Dynamic Adjustment: Based on the preprocessed data, the target pressure value of each drive unit is calculated through the pressure adaptive control algorithm, and the output pressure of each area is dynamically adjusted to achieve human-machine interaction pressure adaptive matching. Step S4: Intelligent generation of personalized rehabilitation training plans: Introducing a time-series predictive reasoning model, combining the user's historical rehabilitation data, real-time status data, and injury parameters, iteratively optimizing and generating customized rehabilitation parameters such as training angle, training frequency, stress threshold, and training duration; Step S5, Real-time Rehabilitation Training and Dynamic Correction: Execute rehabilitation training according to the customized training plan, collect feedback data in real time, and dynamically correct stress and exercise parameters; Step S6: Multi-dimensional quantitative assessment of rehabilitation effect: Quantitative indicators of joint recovery are calculated through rehabilitation assessment algorithms to determine the user's joint rehabilitation progress and recovery level, and the rehabilitation training plan is updated synchronously.

[0010] Preferably, step S1 involves personalized device wearable adaptation, and the specific process is as follows: Step S11: The main body of the equipment adopts a double-layer composite flexible structure. The inner layer is a breathable and antibacterial flexible knitted layer, and the outer layer is a high-toughness elastic silicone support layer. The overall structure has no hard edges and corners, and deforms synchronously with the flexion, extension and rotation of the joints, conforming to the movement law of the human limbs. Step S12: The modular components adopt standardized snap-fit ​​interfaces, and all module interfaces are unified, supporting free disassembly, replacement, and combination. (1) Size adaptation module: Sets four sizes: S, M, L and XL to adapt to users with different limb thicknesses; (2) Joint movement limiting module: Different maximum flexion and extension limiting angles are set for different joints: elbow joint 0°-135°, knee joint 0°-120°, ankle joint 0°-45°. The limiting module can be manually switched according to the user's injury recovery stage. (3) Pressure zone support module: It is divided into three types: uniform support, local pressure, and decompression buffer. It can achieve pressure or decompression and targeted rehabilitation for joint injury pain points and weak stress areas.

[0011] Preferably, step S2 involves accurate acquisition and preprocessing of multi-dimensional feature data, the specific process of which is as follows: Step S21: The sensors adopt an embedded hidden layout. The pressure sensor array is evenly distributed in the force-bearing area around the joint, the angle sensor is fixed at the joint rotation axis to collect the joint flexion and extension angles, and the electromyography sensor is attached to the muscle group around the joint to collect muscle activation signals. All sensor data are transmitted to the main control module in real time. Step S22: Collect real-time parameters, including: human-machine contact pressure P at each point. i Real-time joint flexion and extension angle θ, joint motion angular velocity ω, muscle activation EMG, limb temperature T, joint displacement S; Step S23: To eliminate noise data caused by motion jitter and environmental interference, a moving average filtering algorithm is used to complete data preprocessing, as shown below: ; in, This represents the effective data after filtering at time n. The original data collected by the sensor at time i; N is the number of sampling points in the sliding window; At the same time, an outlier removal threshold is set. When the deviation of a single frame of collected data from the average of the previous 5 frames is greater than 20%, it is judged as abnormal interference data and is directly removed.

[0012] Preferably, in step S3, an adaptive PID closed-loop pressure control algorithm is designed, which combines the user's real-time pressure tolerance, joint movement status, and dynamic adjustment of output pressure during the rehabilitation stage to achieve adaptive pressure control. The specific process is as follows: Step S31: Initialize the baseline target pressure value based on the user's rehabilitation level, joint type, and injury duration, as shown below: ; in, The initial baseline target pressure is given in kPa. The corresponding standard reference pressure for the joint is in kPa; Damage level coefficient; For the rehabilitation stage coefficient; Step S32: Combining real-time collected human-machine pressure deviation, joint movement status, and muscle endurance status, calculate the dynamic pressure correction amount to achieve real-time adaptive adjustment, as shown below: ; in, This is the pressure deviation value; For real-time target pressure; The sensor collects pressure in real time; These are the PID proportional, integral, and derivative coefficients, respectively. This represents the change in joint angle. The angle correction factor is kPa / °. This represents the change in muscle activation. Muscle condition correction factor, kPa; Step S33: Calculate the final real-time output pressure, as shown below: ; At the same time, a pressure safety threshold is set: the output pressure range of each area is 0-50kPa. When the calculated value exceeds the threshold, the upper and lower limits are automatically locked to avoid limb compression injury caused by excessive pressure and ineffective rehabilitation training due to excessive pressure. Step S34: To address the differences in force distribution across different areas of the joint, a zoned independent pressure control is adopted, with each drive unit corresponding to an independent pressure closed-loop control, thereby achieving differentiated pressure output on the flexion, extension, and support sides of the joint, which conforms to the biomechanical characteristics of human joints.

[0013] Preferably, in step S4, a temporal long short-term memory (LSTM) predictive inference model is introduced to predict joint rehabilitation trends based on the user's historical rehabilitation data and real-time status data, and to iteratively generate customized rehabilitation training parameters. The specific process is as follows: Step S41: The model feature inputs include: historical 7-day average training stress, average joint range of motion, muscle activation, training duration, joint recovery score, and user pain tolerance. Step S42: The model predicts future rehabilitation adaptation parameters through temporal inference and dynamically updates the training plan; Step S43: After the model automatically generates a rehabilitation plan, based on the human-computer interaction adjustment mechanism, the user can manually fine-tune the pressure, angle, and duration parameters through the human-computer interaction terminal. The main control module records the user's preference parameters and iteratively optimizes the subsequent prediction model to achieve fully customized adaptation.

[0014] Preferably, in step S6, a comprehensive joint rehabilitation assessment model is constructed to determine the joint recovery status through quantitative indicators. The assessment dimensions include four core indicators: joint range of motion, muscle strength recovery, stress tolerance, and movement stability. The specific process is as follows: Step S61: Calculation formula for core evaluation indicators; (1) Joint range of motion recovery rate As shown below: ; in, This represents the user's current maximum joint range of motion. The standard range of motion for joints in a healthy human body; (2) Muscle strength recovery rate As shown below: ; in, This represents the user's current average muscle activation level. Standard activation level for healthy muscles; (3) Pressure tolerance stability coefficient As shown below: ; in, The standard deviation of stress fluctuation during training; To average training pressure; Step S62: Comprehensive rehabilitation scoring and grading; Comprehensive Rehabilitation Score As shown below: ; in, The coefficient represents the motion stability, ranging from 0 to 1; the weights of each component are determined and optimized based on big data from clinical rehabilitation. Rehabilitation level classification: S≥90 indicates complete recovery, 70≤S<90 indicates good recovery, 50≤S<70 indicates effective recovery, and S<50 indicates poor recovery. The main control module automatically updates the next stage of rehabilitation training plan based on the scoring level.

[0015] Therefore, the present invention employs the aforementioned wearable joint rehabilitation trainer and pressure adaptive control method, and the beneficial effects are as follows: (1) Solve the problem of poor wearing comfort and adaptability: The flexible composite material combined with the modular replaceable structure can be adapted to users with different joints, different limb sizes and different degrees of injury, so as to achieve precise fit and wear without local pressure or slippage, greatly improving the comfort of long-term training.

[0016] (2) Achieve customized and precise adjustment of control mode: Abandon the traditional fixed parameter control mode, and use multi-sensor fusion acquisition and adaptive PID pressure control algorithm to dynamically adjust pressure and motion parameters in combination with real-time human body status to meet the personalized control needs of different users and different rehabilitation stages.

[0017] (3) Intelligent customization of rehabilitation plans: Introducing a time-series predictive reasoning model to generate exclusive rehabilitation training plans based on user rehabilitation data. At the same time, it supports manual fine-tuning by humans and machines to achieve personalized rehabilitation training for each individual and avoid the adaptation defects of general plans.

[0018] (4) Quantitative and accurate rehabilitation assessment: A multi-dimensional quantitative assessment system is constructed, and the objective judgment of the joint recovery status is realized through formulaic calculation, replacing the subjective judgment of humans. The rehabilitation assessment results are accurate and reliable, and can accurately guide the iterative optimization of rehabilitation training.

[0019] (5) High safety and practicality: It sets multiple pressure and angle safety thresholds to avoid secondary damage caused by training overload in real time. The equipment is portable and easy to operate, and is suitable for both hospital clinical rehabilitation and home self-rehabilitation scenarios, with a wide range of applications.

[0020] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the wearable knee joint rehabilitation trainer in Embodiment 1 of the present invention; Figure 2 This is a flowchart of the pressure adaptive control method for the wearable joint rehabilitation trainer of the present invention.

[0022] Figure Labels 1. Flexible wearable body; 2. Modular and replaceable structural components; 3. Multi-sensor feature acquisition module; 4. Pressure adaptive drive adjustment module; 5. Main control processing module. Detailed Implementation

[0023] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0024] Example 1 like Figure 1 As shown, a wearable joint rehabilitation training device includes a flexible wearable main body 1, a modular replaceable structural component 2, a multi-sensor feature acquisition module 3, a pressure adaptive drive adjustment module 4, a main control processing module 5, a human-computer interaction terminal, and a rehabilitation assessment module.

[0025] The flexible wearable body 1 is made of medical-grade silicone and high-elasticity knitted flexible composite material. It has no rigid protrusions, conforms to the curvature of the human body, and has the characteristics of breathability, pressure resistance and hypoallergenicity.

[0026] The modular replaceable structural component 2 includes: a joint movement limiting module, a size adaptation adjustment module, and a pressure zone support module. Each module adopts a snap-on detachable structure, which can be freely replaced and combined to fit according to the user's limb size, type of rehabilitation joint, and degree of injury, thus solving the problem of poor wear fit.

[0027] The modular replaceable structural components are available in three specifications: mild, moderate, and severe injury. They are compatible with six major human joints: shoulder, elbow, wrist, knee, and ankle, allowing for flexible replacement across joints and injury levels. The thickness, support strength, and range of motion of each module are differentiated to meet the joint movement restriction needs at different stages of rehabilitation.

[0028] The multi-sensor feature acquisition module 3 includes a flexible pressure sensor array, a high-precision angle sensor, a displacement sensor, an electromyography sensor, and a temperature sensor, which are distributed and integrated on the inner side of the flexible wearable body and the joint movement area. It is used to collect multi-dimensional feature data such as human-machine contact pressure distribution data, joint flexion and extension angle, joint movement displacement, muscle activation degree, and limb surface temperature in real time, providing accurate data support for adaptive control.

[0029] The flexible pressure sensor array adopts a high-density distributed layout, with 8-12 independent pressure acquisition points set in a single joint area. The sampling frequency is ≥100Hz, and the pressure acquisition accuracy is ±0.5kPa, which can accurately capture minute local pressure changes in the joint.

[0030] The pressure adaptive drive adjustment module 4 adopts an array-type micro pneumatic drive unit, with independent control of corresponding sensor zones, which can realize single-zone and multi-zone differentiated pressure adjustment to accurately match the force requirements of different joint positions.

[0031] The main control processing module 5 has built-in data preprocessing algorithms, pressure adaptive control algorithms, time-series prediction inference models, and rehabilitation assessment algorithms. It is electrically connected to the multi-sensor feature acquisition module, pressure adaptive drive adjustment module, human-computer interaction terminal, and rehabilitation assessment module, respectively, to realize full-process management and control of data processing, intelligent control, scheme generation, and effect evaluation.

[0032] The human-computer interaction terminal is used to receive manual adjustment commands from users, display real-time training data, push personalized rehabilitation plans, and display rehabilitation assessment results.

[0033] Example 2 like Figure 2 As shown, the present invention discloses a pressure adaptive control method for a wearable joint rehabilitation trainer, comprising the following steps: Step S1, Personalized Wearing and Adaptation of the Device: Based on the user's rehabilitation joint type, limb size, and injury level, replace the modular structural components to complete the wearing and fixation, achieving a precise fit between the device and the limb.

[0034] Step S11: Overall flexible structure design.

[0035] The main body of the device adopts a double-layer composite flexible structure. The inner layer is a breathable and antibacterial flexible knitted layer that comes into direct contact with the human skin, avoiding stuffiness and allergies from prolonged wear. The outer layer is a high-toughness elastic silicone support layer, ensuring structural support while preserving the freedom of joint movement, avoiding the problems of strong restraint and poor comfort associated with traditional rigid structures. The overall structure has no hard edges and corners, and can deform synchronously with joint flexion, extension, and rotation movements, conforming to the movement patterns of the human limbs.

[0036] Step S12: Modular replaceable structure implementation.

[0037] Modular components use standardized snap-fit ​​interfaces, all modules have the same interface, and support free disassembly, replacement, and combination: (1) Size adaptation module: Sets four sizes: S, M, L and XL to fit users with different limb sizes. It can accurately adjust the tightness of the wear and eliminate the problem of gap slippage. (2) Joint movement limiting module: Different maximum flexion and extension limiting angles are set for different joints, such as elbow joint 0°-135°, knee joint 0°-120°, and ankle joint 0°-45°. The limiting module can be manually switched according to the user's injury recovery stage to avoid excessive joint movement. (3) Pressure zone support module: It is divided into three types: uniform support, local pressure, and pressure reduction buffer. It can accurately apply pressure or reduce pressure to joint injury pain points and weak stress areas, thereby improving wearing comfort and rehabilitation targeting.

[0038] Step S2, Accurate Acquisition and Preprocessing of Multi-Dimensional Feature Data: Through the multi-sensor feature acquisition module, joint motion parameters, human-machine pressure parameters, and muscle state parameters are acquired in real time, and data preprocessing is completed through filtering and noise reduction and outlier removal algorithms.

[0039] Step S21, Multi-sensor layout structure.

[0040] The sensors feature an embedded, hidden design with no exposed protrusions, ensuring comfort during wear. An array of pressure sensors is evenly distributed around the joint's stress-bearing areas, while angle sensors are fixed to the joint's rotation axis for precise measurement of flexion and extension angles. Electromyography (EMG) sensors are integrated with the surrounding muscle groups to collect muscle activation signals. All sensor data is transmitted synchronously to the main control module in real time.

[0041] Step S22: Core parameter acquisition.

[0042] Real-time collected parameters include: human-machine contact pressure P at each point. i Real-time joint flexion and extension angle θ, joint motion angular velocity ω, muscle activation EMG, limb temperature T, and joint displacement S.

[0043] Step S23: Data preprocessing.

[0044] To eliminate noise data caused by motion jitter and environmental interference, a moving average filtering algorithm is used for data preprocessing, as shown below: ; in, This represents the effective data after filtering at time n. The original data collected by the sensor at time i is denoted as N; N is the number of sampling points in the sliding window. In this invention, N=10 is preferred to balance the real-time performance and stability of the data.

[0045] At the same time, an outlier removal threshold is set. When the deviation of a single frame of collected data from the average of the previous 5 frames is greater than 20%, it is judged as abnormal interference data and is directly removed. Valid feature data is retained to provide accurate input for subsequent control algorithms.

[0046] Step S3, Pressure Adaptive Closed-Loop Dynamic Adjustment: Based on the pre-processed accurate data, the target pressure value of each drive unit is calculated through the pressure adaptive control algorithm, and the output pressure of each area is dynamically adjusted to achieve human-machine interaction pressure adaptive matching.

[0047] This invention designs an adaptive PID closed-loop pressure control algorithm, which combines the user's real-time pressure tolerance, joint movement status, and dynamic adjustment of output pressure during the rehabilitation stage to achieve precise adaptive pressure control, solving the problem that traditional fixed pressure control cannot be customized.

[0048] Step S31: Calculate the target pressure benchmark value.

[0049] Based on the user's rehabilitation level, joint type, and injury duration, initialize the baseline target pressure value as follows: ; in, The initial baseline target pressure (kPa); To correspond to the standard reference pressure for joints, the pressure is 30 kPa for the knee joint, 25 kPa for the elbow joint, and 20 kPa for the ankle joint. The damage severity coefficients are: 0.6 for mild injury, 0.8 for moderate injury, and 1.0 for severe injury. The coefficient represents the rehabilitation stage: 0.7 for the initial stage, 0.9 for the middle stage, and 1.0 for the later stage.

[0050] Step S32: Calculation of dynamic pressure correction.

[0051] By combining real-time data on human-machine pressure deviation, joint movement status, and muscle endurance status, dynamic pressure correction is calculated to achieve real-time adaptive adjustment, as shown below: ; in, This is the pressure deviation value; For real-time target pressure; The sensor collects pressure in real time; These are the proportional, integral, and derivative coefficients of the PID controller. After simulation and debugging, the preferred embodiment of this invention is... ; This represents the change in joint angle. This is the angle correction factor, with a value of 0.15 kPa / °; This represents the change in muscle activation. This is the muscle condition correction factor, with a value of 0.2 kPa.

[0052] Step S33: Calculate the final real-time output pressure, as shown below: ; At the same time, a pressure safety threshold is set: the output pressure range of each area is 0-50kPa. When the calculated value exceeds the threshold, the upper and lower limits are automatically locked to avoid limb compression injury caused by excessive pressure and ineffective rehabilitation training due to excessive pressure.

[0053] Step S34: Independent control logic for pressure zones.

[0054] To address the different stress distribution in different areas of the joint, a zoned independent pressure control system is adopted, with each drive unit corresponding to an independent pressure closed-loop control. This achieves differentiated pressure output on the flexion, extension, and support sides of the joint, conforming to the biomechanical characteristics of human joints and significantly improving wearing comfort and rehabilitation accuracy.

[0055] Step S4: Intelligent generation of personalized rehabilitation training plans: Introducing a time-series predictive reasoning model, combining the user's historical rehabilitation data, real-time status data, and injury parameters, iteratively optimizing and generating customized rehabilitation parameters such as training angle, training frequency, stress threshold, and training duration.

[0056] This invention introduces a temporal long short-term memory (LSTM) predictive reasoning model, which predicts joint rehabilitation trends based on users' historical rehabilitation data and real-time status data, and iteratively generates customized rehabilitation training parameters to replace the traditional fixed training mode.

[0057] Step S41: Input the feature set into the model.

[0058] Input features include: historical 7-day average training stress, average joint range of motion, muscle activation, training duration, joint recovery score, and user pain tolerance.

[0059] Step S42: Iterative update rules for rehabilitation parameters.

[0060] The model predicts future rehabilitation adaptation parameters through temporal inference and dynamically updates the training plan accordingly. (1) When the predicted range of motion of the joint continues to improve and the muscle endurance is good, gradually increase the training angle, increase the output pressure, and extend the duration of a single training session; (2) When it is predicted that muscle fatigue will increase, local pressure will exceed the standard, or user pain feedback will exceed the standard, the pressure will be automatically reduced, the range of motion of joints will be reduced, and the training time will be shortened to avoid training overload; (3) The frequency of training should be dynamically adjusted according to the rehabilitation stage: 1-2 times a day in the early stage, 2-3 times a day in the middle stage, and 1 time a day in the later stage for consolidation training.

[0061] Step S43, Human-computer interaction adjustment mechanism.

[0062] Supports both automatic and manual fine-tuning modes: After the model automatically generates a rehabilitation plan, users can manually fine-tune the pressure, angle, and duration parameters through the human-computer interaction terminal. The main control module records the user's preference parameters and iteratively optimizes the subsequent prediction model to achieve fully customized adaptation.

[0063] Step S5, Real-time Rehabilitation Training and Dynamic Correction: Execute rehabilitation training according to the customized training plan, collect feedback data in real time, and dynamically correct stress and exercise parameters.

[0064] Step S6: Multi-dimensional quantitative assessment of rehabilitation effect: Quantitative indicators of joint recovery are calculated through rehabilitation assessment algorithms to determine the user's joint rehabilitation progress and recovery level, and the rehabilitation training plan is updated synchronously.

[0065] This invention constructs a comprehensive assessment model for joint rehabilitation, which accurately judges the joint recovery status through quantitative indicators, replacing the traditional subjective assessment method. The assessment dimensions include four core indicators: joint range of motion, muscle strength recovery, pressure tolerance, and movement stability.

[0066] Step S61: Calculation formula for core evaluation indicators.

[0067] (1) Joint range of motion recovery rate As shown below: ; in, This represents the user's current maximum joint range of motion. The standard range of motion for joints in a healthy human body.

[0068] (2) Muscle strength recovery rate As shown below: ; in, This represents the user's current average muscle activation level. This represents the standard activation level for healthy muscles.

[0069] (3) Pressure tolerance stability coefficient As shown below: ; in, The standard deviation of stress fluctuation during training; The coefficient represents the average training stress. The closer the coefficient is to 1, the better the joint stability under stress and the better the recovery state.

[0070] Step S62: Comprehensive rehabilitation scoring and grading.

[0071] Comprehensive Rehabilitation Score As shown below: ; in, The coefficient represents the motion stability coefficient, with a value ranging from 0 to 1; the weights of each component are determined based on optimized clinical rehabilitation big data.

[0072] Rehabilitation level classification: S≥90 indicates complete recovery, 70≤S<90 indicates good recovery, 50≤S<70 indicates effective recovery, and S<50 indicates poor recovery. The main control module automatically updates the next stage of rehabilitation training plan based on the scoring level.

[0073] Therefore, the present invention adopts the above-mentioned wearable joint rehabilitation trainer and pressure adaptive control method to achieve comfortable wear of the device, personalized and customized adjustment of rehabilitation parameters, and accurate pressure adaptive output, which effectively improves the pertinence, safety and efficiency of joint rehabilitation training, and is suitable for rehabilitation training scenarios of various joint injuries after surgery, sports injuries and degenerative joint diseases.

[0074] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A wearable joint rehabilitation training device, characterized in that, It includes a flexible wearable main body, modular and replaceable structural components, a multi-sensor feature acquisition module, a pressure adaptive drive adjustment module, a main control processing module, a human-computer interaction terminal, and a rehabilitation assessment module; The flexible wearable body is made of medical-grade silicone and highly elastic knitted flexible composite material, with no rigid protrusions, and conforms to the curvature of the human limbs; The modular replaceable structural components are available in three specifications: mild, moderate, and severe damage, to fit human joints and enable replacement across joints and damage levels. The multi-sensor feature acquisition module is used to collect multi-dimensional feature data such as human-machine contact pressure distribution, joint flexion and extension angle, joint movement displacement, muscle activation, and limb surface temperature in real time. The pressure adaptive drive adjustment module adopts an array of miniature pneumatic drive units, with independent control of corresponding sensor zones to achieve differentiated pressure adjustment in single or multiple areas, matching the force requirements of different joint positions; The main control processing module has built-in data preprocessing algorithm, pressure adaptive control algorithm, time series prediction inference model and rehabilitation assessment algorithm, which are electrically connected to the multi-sensor feature acquisition module, pressure adaptive drive adjustment module, human-computer interaction terminal and rehabilitation assessment module respectively, to realize full-process management and control of data processing, intelligent control, plan generation and effect evaluation; The human-computer interaction terminal is used to receive manual adjustment commands from users, display real-time training data, push personalized rehabilitation plans, and display rehabilitation assessment results.

2. The wearable joint rehabilitation trainer according to claim 1, characterized in that, The modular and replaceable structural components include: a joint movement limiting module, a size adaptation and adjustment module, and a pressure zone support module. Each module adopts a snap-on detachable structure, and the thickness, support strength, and movement limiting angle of each module are set differently. They can be freely replaced and combined to fit the user's limb size, type of rehabilitated joint, and degree of injury.

3. The wearable joint rehabilitation trainer according to claim 1, characterized in that, The multi-sensor feature acquisition module includes: a flexible pressure sensor array, an angle sensor, a displacement sensor, an electromyography sensor, and a temperature sensor, which are distributed and integrated on the inner side of the flexible wearable body and the joint movement area; The flexible pressure sensor array adopts a high-density distributed layout, with 8-12 independent pressure acquisition points set in a single joint area. The sampling frequency is greater than or equal to 100Hz, and the pressure acquisition accuracy is ±0.5kPa, capturing minute local pressure changes in the joint.

4. A pressure adaptive control method for a wearable joint rehabilitation trainer, used to execute the wearable joint rehabilitation trainer according to any one of claims 1-3, characterized in that, Includes the following steps: Step S1, Personalized Wearing and Adaptation of the Device: Based on the user's rehabilitation joint type, limb size, and injury level, replace the modular structural components to complete the wearing and fixation, and achieve a close fit between the device and the limb; Step S2, Accurate Acquisition and Preprocessing of Multi-Dimensional Feature Data: Through the multi-sensor feature acquisition module, joint motion parameters, human-machine pressure parameters, and muscle state parameters are acquired in real time, and data preprocessing is completed through filtering and noise reduction and outlier removal algorithms; Step S3, Pressure Adaptive Closed-Loop Dynamic Adjustment: Based on the preprocessed data, the target pressure value of each drive unit is calculated through the pressure adaptive control algorithm, and the output pressure of each area is dynamically adjusted to achieve human-machine interaction pressure adaptive matching. Step S4: Intelligent generation of personalized rehabilitation training plans: Introducing a time-series predictive reasoning model, combining the user's historical rehabilitation data, real-time status data, and injury parameters, iteratively optimizing and generating customized rehabilitation parameters such as training angle, training frequency, stress threshold, and training duration; Step S5, Real-time Rehabilitation Training and Dynamic Correction: Execute rehabilitation training according to the customized training plan, collect feedback data in real time, and dynamically correct stress and exercise parameters; Step S6: Multi-dimensional quantitative assessment of rehabilitation effect: Quantitative indicators of joint recovery are calculated through rehabilitation assessment algorithms to determine the user's joint rehabilitation progress and recovery level, and the rehabilitation training plan is updated synchronously.

5. The wearable joint rehabilitation trainer pressure adaptive control method according to claim 4, characterized in that, Step S1: Personalized device wearable adaptation. The specific process is as follows: Step S11: The main body of the equipment adopts a double-layer composite flexible structure. The inner layer is a breathable and antibacterial flexible knitted layer, and the outer layer is a high-toughness elastic silicone support layer. The overall structure has no hard edges and corners, and deforms synchronously with the flexion, extension and rotation of the joints, conforming to the movement law of the human limbs. Step S12: The modular components adopt standardized snap-fit ​​interfaces, and all module interfaces are unified, supporting free disassembly, replacement, and combination. (1) Size adaptation module: Sets four sizes: S, M, L and XL to adapt to users with different limb thicknesses; (2) Joint movement limiting module: Different maximum flexion and extension limiting angles are set for different joints: elbow joint 0°-135°, knee joint 0°-120°, ankle joint 0°-45°. The limiting module can be manually switched according to the user's injury recovery stage. (3) Pressure zone support module: It is divided into three types: uniform support, local pressure, and decompression buffer. It can achieve pressure or decompression and targeted rehabilitation for joint injury pain points and weak stress areas.

6. The wearable joint rehabilitation trainer pressure adaptive control method according to claim 4, characterized in that, Step S2 involves accurate acquisition and preprocessing of multi-dimensional feature data, the specific process of which is as follows: Step S21: The sensor adopts an embedded hidden layout. The pressure sensor array is evenly distributed in the force area around the joint, and the angle sensor is fixed at the joint rotation axis to collect the joint flexion and extension angles. The electromyography (EMG) sensor is attached to the muscle group around the joint to collect muscle activation signals, and all sensor data is transmitted to the main control module in real time. Step S22: Collect real-time parameters, including: human-machine contact pressure P at each point. i Real-time joint flexion and extension angle θ, joint motion angular velocity ω, muscle activation EMG, limb temperature T, joint displacement S; Step S23: To eliminate noise data caused by motion jitter and environmental interference, a moving average filtering algorithm is used to complete data preprocessing, as shown below: ； in, This represents the effective data after filtering at time n. The original data collected by the sensor at time i; N is the number of sampling points in the sliding window; At the same time, an outlier removal threshold is set. When the deviation of a single frame of collected data from the average of the previous 5 frames is greater than 20%, it is judged as abnormal interference data and is directly removed.

7. The wearable joint rehabilitation trainer pressure adaptive control method according to claim 4, characterized in that, In step S3, an adaptive PID closed-loop pressure control algorithm is designed. This algorithm combines the user's real-time pressure tolerance, joint movement status, and dynamic adjustment of output pressure during the rehabilitation phase to achieve adaptive pressure control. The specific process is as follows: Step S31: Initialize the baseline target pressure value based on the user's rehabilitation level, joint type, and injury duration, as shown below: ； in, The initial baseline target pressure is given in kPa. The corresponding standard reference pressure for the joint is in kPa; Damage level coefficient; For the rehabilitation stage coefficient; Step S32: Combining real-time collected human-machine pressure deviation, joint movement status, and muscle endurance status, calculate the dynamic pressure correction amount to achieve real-time adaptive adjustment, as shown below: ； in, This is the pressure deviation value; For real-time target pressure; The sensor collects pressure in real time; These are the PID proportional, integral, and derivative coefficients, respectively. This represents the change in joint angle. The angle correction factor is kPa / °. This represents the change in muscle activation. Muscle condition correction factor, kPa; Step S33: Calculate the final real-time output pressure, as shown below: ； At the same time, a pressure safety threshold is set: the output pressure range of each area is 0-50kPa. When the calculated value exceeds the threshold, the upper and lower limits are automatically locked to avoid limb compression injury caused by excessive pressure and ineffective rehabilitation training due to excessive pressure. Step S34: To address the differences in force distribution across different areas of the joint, a zoned independent pressure control is adopted, with each drive unit corresponding to an independent pressure closed-loop control, thereby achieving differentiated pressure output on the flexion, extension, and support sides of the joint, which conforms to the biomechanical characteristics of human joints.

8. The wearable joint rehabilitation trainer pressure adaptive control method according to claim 4, characterized in that, In step S4, a temporal long short-term memory (LSTM) predictive inference model is introduced. Based on the user's historical rehabilitation data and real-time status data, the joint rehabilitation trend is predicted, and customized rehabilitation training parameters are iteratively generated. The specific process is as follows: Step S41: The model feature inputs include: historical 7-day average training stress, average joint range of motion, muscle activation, training duration, joint recovery score, and user pain tolerance. Step S42: The model predicts future rehabilitation adaptation parameters through temporal inference and dynamically updates the training plan; Step S43: After the model automatically generates a rehabilitation plan, based on the human-computer interaction adjustment mechanism, the user can manually fine-tune the pressure, angle, and duration parameters through the human-computer interaction terminal. The main control module records the user's preference parameters and iteratively optimizes the subsequent prediction model to achieve fully customized adaptation.

9. The wearable joint rehabilitation training device pressure adaptive control method according to claim 4, characterized in that, In step S6, a comprehensive assessment model for joint rehabilitation is constructed. The model uses quantitative indicators to determine the extent of joint recovery. The assessment dimensions include four core indicators: joint range of motion, muscle strength recovery, stress tolerance, and movement stability. The specific process is as follows: Step S61: Calculation formula for core evaluation indicators; (1) Joint range of motion recovery rate As shown below: ； in, This represents the user's current maximum joint range of motion. The standard range of motion for joints in a healthy human body; (2) Muscle strength recovery rate As shown below: ； in, This represents the user's current average muscle activation level. Standard activation level for healthy muscles; (3) Pressure tolerance stability coefficient As shown below: ； in, The standard deviation of stress fluctuation during training; To average training pressure; Step S62: Comprehensive rehabilitation scoring and grading; Comprehensive Rehabilitation Score As shown below: ； in, The coefficient represents the motion stability, ranging from 0 to 1; the weights of each component are determined and optimized based on big data from clinical rehabilitation. Rehabilitation level classification: S≥90 indicates complete recovery, 70≤S<90 indicates good recovery, 50≤S<70 indicates effective recovery, and S<50 indicates poor recovery. The main control module automatically updates the next stage of rehabilitation training plan based on the scoring level.

Images