Intelligent rehabilitation training scheme generation system and method and medium
By establishing a multi-dimensional labeled digital training video library and intelligent assessment, combined with multi-source data fusion modeling and real-time movement correction, personalized training programs are generated, solving the problems of insufficient movement guidance and weak risk control in existing rehabilitation training systems, and achieving efficient and accurate rehabilitation training results.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-29
- Publication Date
- 2026-03-27
AI Technical Summary
Existing rehabilitation training systems suffer from a lack of standardized movement guidance, insufficient personalized adaptation, and weak risk control, resulting in poor rehabilitation outcomes and difficulty in ensuring patient safety.
By establishing a digital training video library with multi-dimensional labels, combined with intelligent evaluation, multi-source data fusion, and real-time motion correction, personalized training programs are generated. Through multi-source data fusion modeling and dynamic intelligent optimization, real-time adjustments are made to achieve precise motion correction and risk control.
It significantly improves the safety, accuracy, and adaptability of rehabilitation training, provides efficient and personalized rehabilitation services, and ensures patient compliance and safety.
Smart Images
Figure CN121747832A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of intelligent medical treatment, in particular to an intelligent rehabilitation training scheme generation system, method and medium. BACKGROUND
[0002] With the aggravation of population aging and the surge in demand for rehabilitation treatment, remote rehabilitation has become a new health management method.
[0003] At present, the existing rehabilitation training system generally has problems such as shortage of medical resources and insufficient individual adaptation. Although the existing remote rehabilitation system partially solves the geographical restriction, it still has significant deficiencies: first, there is a lack of standardized action guidance: without professional video reference for patient training, it is easy to cause deformation or compensation of action, affecting the rehabilitation effect. Second, the individual adaptation is insufficient, and it is impossible to accurately adapt the rehabilitation training scheme and dynamically adjust it. Third, the risk control mechanism is weak, and it is impossible to identify hidden dangers such as muscle compensation and imbalance and fall in real time, and the patient compliance and safety are difficult to guarantee.
[0004] In addition, in the face of training data, the rehabilitation therapist lacks intelligent analysis and decision support tools. SUMMARY
[0005] In view of the defects in the prior art, the purpose of the present application is to provide an intelligent rehabilitation training scheme generation system, method and medium, which significantly improves the safety, accuracy and adaptability of rehabilitation training through the cooperation of intelligent matching of training videos, multi-source data fusion modeling, real-time action correction, dynamic intelligent optimization and efficient remote intervention.
[0006] In a first aspect, the present application provides an intelligent rehabilitation training scheme generation method, comprising the following steps: Step S100, establishing a digital training video library: establishing a digital training video library, and adding multi-dimensional tags to each digital training video; Step S200, intelligent evaluation: obtaining the rehabilitation scene type selected by the user and the clinical data uploaded, evaluating to generate a user ability baseline report, and automatically associating the digital training video tags adapted according to the user ability baseline report; Step S300, scheme customization: based on the automatically associated and adapted digital training video tags, automatically screening the matched digital training videos from the digital training video library; the rehabilitation therapist manually adjusts the combination of digital training videos on the basis of intelligent recommendation, and generates an individual training scheme to send to the user; Step S400, training data acquisition: the user acquires training data through a terminal device, and the terminal device includes a camera device, a wearable sensor group and an environmental Internet of Things device; Step S500, multi-source data fusion: synchronously acquiring the data stream of the training data in real time, and constructing a user digital model; Step S600, regimen optimization: automatically adjust the training regimen based on real-time feedback of the data stream of the collected training data through real-time dynamic optimization; predict the next stage of ability improvement curve by analyzing historical training data, and automatically recommend video combination scheme and optimal training duration and frequency; Step S700, risk control: real-time feedback output through hierarchical feedback strategy; Step S800, decision support: generate a visual rehabilitation decision panel by integrating data; the therapist adjusts the training video combination by dragging and real-time predicts the effect after adjustment, and adjusts the training regimen.
[0007] In a preferred embodiment of the present application, the multi-dimensional label includes but is not limited to indication type, instrument requirement, target body part, balance level and muscle strength rating.
[0008] In a preferred embodiment of the present application, the intelligent assessment of step S200 further includes pushing a differentiated assessment questionnaire according to the user-selected rehabilitation scene type and uploaded clinical data.
[0009] In a preferred embodiment of the present application, in the training data collection of step S400, the training action video stream is captured through the camera of the user terminal device; the angular velocity of the limb joint is collected through the IMU binding belt included in the wearable sensor group, the center of gravity distribution is monitored through the pressure sensing insole, and the muscle activation is detected through the electromyography bracelet; the inclination angle data is fed back through the intelligent rehabilitation bed included in the environmental Internet of Things device, the resistance value is output through the electric balance board, and the body posture inclination is monitored through the safety protection radar.
[0010] In a preferred embodiment of the present application, in the multi-source data fusion of step S500, the action stream of the camera and the IMU joint data are fused to generate a 3D skeletal motion trajectory; the pressure sensing insole data is mapped to a virtual center of gravity projection model to quantify the balance stability; the electromyography signal is associated with the rehabilitation bed angle data to calculate the muscle activation efficiency.
[0011] In a preferred embodiment of the present application, in the risk control of step S700, when the safety protection radar detects slight imbalance, it real-time voice prompts "Please tighten the core muscle group"; when it detects 3 consecutive action errors, it pauses the training and plays the error correction video; when the electromyography bracelet recognizes muscle compensation and heart rate rises sharply, it immediately terminates the training and notifies the therapist.
[0012] In a preferred embodiment of the present application, in the risk control of step S700, when the safety protection radar detects slight imbalance, it real-time voice prompts "Please tighten the core muscle group"; when it detects 3 consecutive action errors, it pauses the training and plays the error correction video; when the electromyography bracelet recognizes muscle compensation and heart rate rises sharply, it immediately terminates the training and notifies the therapist. The digital training video library module is used for storing digital training videos and adding multi-dimensional labels to each digital training video. An intelligent evaluation module is configured to acquire a rehabilitation scene type selected by a user and uploaded clinical data, generate a user ability baseline report, and automatically associate an adaptive digital training video tag based on the user ability baseline report; A scheme customization module is configured to automatically filter a matched digital training video from a digital training video library based on the intelligent evaluation module automatically associating an adaptive digital training video tag, manually adjust a digital training video combination by a rehabilitation therapist based on intelligent recommendation, and generate a personalized training scheme and send the personalized training scheme to a user terminal device. A training data acquisition module includes a user terminal device integrated with a camera device, a wearable sensor group including an inertial measurement unit (IMU) joint strap, a pressure sensor insole, and an electromyography bracelet, and an environmental Internet of Things device including an intelligent rehabilitation bed, an electric balance board, and a safety protection radar. A multi-source data fusion module is configured to synchronize data streams from the training data acquisition module in real time and construct a user digital model. A scheme optimization module includes a real-time dynamic optimization unit configured to automatically adjust a training scheme based on real-time feedback of the multi-source data fusion module, and a long-term adaptability planning unit configured to analyze historical training data, predict a next-stage ability improvement curve, and automatically recommend a video combination scheme and an optimal training duration and frequency. A risk control module is configured to perform real-time feedback output through a hierarchical feedback strategy. A decision support module includes an intelligent report generation unit configured to integrate data to generate a visual rehabilitation decision board, a scheme simulation unit configured to allow a rehabilitation therapist to drag and adjust a training video combination, and a remote intervention unit configured to allow the rehabilitation therapist to adjust a training scheme.
[0013] In a preferred embodiment of the present application, the multi-dimensional tag includes an indication type, an instrument requirement, a target body part, a balance level, and a muscle strength rating.
[0014] In a preferred embodiment of the present application, the digital training video is classified into four stages according to PT physical therapy, including a lying position training video, a sitting position training video, a standing position training video, and a walking position training video.
[0015] In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which is executed by a processor to implement the intelligent rehabilitation training scheme generation method.
[0016] Compared with the prior art, the present application has the following advantages: (1) Multi-dimensional video intelligent matching, structured video library through multi-dimensional tags (muscle strength rating, balance rating, etc.), combined with user ability baseline report automatically associated with adaptive training video generated by intelligent evaluation, improve the scientific nature and precision of individualization of the scheme; (2) Real-time action correction, through multi-source data fusion modeling, fusion of training data collected by cameras, wearable sensor groups and environmental Internet of Things devices, build user digital model, realize real-time accurate correction of action; (3) Dynamic intelligent optimization, based on real-time feedback of multi-source data fusion, intelligent adjustment of training scheme, and analysis of historical training data, prediction and recommendation of video training scheme; (4) Dynamic remote intervention mechanism, the rehabilitation therapist can manage multiple patients at the same time, remotely and efficiently adjust the scheme, view data parameters and generate rehabilitation reports, providing efficient, accurate and personalized rehabilitation services for patients.
[0017] The features of the present application can be clearly understood by referring to the drawings and the detailed description of the preferred embodiments below. BRIEF DESCRIPTION OF DRAWINGS
[0018] Figure 1 The flowchart of the intelligent rehabilitation training scheme generation method of the present application is shown in the figure. Figure 2 The structure diagram of the intelligent rehabilitation training scheme generation system of the present application is shown in the figure. DETAILED DESCRIPTION
[0019] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme of the embodiments of the present application will be described clearly and completely below in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all the embodiments. The components of the embodiments of the present application described and shown in the drawings can be arranged and designed in various different configurations.
[0020] Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor are within the scope of protection of the present application.
[0021] It should be noted that similar reference numerals and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, all directional indications (such as up, down, left, right, front, back, bottom, etc.) in this application are only used to explain the relative positional relationships and movements between components in a specific orientation (as shown in the figures). If the specific orientation changes, the directional indication will also change accordingly. Furthermore, descriptions involving "first," "second," etc., in this application are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated.
[0022] like Figure 1 As shown, the intelligent rehabilitation training program generation method of the present invention includes the following steps: Step S100: Establish a digital training video library: Establish a digital training video library and add multi-dimensional tags to each digital training video, such as indication type, equipment requirements, target body part, balance level, muscle strength rating, etc.; classify the digital training videos according to the four stages of PT physical therapy, which include supine training videos, sitting training videos, standing training videos, and walking training videos; supine training videos include but are not limited to joint mobility training videos and turning training videos; sitting training videos include but are not limited to balance training videos and transfer training videos; standing training videos include but are not limited to weight transfer videos and resistance training videos; walking training videos include but are not limited to gait correction videos and step training videos. Step S200, Intelligent Assessment: Obtain the rehabilitation scenario type selected by the user and the uploaded clinical data, assess and generate a user ability baseline report (including muscle strength rating, balance level, etc.), and automatically associate appropriate digital training video tags (such as muscle strength grade 3, weak balance, etc.) based on the user ability baseline report; the rehabilitation scenario type includes one of the following: inpatient scenario, home scenario, community scenario, or elderly care institution scenario; the clinical data includes the user's basic information, medical record information, diagnostic report, etc. The intelligent assessment in step S200 also includes pushing differentiated assessment questionnaires based on the type of rehabilitation scenario selected by the user and the uploaded clinical data. Step S300, Solution Customization: Based on the automatically associated and adapted digital training video tags in the intelligent assessment in step S200, matching digital training videos are automatically selected from the digital training video library; the rehabilitation therapist manually adjusts the combination of digital training videos based on the intelligent recommendation to generate a personalized training plan and send it to the user terminal; the solution customization step may include setting the duration of a single training session, the number of repetitions of the movements, etc. Step S400, training data collection: capture training action video stream through the camera of the user terminal device 410; collect the angular velocity of the limb joint through the IMU strap included in the wearable sensor group 420, monitor the gravity distribution through the pressure sensing insole, and detect the muscle activation through the electromyography bracelet; feedback the inclination angle data through the smart rehabilitation bed included in the environmental Internet of Things device 430, output the resistance value through the electric balance board, and monitor the body posture inclination through the safety protection radar; Step S500, multi-source data fusion: real-time synchronization of the data stream from step S400 training data collection, and construction of a user digital model; specifically including: camera action stream and IMU joint data fusion, generation of 3D skeletal motion trajectory; pressure sensing insole data mapping to a virtual gravity projection model, quantification of balance stability; electromyography signal and rehabilitation bed angle data correlation, calculation of muscle activation efficiency; Step S600, scheme optimization: through the real-time dynamic optimization unit 510, based on the real-time feedback in step S500 multi-source data fusion, automatically adjust the training scheme; for example, when the joint activity degree standard rate is detected to be greater than or equal to 90%, automatically upgrade the resistance intensity (such as balance board resistance + 15%); if the electromyography activation is continuously lower than the preset threshold, replace it with a low difficulty video (such as reducing the standing resistance training to a seated resistance); also analyze the historical training data through the long-term adaptive planning unit 620, predict the next stage ability improvement curve, and automatically recommend the video combination scheme and the optimal training time length, frequency, etc.
[0023] Step S700, risk control: real-time feedback output through a hierarchical feedback strategy; when the safety protection radar detects slight imbalance, real-time voice prompt “Please tighten the core muscle group”; when 3 consecutive action errors are detected, pause the training and play the error correction video; when the electromyography bracelet recognizes muscle compensation and the heart rate suddenly rises, immediately terminate the training and notify the rehabilitation therapist; Step S800, decision support: generate a visual rehabilitation decision board through the intelligent report generation unit 810; the rehabilitation therapist adjusts the training video combination through the scheme simulation unit 820, the system real-time predicts the effect after adjustment, and adjusts the training scheme through the remote intervention unit 830.
[0024] As shown in Figure 2 The intelligent rehabilitation training scheme generation system of the present application comprises: The digital training video library module 100 is used to store digital training videos and add multi-dimensional tags to each video, such as indication type, equipment requirements, target body part, balance level, and muscle strength rating. The digital training videos are classified according to the four stages of PT physical therapy: supine training videos include but are not limited to joint mobility training videos and turning training videos; seated training videos include but are not limited to balance training videos and transfer training videos; standing training videos include but are not limited to weight transfer videos and resistance training videos; and walking training videos include but are not limited to gait correction videos and step training videos. The intelligent assessment module 200 is used to acquire the rehabilitation scenario type selected by the user and the uploaded clinical data, assess and generate a user ability baseline report (including muscle strength rating, balance level, etc.), and automatically associate appropriate digital training video tags (such as muscle strength grade 3, weak balance, etc.) based on the user ability baseline report; the rehabilitation scenario type includes one of the following: inpatient scenario, home scenario, community scenario, or elderly care institution scenario; the clinical data includes the user's basic information, medical record information, diagnostic report, etc. The solution customization module 300 is used to automatically associate and adapt digital training video tags based on the intelligent assessment module 200, automatically filter matching digital training videos from the digital training video library, and allow rehabilitation therapists to manually adjust the combination of digital training videos based on intelligent recommendations to generate personalized training plans and send them to the user terminal device 410; the solution customization module 300 supports setting the duration of a single training session, the number of repetitions of the movements, etc. The training data acquisition module 400 includes a user terminal device 410, which is a mobile phone, tablet computer, or smart TV with an integrated camera, used for video training interaction; a wearable sensor group 420, which includes an inertial measurement unit (IMU), joint straps, pressure-sensing insoles, electromyography (EMG) bracelets, etc.; and an environmental IoT device 430, which includes a smart rehabilitation bed (adjustable angle), an electric balance board (controllable resistance), and a safety radar (monitoring fall risk), etc. The multi-source data fusion module 500 is used to synchronize data streams from user terminal device 410, wearable sensor group 420, and environmental IoT device 430 in real time to construct a user digital model. Specifically, this includes fusing camera motion streams and IMU joint data to generate 3D skeletal motion trajectories; mapping pressure-sensing insole data to a virtual center of gravity projection model to quantify balance stability; and correlating electromyographic signals with rehabilitation bed angle data to calculate muscle activation efficiency. The scheme optimization module 600 comprises a real-time dynamic optimization unit 610, which automatically adjusts the training scheme based on the real-time feedback of the multi-source data fusion module 500; for example, when it is detected that the joint activity reaches a standard rate of 90%, the resistance intensity is automatically upgraded (for example, the balance plate resistance is increased by 15%); if the muscle activation continues to be lower than the preset threshold, a low-difficulty video is replaced (for example, standing resistance training is reduced to seated resistance); and the long-term adaptive planning unit 620 is further included, which is used for analyzing historical training data, predicting the next-stage ability improvement curve, and automatically recommending a video combination scheme and optimal training duration, frequency and the like.
[0025] The risk control module 700 is used for real-time feedback output through a hierarchical feedback strategy; when the safety protection radar detects slight imbalance, real-time voice prompts “Please tighten the core muscle group”; when it is detected that the action is continuously wrong for three times, the training is paused and a correction video is played; when the electromyographic bracelet recognizes muscle compensation and the heart rate suddenly rises, the training is immediately terminated and the physiotherapist is notified; The decision support module 800 comprises an intelligent report generation unit 810 for generating a visual rehabilitation decision board by integrating data; a scheme simulation unit 820 for enabling the physiotherapist to drag and adjust the training video combination, and the system real-time predicts the effect after adjustment; and a remote intervention unit 830 for enabling the physiotherapist to adjust the training scheme. In the embodiment, the intelligent evaluation module 200 further comprises a questionnaire evaluation unit, which is used for pushing a differentiated evaluation questionnaire according to the rehabilitation scene type selected by the user and the uploaded clinical data.
[0026] The computer readable storage medium of the present application has a computer program stored thereon, and the program is executed by a processor to implement the intelligent rehabilitation training scheme generation method of the present application.
[0027] The basic principles and main features of the present application and the advantages of the present application are shown and described. It should be understood by those skilled in the art that the present application is not limited by the above embodiments, and the above embodiments and descriptions in the specification are only to illustrate the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the present application. The scope of protection of the present application is defined by the appended claims and their equivalents.
Claims
1. A method for generating an intelligent rehabilitation training program, characterized in that, Includes the following steps: Step S100: Establish a digital training video library: Establish a digital training video library and add multidimensional tags to each digital training video. Step S200, Intelligent Assessment: Obtain the rehabilitation scenario type selected by the user and the uploaded clinical data, assess and generate a user ability baseline report, and automatically associate the appropriate digital training video tags based on the user ability baseline report; Step S300, Solution Customization: Based on automatically associated and adapted digital training video tags, matching digital training videos are automatically selected from the digital training video library; the rehabilitation therapist manually adjusts the combination of digital training videos based on the intelligent recommendation to generate a personalized training plan and send it to the user; Step S400, Training Data Acquisition: The user collects training data through a terminal device, which includes a camera device, a wearable sensor group, and an environmental IoT device; Step S500, Multi-source data fusion: Real-time synchronization of the collected training data stream to build the user's digital model; Step S600, Scheme Optimization: Through real-time dynamic optimization, the training scheme is automatically adjusted based on the real-time feedback of the data stream of the collected training data; by analyzing historical training data, the capability improvement curve of the next stage is predicted, and video combination schemes and optimal training duration and frequency are automatically recommended; Step S700, Risk Control: Real-time feedback output is achieved through a tiered feedback strategy; Step S800, Decision Support: A visual rehabilitation decision panel is generated by integrating data; rehabilitation therapists can adjust the training video combination by dragging and dropping and predict the effect after adjustment in real time, and adjust the training plan accordingly.
2. The intelligent rehabilitation training program generation method according to claim 1, characterized in that, The multidimensional labels include, but are not limited to, indication type, device requirements, target body part, balance level, and muscle strength rating.
3. The method for generating an intelligent rehabilitation training program according to claim 1 or 2, characterized in that, The intelligent assessment in step S200 also includes pushing differentiated assessment questionnaires based on the rehabilitation scenario type selected by the user and the uploaded clinical data.
4. The method for generating an intelligent rehabilitation training program according to claim 1 or 2, characterized in that, The training data acquisition in step S400 includes capturing training motion video streams through the camera device of the user terminal device; collecting angular velocities of limb joints through the IMU strap included in the wearable sensor group, monitoring center of gravity distribution through pressure-sensing insoles, and detecting muscle activation through electromyography bracelets; and providing tilt angle data through the smart rehabilitation bed included in the environmental IoT device, outputting resistance values through the electric balance board, and monitoring body tilt angle through safety radar.
5. The intelligent rehabilitation training program generation method according to claim 4, characterized in that, The multi-source data fusion in step S500 includes the fusion of the motion flow of the camera device and the joint data of the IMU to generate a 3D skeleton motion trajectory. Pressure-sensing insole data is mapped to a virtual center-of-gravity projection model to quantify balance stability; Electromyographic signals are correlated with the angle data of the rehabilitation bed to calculate muscle activation efficiency.
6. The intelligent rehabilitation training program generation method according to claim 4, characterized in that, The risk control in step S700 includes real-time voice prompts such as "Please tighten your core muscles" when the safety radar detects a slight imbalance; pausing training and playing a correction video when three consecutive incorrect movement matches are detected; and immediately terminating training and notifying the rehabilitation therapist when the electromyography bracelet recognizes muscle compensation and a sudden increase in heart rate.
7. An intelligent rehabilitation training program generation system, characterized in that, include: The digital training video library module is used to store digital training videos and add multidimensional tags to each digital training video. The intelligent assessment module is used to obtain the rehabilitation scenario type selected by the user and the uploaded clinical data, assess and generate a user ability baseline report, and automatically associate and adapt digital training video tags based on the user ability baseline report. The solution customization module is used to automatically associate and adapt digital training video tags based on the intelligent assessment module, automatically filter matching digital training videos from the digital training video library, and allow rehabilitation therapists to manually adjust the combination of digital training videos based on intelligent recommendations to generate personalized training plans and send them to user terminal devices. The training data acquisition module includes a user terminal device with an integrated camera for video training interaction; a wearable sensor group including an inertial measurement unit (IMU), a joint strap, a pressure-sensing insole, and an electromyography (EMG) bracelet; and environmental IoT devices including a smart rehabilitation bed, an electric balance board, and a safety radar. The multi-source data fusion module is used to synchronize data streams from the training data acquisition module in real time to build user digital models; The scheme optimization module includes a real-time dynamic optimization unit, which automatically adjusts the training scheme based on real-time feedback from the multi-source data fusion module. The long-term adaptive planning unit is used to analyze historical training data, predict the capability improvement curve for the next stage, and automatically recommend video combination schemes and optimal training duration and frequency. The risk control module is used to provide real-time feedback output through a tiered feedback strategy. The decision support module includes an intelligent report generation unit, which integrates data to generate a visual rehabilitation decision dashboard; a program simulation unit, which allows rehabilitation therapists to drag and adjust training video combinations, and the system predicts the effects after adjustment in real time; and a remote intervention unit, which allows rehabilitation therapists to adjust training programs.
8. The intelligent rehabilitation training program generation system according to claim 7, characterized in that, The multidimensional labels include indication type, device requirements, target body part, balance level, and muscle strength rating.
9. The intelligent rehabilitation training program generation system according to claim 7 or 8, characterized in that, The digital training videos are categorized into four stages of physical therapy (PT): training videos for the supine, sitting, standing, and walking phases.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by the processor to implement the intelligent rehabilitation training program generation method as described in claim 1.