Rehabilitation evaluation system for central nervous system dysfunction patient

Abstract

A rehabilitation evaluation system for a patient with central nervous system dysfunction comprises the following steps of S1, collecting videos, motion data and pressure data of a user through a data collection layer, S2, uploading the video data to an edge computing server, performing preliminary posture estimation, and S3, performing rehabilitation evaluation. The method comprises the following steps: S1, carrying out edge calculation on the data, S3, transmitting the data subjected to edge calculation to a cloud server for deep modeling analysis of gait and balance parameters, S4, carrying out abnormal mode identification and comparison after parameter analysis, S5, uploading the analyzed data after comparison to a decision application layer, S6, transmitting each parameter and a corresponding processing scheme to a doctor end for a doctor to check, and S5, carrying out data processing. The method has the advantages that through full-link innovation of multi-mode lightweight acquisition, edge cloud hierarchical processing, double-model intelligent diagnosis, reinforcement learning dynamic optimization and visual remote decision, central nervous rehabilitation evaluation extends to a daily life scene from a laboratory, and core breakthrough of precision improvement, efficiency multiplication and personalized enhancement is achieved.

Description

Technical Field

[0001] This invention relates to the field of rehabilitation medicine technology, specifically a rehabilitation assessment system for patients with central nervous system dysfunction. Background Technology

[0002] The rehabilitation assessment and treatment of patients with central nervous system dysfunction rely on the accurate monitoring and analysis of gait, balance ability and abnormal posture. Traditional methods mainly rely on visual assessment by doctors, laboratory-grade motion capture systems (such as infrared optical devices) or single sensors (such as accelerometers) to collect data.

[0003] However, existing data acquisition and analysis technologies still have the following drawbacks:

[0004] Current technologies typically involve assessments conducted in specialized medical institutions, which are subject to time and space limitations. Data analysis often relies on basic algorithms, simple threshold judgments, or linear models, making it difficult to integrate multi-dimensional and heterogeneous data such as video, physiological parameters, and joint angles. This results in insufficient personalization of rehabilitation plans, weak real-time feedback, and limited remote monitoring capabilities. Summary of the Invention

[0005] The technical problem to be solved by this invention is to overcome the above-mentioned technical defects and provide a rehabilitation assessment system for patients with central nervous system dysfunction.

[0006] To address the above problems, the technical solution of this invention is: a rehabilitation assessment system for patients with central nervous system dysfunction, comprising the following steps:

[0007] S1. The data acquisition layer collects video, motion, and stress data from users to provide intuitive and realistic audio-visual materials for rehabilitation assessment and treatment.

[0008] S2. Upload the video data to the edge computing server for preliminary pose estimation, reduce data transmission pressure, improve processing speed, and provide basic data support for deep modeling.

[0009] S3. The data processed by edge computing is transmitted to the cloud server for in-depth modeling and analysis of gait and balance parameters;

[0010] S4. After analyzing the parameters, perform anomaly pattern identification and comparison;

[0011] S5. Upload the compared and analyzed data to the decision application layer, realize clinical rule reasoning through decision-making, match the corresponding treatment plan, and provide visualized data for doctors' observation and research.

[0012] S6. Transmit each parameter and corresponding processing plan to the doctor's end for the doctor to view.

[0013] Furthermore, in step S1, the data acquisition layer adopts multi-mimicry data input, records daily activity videos through a mobile phone camera to facilitate remote detection and follow-up, analyzes gait symmetry by capturing the full gait from the front, provides high-precision data on physiological indicators such as blood pressure and heart rate, emotional and psychological characteristics, and physical activity indicators through wearable sensors, records joint angle changes and acceleration data in real time, and evaluates foot support distribution, gait stability, and load conditions through analysis by pressure sensors inside the shoe sole.

[0014] Furthermore, in step S2, the edge computing server preprocesses the video using the OpenPose algorithm to extract motion parameters. The motion parameter extraction uses spatiotemporal parameters to analyze and extract step length, step width, and support phase duration, uses angle parameters to analyze and extract knee flexion and extension angles and pelvic tilt, and uses a symmetry index to compare the differences in joint activity between the left and right sides.

[0015] Furthermore, in step S3, the cloud analysis layer extracts motion parameters and performs spatiotemporal alignment, captures joint movement trajectories through video, and simultaneously synchronizes sensor data to accurately correspond multi-dimensional information, achieving multi-dimensional information synchronization. During the spatiotemporal alignment, feature integration is performed. The feature integration is carried out simultaneously by constructing a fusion network using PyTorch, integrating visual features, and integrating sensor features to enhance information complementarity.

[0016] Furthermore, in step S4, abnormal pattern recognition is divided into a classification model and a clustering model. The classification model uses the SVM algorithm to classify small sample high-dimensional data, accurately distinguishing abnormal postures and guiding personalized rehabilitation strategies. The K-means algorithm is used for data mining, which automatically divides rehabilitation stages and sets stage-based rehabilitation goals. Through structured rules, efficient and interpretable posture analysis and decision-making are achieved, which is widely used in rehabilitation assessment, fall warning, and personalized treatment plan formulation.

[0017] Furthermore, in step S5, the decision application layer includes a decision tree rule engine and reinforcement learning optimization. The decision tree rule engine has rehabilitation assessment, fall warning, and treatment plan generation functions. The reinforcement learning optimization includes an action space and a reward function. The action space is trained for VR duration by selecting orthotic types, providing diverse training strategies for the intelligent training system. In reinforcement learning training, the reward function quantifies the feedback mechanism of action quality to guide rehabilitation training or fall prevention.

[0018] Furthermore, in step S6, after the data is transmitted to the doctor's end, the doctor reviews it and decides to implement a specific plan.

[0019] The advantages of this invention compared to existing technologies are:

[0020] 1. This invention solves the four major pain points of existing technologies—data bias, inefficient analysis, static decision-making, and scenario limitations—through a full-chain innovation of multimodal lightweight acquisition, edge cloud layered processing, dual-model intelligent diagnosis, reinforcement learning dynamic optimization, and visualized remote decision-making. It extends central nervous system rehabilitation assessment from the laboratory to daily life scenarios, achieving core breakthroughs in improving accuracy, doubling efficiency, and enhancing personalization. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the system method flow of the present invention.

[0022] Figure 2 This is a flowchart of the data acquisition layer of the present invention.

[0023] Figure 3 This is a schematic diagram of the edge computing server process of the present invention.

[0024] Figure 4 This is a schematic diagram of the cloud analysis layer process of the present invention.

[0025] Figure 5 This is a schematic diagram of the abnormal pattern recognition process of the present invention.

[0026] Figure 6 This is a schematic diagram of the decision application layer process of the present invention. Detailed Implementation

[0027] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0028] To make the content of this invention easier to understand, the technical solutions in the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings.

[0029] like Figure 1 As shown, the technical solution of the present invention is: a rehabilitation assessment system for patients with central nervous system dysfunction, comprising the following steps:

[0030] S1. The data acquisition layer collects video, motion, and stress data from users to provide intuitive and realistic audio-visual materials for rehabilitation assessment and treatment.

[0031] S2. Upload the video data to the edge computing server for preliminary pose estimation, reduce data transmission pressure, improve processing speed, and provide basic data support for deep modeling.

[0032] S3. The data processed by edge computing is transmitted to the cloud server for in-depth modeling and analysis of gait and balance parameters;

[0033] S4. After analyzing the parameters, perform anomaly pattern identification and comparison;

[0034] S5. Upload the compared and analyzed data to the decision application layer, realize clinical rule reasoning through decision-making, match the corresponding treatment plan, and provide visualized data for doctors' observation and research.

[0035] S6. Transmit each parameter and corresponding processing plan to the doctor's end for the doctor to view.

[0036] In step S1, the data acquisition layer uses multi-mimicry data input. It records daily activity videos through a mobile phone camera to facilitate remote detection and follow-up. It analyzes gait symmetry by capturing the full gait from the front. Wearable sensors provide high-precision data on physiological indicators such as blood pressure and heart rate, emotional and psychological characteristics, and physical activity indicators. It records joint angle changes and acceleration data in real time. It analyzes the pressure sensor inside the shoe sole to evaluate the distribution of foot support, gait stability, and load conditions.

[0037] In step S2, the edge computing server preprocesses the video using the OpenPose algorithm to extract motion parameters. The motion parameter extraction uses spatiotemporal parameters to analyze and extract step length, step width, and support phase duration, and uses angle parameters to analyze and extract knee flexion and extension angles and pelvic tilt. The symmetry index is used to compare the differences in joint activity between the left and right sides.

[0038] In the S3 steps, the cloud analysis layer extracts motion parameters and performs spatiotemporal alignment. It captures joint movement trajectories through video and synchronizes sensor data to accurately correspond multi-dimensional information. Simultaneously, feature integration is performed while spatiotemporal alignment is achieved. Feature integration is carried out by building a fusion network using PyTorch, integrating visual features, and integrating sensor features to enhance information complementarity.

[0039] In step S4, abnormal pattern recognition is divided into a classification model and a clustering model. The classification model uses the SVM algorithm to classify small sample high-dimensional data, accurately distinguishing abnormal postures and guiding personalized rehabilitation strategies. The K-means algorithm is used for data mining, which automatically divides rehabilitation stages and sets stage-based rehabilitation goals. Through structured rules, it achieves efficient and interpretable posture analysis and decision-making, and is widely used in rehabilitation assessment, fall warning, and personalized treatment plan development.

[0040] The decision application layer in S5 includes a decision tree rule engine and reinforcement learning optimization. The decision tree rule engine has rehabilitation assessment, fall warning and treatment plan generation. The reinforcement learning optimization includes action space and reward function. The action space provides diverse training strategies for the intelligent training system by selecting the type of orthosis for VR training. In reinforcement learning training, the reward function quantifies the feedback mechanism of action quality to guide rehabilitation training or fall prevention.

[0041] In step S6, after the data is transmitted to the doctor's end, the doctor reviews it and decides on the specific plan to be implemented.

[0042] In the specific usage process, the patient wears the corresponding device and records corresponding videos through the mobile phone camera for analysis and monitoring. At the same time, the wearable device collects motion data and pressure data, which is then uploaded to the edge computing server for preliminary posture estimation. This reduces data transmission pressure, improves processing speed, and provides basic data support for deep modeling. The data after edge computing is transmitted to the cloud server for deep modeling analysis of gait and balance parameters. Subsequently, after parameter analysis, abnormal pattern recognition is performed for comparison. The data after comparison and analysis is uploaded to the decision application layer. Through decision-making, clinical rule reasoning is realized to match the corresponding treatment plan, providing doctors with visualized data for observation and research. Finally, the data is transmitted to the doctor's end for review and decision on the specific treatment plan.

[0043] It will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

[0044] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. A rehabilitation assessment system for patients with central nervous system dysfunction, characterized in that, Includes the following steps: S1. The data acquisition layer collects video, motion, and stress data from users to provide intuitive and realistic audio-visual materials for rehabilitation assessment and treatment. S2. Upload the video data to the edge computing server for preliminary pose estimation, reduce data transmission pressure, improve processing speed, and provide basic data support for deep modeling. S3. The data processed by edge computing is transmitted to the cloud server for in-depth modeling and analysis of gait and balance parameters; S4. After analyzing the parameters, perform anomaly pattern identification and comparison; S5. Upload the compared and analyzed data to the decision application layer, realize clinical rule reasoning through decision-making, match the corresponding treatment plan, and provide visualized data for doctors' observation and research. S6. Transmit each parameter and corresponding processing plan to the doctor's end for the doctor to view.

2. The rehabilitation assessment system for patients with central nervous system dysfunction according to claim 1, characterized in that: In step S1, the data acquisition layer uses multi-mimicry data input. It records daily activity videos through a mobile phone camera to facilitate remote detection and follow-up. It analyzes gait symmetry by capturing the full gait from the front. It provides high-precision data on physiological indicators such as blood pressure and heart rate, emotional and psychological characteristics, and physical activity indicators through wearable sensors. It records joint angle changes and acceleration data in real time by collecting joint angles and acceleration data. It analyzes the foot support distribution, gait stability, and load conditions through pressure sensors inside the shoe sole.

3. The rehabilitation assessment system for patients with central nervous system dysfunction according to claim 1, characterized in that: In step S2, the edge computing server preprocesses the video using the OpenPose algorithm to extract motion parameters. The motion parameter extraction uses spatiotemporal parameters to analyze and extract step length, step width, and support phase duration, and uses angle parameters to analyze and extract knee flexion and extension angles and pelvic tilt. The symmetry index is used to compare the differences in joint activity between the left and right sides.

4. The rehabilitation assessment system for patients with central nervous system dysfunction according to claim 1, characterized in that: In step S3, the cloud analysis layer extracts motion parameters and performs spatiotemporal alignment. It captures joint movement trajectories through video and synchronizes sensor data to accurately correspond multi-dimensional information, achieving multi-dimensional information synchronization. Simultaneously with the spatiotemporal alignment, feature integration is performed. The feature integration is carried out synchronously by constructing a fusion network using PyTorch, integrating visual features, and integrating sensor features to enhance information complementarity.

5. The rehabilitation assessment system for patients with central nervous system dysfunction according to claim 1, characterized in that: In step S4, abnormal pattern recognition is divided into a classification model and a clustering model. The classification model uses the SVM algorithm to classify small sample high-dimensional data, accurately distinguishing abnormal postures and guiding personalized rehabilitation strategies. The K-means algorithm is used for data mining, which automatically divides rehabilitation stages and sets stage-based rehabilitation goals. Through structured rules, it achieves efficient and interpretable posture analysis and decision-making, and is widely used in rehabilitation assessment, fall warning, and personalized treatment plan formulation.

6. The rehabilitation assessment system for patients with central nervous system dysfunction according to claim 1, characterized in that: In step S5, the decision application layer includes a decision tree rule engine and reinforcement learning optimization. The decision tree rule engine has rehabilitation assessment, fall warning, and treatment plan generation functions. The reinforcement learning optimization includes an action space and a reward function. The action space is trained for VR duration by selecting orthotic types, providing diverse training strategies for the intelligent training system. In reinforcement learning training, the reward function quantifies the feedback mechanism of action quality to guide rehabilitation training or fall prevention. A rehabilitation assessment system for patients with central nervous system dysfunction according to claim 1, characterized in that: In step S6, after the data is transmitted to the doctor's end, the doctor reviews it and decides to implement a specific plan.

Images