Intelligent knee joint rehabilitation system based on Internet of Things

Abstract

The invention relates to the technical field of intelligent medical treatment, and discloses an intelligent knee joint rehabilitation system based on Internet of Things, comprising: a flexible multi-parameter sensing module for collecting a multi-modal biological signal of a knee joint; the edge calculation module is used for processing and analyzing sensing data in real time; the dynamic impedance adjusting module is used for dynamically adjusting impedance parameters of rehabilitation training according to the real-time data; the cloud-side collaborative decision module is used for realizing efficient transmission and collaborative decision of data; and the self-adaptive rehabilitation strategy engine is used for optimizing rehabilitation training parameters and comprises a real-time intensity adjustment module based on an improved TD3 reinforcement learning algorithm. According to the system, the flexible strain sensor array unit, the 9-axis inertial measurement unit and the surface myoelectricity sensing unit are integrated through the flexible multi-parameter sensing module, and multi-dimensional real-time monitoring of knee joint strain distribution, three-dimensional acceleration, angular velocity, magnetic field data and myoelectricity signals is achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent medical treatment, and particularly relates to an intelligent knee joint rehabilitation system based on Internet of Things. BACKGROUND

[0002] Knee joint rehabilitation is an important research direction in the field of orthopedics and rehabilitation medicine, especially after knee joint surgery, patients need to recover joint function through systematic rehabilitation training. Traditional knee joint rehabilitation methods mainly rely on the guidance of rehabilitation physicians and the subjective feedback of patients, lacking real-time and accurate monitoring and adjustment mechanism, resulting in uneven rehabilitation effect. In addition, the existing rehabilitation equipment mostly adopts fixed impedance mode, which cannot dynamically adjust the training intensity according to the real-time state of the patient, and is difficult to meet the individualized rehabilitation needs.

[0003] With the development of Internet of Things, edge computing and artificial intelligence technology, intelligent rehabilitation system has gradually become a research hotspot, however, the existing technology still has the following shortcomings:

[0004] 1. The precision and real-time performance of multi-modal biological signal acquisition are insufficient;

[0005] 2. The data processing and decision-making delay is high, which is difficult to meet the real-time rehabilitation needs;

[0006] 3. The individualization and dynamic adjustment capability of rehabilitation strategy is limited;

[0007] 4. There is a lack of efficient remote monitoring and emergency braking mechanism. SUMMARY

[0008] In order to make up for the above shortcomings, the present application provides an intelligent knee joint rehabilitation system based on Internet of Things, which aims to solve the problem of "difficult to meet individualized rehabilitation needs" mentioned in the prior art.

[0009] In order to achieve the above purpose, the present application adopts the following technical scheme: an intelligent knee joint rehabilitation system based on Internet of Things, comprising:

[0010] Flexible multi-parameter sensing module: for collecting multi-modal biological signals of knee joint;

[0011] Edge computing module: for real-time processing and analyzing sensing data;

[0012] Dynamic impedance adjustment module: for dynamically adjusting the impedance parameters of rehabilitation training according to real-time data;

[0013] Cloud-edge collaborative decision-making module: for realizing efficient data transmission and collaborative decision-making;

[0014] Adaptive rehabilitation strategy engine: for optimizing rehabilitation training parameters, including real-time intensity adjustment module based on improved TD3 reinforcement learning algorithm;

[0015] Three-dimensional visualization rehabilitation module: for providing real-time monitoring and remote guidance of rehabilitation training, developed based on Unity3D engine;

[0016] The flexible multi-parameter sensing module, edge computing module, dynamic impedance adjustment module, cloud-edge collaborative decision module, adaptive rehabilitation strategy engine and three-dimensional visualization rehabilitation module are connected through network communication.

[0017] Further description of the above technical solutions:

[0018] The flexible multi-parameter sensing module includes a flexible strain sensor array unit, a 9-axis inertial measurement unit and a surface electromyography sensing unit, the flexible strain sensor array unit is designed with multiple layers of flexible materials and can be attached to the surface of the knee joint to monitor the strain distribution of the knee joint in real time, the 9-axis inertial measurement unit is used to collect three-dimensional acceleration, angular velocity and magnetic field data of the knee joint, in combination with the surface electromyography sensor, to realize multidimensional monitoring of the kinematics and dynamics of the knee joint, and the surface electromyography sensing unit is placed on the skin surface to capture the electrical signals generated when the muscle contracts, reflecting the activation degree and activity pattern of the muscle.

[0019] Further description of the above technical solutions:

[0020] The edge computing module realizes real-time prediction and abnormal motion recognition of the knee joint motion state by deploying a lightweight LSTM model, the response time is less than 50ms, and the LSTM model is trained through multi-source data fusion, which can combine the data of the flexible multi-parameter sensing module, the 9-axis inertial measurement unit and the surface electromyography sensing unit to improve the prediction accuracy.

[0021] Further description of the above technical solutions:

[0022] The dynamic impedance adjustment module realizes continuous torque adjustment of 0-200Nm through the innovative combination of a magneto-rheological damper and a harmonic reducer, the response time is less than 50ms, the magneto-rheological damper adopts an intelligent control algorithm and can dynamically adjust the damping coefficient according to real-time data to optimize the mechanical properties of rehabilitation training.

[0023] Further description of the above technical solutions:

[0024] The cloud-edge collaborative decision module supports dual-mode communication and can realize efficient transmission of data in low-power and high-bandwidth scenarios.

[0025] Further description of the above technical solutions:

[0026] The edge computing module cooperates with the cloud to realize privacy protection and collaborative optimization of multi-agency data through federated learning, and the adaptive rehabilitation strategy engine realizes real-time optimization of training intensity parameters by improving the TD3 reinforcement learning algorithm, and realizes dynamic adjustment in combination with a PID controller.

[0027] As a further description of the above technical solution:

[0028] The three-dimensional visual rehabilitation module is developed based on a Unity3D engine, supports three-dimensional visualization and remote guidance of real-time data, and provides a doctor terminal interactive interface, supports real-time monitoring of rehabilitation progress and an emergency braking mechanism.

[0029] As a further description of the above technical solution:

[0030] The adaptive rehabilitation strategy engine further includes a multi-source data fusion algorithm and an attention mechanism, establishes a knee joint six-degree-of-freedom kinematics model, combines a sensor data fusion framework based on the attention mechanism, improves the accuracy and efficiency of data processing, the TD3 algorithm can adjust the training scheme according to the real-time state of the patient through the training of the knee joint six-degree-of-freedom kinematics model, improves the rehabilitation effect, and the attention mechanism can dynamically allocate weights according to the importance of sensor data, and optimize the data fusion effect.

[0031] As a further description of the above technical solution:

[0032] The edge computing module further includes a 5G+ edge computing collaborative architecture for realizing low-delay transmission and processing of real-time data, supporting a remote rehabilitation guidance mode, and the 5G network cooperates with the edge computing node to ensure the real-time and reliability of data processing.

[0033] As a further description of the above technical solution:

[0034] The system further includes an emergency braking mechanism module, mainly used for automatically triggering a safety circuit when detecting abnormal motion, and ensuring patient safety.

[0035] The present application has the following beneficial effects:

[0036] 1、In the present application, the flexible multi-parameter sensing module integrates a flexible strain sensor array unit, a 9-axis inertial measurement unit and a surface electromyography sensing unit, realizes multidimensional real-time monitoring of knee joint strain distribution, three-dimensional acceleration, angular velocity, magnetic field data and muscle electrical signals, and in combination with a lightweight LSTM model deployed by an edge computing module, can predict the knee joint motion state and identify abnormal motion with a response time of less than 50ms, the prediction accuracy reaches 95%, and the real-time and safety of rehabilitation training are significantly improved.

[0037] 2、In the application, through the innovative combination of the magneto-rheological damper of the dynamic impedance adjustment module and the harmonic reducer, 0-200Nm continuous torque adjustment is realized, the response time is less than 50ms, the adaptive rehabilitation strategy engine based on the improved TD3 reinforcement learning algorithm and the PID controller can dynamically adjust the training intensity parameters according to the real-time state of the patient, and the mechanical properties of the rehabilitation training are optimized. Experimental results show that the optimized training intensity parameters are more in line with the rehabilitation needs of the patient, the average knee joint range of motion is increased by 20%, and the average muscle strength is increased by 15%.

[0038] 3、In the application, through the cloud-edge collaborative decision module and the 5G+edge computing collaborative architecture, efficient data transmission and low delay processing are realized, and a remote rehabilitation guidance mode is supported. The three-dimensional visual rehabilitation module is developed based on the Unity3D engine, provides a three-dimensional visualization of real-time data and a doctor terminal interactive interface, supports real-time monitoring of rehabilitation progress and an emergency braking mechanism, the emergency braking mechanism module automatically triggers a safety circuit when detecting abnormal motion, ensures patient safety, the consistency of the abnormal motion identified by the system and the assessment results of the physician reaches 85%, and the reliability of the rehabilitation training and the convenience of remote guidance are significantly improved. BRIEF DESCRIPTION OF DRAWINGS

[0039] Figure 1 It is a whole system framework schematic diagram of an intelligent knee joint rehabilitation system based on the Internet of Things in the application;

[0040] Figure 2 It is a flexible multi-parameter sensing module detailed framework schematic diagram of an intelligent knee joint rehabilitation system based on the Internet of Things in the application.

[0041] LEGEND:

[0042] 1, flexible multi-parameter sensing module; 2, edge computing module; 3, dynamic impedance adjustment module; 4, cloud-edge collaborative decision module; 5, adaptive rehabilitation strategy engine; 6, three-dimensional visual rehabilitation module; 7, emergency braking mechanism module; 101, flexible strain sensor array unit; 102, 9-axis inertial measurement unit; 103, surface electromyography sensing unit. DETAILED DESCRIPTION

[0043] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the application.

[0044] REFERENCE Figure 1 - Figure 2The present invention provides an embodiment of an intelligent knee joint rehabilitation system based on the Internet of Things, comprising a flexible multi-parameter sensing module 1, an edge computing module 2, a dynamic impedance adjustment module 3, a cloud-edge collaborative decision-making module 4, an adaptive rehabilitation strategy engine 5, a three-dimensional visualization rehabilitation module 6, and an emergency braking mechanism module 7. The flexible multi-parameter sensing module 1, the edge computing module 2, the dynamic impedance adjustment module 3, the cloud-edge collaborative decision-making module 4, the adaptive rehabilitation strategy engine 5, the three-dimensional visualization rehabilitation module 6, and the emergency braking mechanism module 7 are connected via network communication.

[0045] The flexible multi-parameter sensing module 1 is used to collect multimodal biosignals of the knee joint. Specifically, the flexible multi-parameter sensing module 1 includes a flexible strain sensor array unit 101, a 9-axis inertial measurement unit 102, and a surface electromyography (EMG) sensing unit 103. The flexible strain sensor array unit 101 is designed with multi-layer flexible materials, which can conform to the surface of the knee joint and monitor the strain distribution of the knee joint in real time. The 9-axis inertial measurement unit 102 is used to collect three-dimensional acceleration, angular velocity, and magnetic field data of the knee joint. Combined with the surface EMG sensor, it realizes multi-dimensional monitoring of the kinematics and dynamics of the knee joint. The surface EMG sensing unit 103 is a sensor that is placed on the skin surface to capture the electrical signals generated when the muscles contract, reflecting the degree of muscle activation and activity pattern.

[0046] Edge computing module 2 is used for real-time processing and analysis of sensor data. By deploying a lightweight LSTM model, it achieves real-time prediction of knee joint motion status and abnormal movement recognition with a response time of less than 50ms. The LSTM model is trained with multi-source data fusion and can combine data from flexible multi-parameter sensing module 1, 9-axis inertial measurement unit 102, and surface electromyography sensing unit 103 to improve prediction accuracy. The following formula can be used as a reference:

[0047] i t =σ(W i ·[h t-1 ,x t ]+b i )

[0048] Among them, i t It is the output of the input gate, σ is the Sigmoid activation function, and W is the output of the input gate. i It is the weight matrix, h t-1 It is the hidden state from the previous moment, x t This is the current input, b i It is a bias term;

[0049] f t =σ(W f ·[h t-1 ,x t ]+b f )

[0050] wherein f t is the output of the forget gate, which determines how much memory from the previous time step is retained;

[0051]

[0052] wherein, is the candidate memory cell, and tanh is the hyperbolic tangent activation function;

[0053]

[0054] wherein C t is the memory cell at the current time step;

[0055] o t = sigma(W o · [h t-1 , x t ] + b o ) h t = o t · tanh(C t )

[0056] wherein h t is the hidden state at the current time step.

[0057] In addition, the edge computing module 2 cooperates with the cloud to realize privacy protection and collaborative optimization of multi-institution data through federated learning. The edge computing module 2 further comprises a 5G+ edge computing collaborative architecture for realizing low-delay transmission and processing of real-time data, supporting a remote rehabilitation guidance mode. The 5G network cooperates with the edge computing node to ensure the real-time and reliability of data processing.

[0058] The dynamic impedance adjustment module 3 is used for dynamically adjusting the impedance parameters of the rehabilitation training according to real-time data. Through the innovative combination of the magneto-rheological damper and the harmonic reducer, the dynamic impedance adjustment module 3 realizes continuous torque adjustment of 0-200 Nm with a response time less than 50 ms. The magneto-rheological damper adopts an intelligent control algorithm, which can dynamically adjust the damping coefficient according to real-time data to optimize the mechanical properties of the rehabilitation training.

[0059] The cloud-edge collaborative decision module (4) is used for realizing efficient transmission and collaborative decision of data, and supports dual-mode communication, namely BLE 5.2 and LoRaWAN, which can realize efficient transmission of data in low-power and high-bandwidth scenarios.

[0060] The adaptive rehabilitation strategy engine 5 is used to optimize the rehabilitation training parameters, including a real-time intensity adjustment module based on an improved TD3 reinforcement learning algorithm, and it optimizes the training intensity parameters in real time through the improved TD3 reinforcement learning algorithm, and realizes dynamic adjustment combined with a PID controller, the adaptive rehabilitation strategy engine 5 also includes a multi-source data fusion algorithm and an attention mechanism, establishes a six-degree-of-freedom kinematics model of the knee joint, combines a sensor data fusion framework based on the attention mechanism, and improves the accuracy and efficiency of data processing, the attention mechanism refers to the following formula:

[0061]

[0062] Where, a i is the weight of the i th sensor, e i is the importance score of the i th sensor.

[0063] The TD3 algorithm can adjust the training scheme according to the real-time state of the patient through the training of the six-degree-of-freedom kinematics model of the knee joint, improve the rehabilitation effect, and the attention mechanism can dynamically allocate weights according to the importance of the sensor data to optimize the data fusion effect;

[0064] The three-dimensional visual rehabilitation module 6 is used to provide real-time monitoring and remote guidance for rehabilitation training, which is developed based on the Unity3D engine, supports real-time data three-dimensional visualization and remote guidance, and provides a doctor terminal interactive interface, supports real-time monitoring of rehabilitation progress and emergency braking mechanism, and the emergency braking mechanism module 7 is mainly used to automatically trigger the safety circuit when detecting abnormal motion to ensure patient safety.

[0065] Example two

[0066] Based on example one, experiment 1 is carried out:

[0067] Experimental subjects: 10 patients with knee joint rehabilitation, aged 25-60 years old, all in the postoperative rehabilitation stage of knee joint.

[0068] Experimental equipment: intelligent knee joint rehabilitation system equipped with flexible multi-parameter sensing module 1, edge computing module 2, dynamic impedance adjustment module 3, cloud-edge collaborative decision module 4, adaptive rehabilitation strategy engine 5, three-dimensional visual rehabilitation module 6 and emergency braking mechanism module 7.

[0069] Experimental process:

[0070] S1: Each patient wears the intelligent knee joint rehabilitation system for rehabilitation training for one week, 30 minutes a day.

[0071] S2: The flexible multi-parameter sensing module 1 collects real-time strain distribution, three-dimensional acceleration, angular velocity, magnetic field data and surface electromyography signals of the knee joint.

[0072] S3: The edge computing module 2 processes and analyzes the sensing data in real time through a lightweight LSTM model, predicts the motion state of the knee joint and identifies abnormal movements.

[0073] S4: The system records the identification of abnormal movements in each training and compares it with the assessment results of professional rehabilitation physicians.

[0074] Experimental results:

[0075] Prediction accuracy: The prediction accuracy of the LSTM model for the motion state of the knee joint reaches 95%, and the accuracy rate of abnormal movement recognition is 90%;

[0076] Response time: The response time of the edge computing module is less than 50ms, meeting the real-time requirements;

[0077] Physician assessment comparison: The consistency of the system-identified abnormal movements with the physician assessment results is 85%, showing high reliability.

[0078] Example three

[0079] Based on example one and example two, experiment 2 is conducted:

[0080] Experimental subjects: 15 patients with knee joint rehabilitation, aged 30-65 years old, all in the postoperative rehabilitation stage of knee joint.

[0081] Experimental equipment: consistent with example two.

[0082] Experimental process:

[0083] S1: Each patient wears the intelligent knee joint rehabilitation system for two weeks of rehabilitation training, 45 minutes a day.

[0084] S2: The adaptive rehabilitation strategy engine optimizes the training intensity parameters in real time through the improved TD3 reinforcement learning algorithm, and dynamically adjusts the impedance parameters combined with the PID controller.

[0085] S3: The system records the changes of training intensity parameters in each training and evaluates the rehabilitation effect of the patients.

[0086] S4: The rehabilitation effect is evaluated by the range of motion of the knee joint, muscle strength and subjective feeling of the patients.

[0087] Experimental results

[0088] Training intensity optimization: The improved TD3 algorithm can dynamically adjust the training intensity according to the real-time state of the patient, and the optimized training intensity parameters are more in line with the rehabilitation needs of the patient.

[0089] Rehabilitation effect: After two weeks of training, the knee joint activity of the patient increased by an average of 20%, and the muscle strength increased by an average of 15%.

[0090] Patient subjective experience: 90% of patients feel that the training intensity is moderate, and the rehabilitation process is comfortable, without obvious discomfort.

[0091] Finally, it should be pointed out that the above-mentioned is only the preferred embodiment of the present application, and is not used to limit the present application, although the present application has been described in detail with reference to the foregoing embodiments, for those skilled in the art, the technical solutions recorded in the foregoing embodiments can still be modified, or some technical features can be replaced, any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application, should be included in the protection scope of the present application.

Claims

1. An Internet of Things based intelligent knee rehabilitation system, characterized in that: Comprise: Flexible multi-parameter sensing module (1): for collecting multi-modal biological signals of knee joint; Edge computing module (2): for real-time processing and analysis of sensing data; Dynamic impedance adjustment module (3): for dynamically adjusting the impedance parameters of rehabilitation training according to real-time data; Cloud-edge collaborative decision module (4): for efficient data transmission and collaborative decision-making; Adaptive rehabilitation strategy engine (5): for optimizing rehabilitation training parameters, including real-time intensity adjustment module based on improved TD3 reinforcement learning algorithm; Three-dimensional visual rehabilitation module (6): for providing real-time monitoring and remote guidance of rehabilitation training, developed based on Unity3D engine; The flexible multi-parameter sensing module (1), edge computing module (2), dynamic impedance adjustment module (3), cloud-edge collaborative decision module (4), adaptive rehabilitation strategy engine (5) and three-dimensional visual rehabilitation module (6) are connected through network communication.

2. The intelligent knee rehabilitation system based on the Internet of Things according to claim 1, characterized in that: The flexible multi-parameter sensing module (1) comprises a flexible strain sensor array unit (101), a 9-axis inertial measurement unit (102) and a surface electromyography sensing unit (103). The flexible strain sensor array unit (101) is designed with multiple layers of flexible material, which can conform to the surface of the knee joint and monitor the strain distribution of the knee joint in real time. The 9-axis inertial measurement unit (102) is used to collect three-dimensional acceleration, angular velocity and magnetic field data of the knee joint, combined with surface electromyography sensors, to realize multi-dimensional monitoring of knee joint kinematics and dynamics. The surface electromyography sensing unit (103) is placed on the skin surface to capture the electrical signals generated during muscle contraction, reflecting the degree of muscle activation and activity pattern. 3.The smart knee rehabilitation system based on the Internet of Things according to claim 2, characterized in that: The edge computing module (2) realizes real-time prediction and abnormal motion recognition of knee joint motion state by deploying a lightweight LSTM model, with a response time less than 50ms. The LSTM model is trained with multi-source data fusion, which can combine the data of the flexible multi-parameter sensing module (1), the 9-axis inertial measurement unit (102) and the surface electromyography sensing unit (103) to improve the prediction accuracy.

4. The intelligent knee rehabilitation system based on the Internet of Things according to claim 1, characterized in that: The dynamic impedance adjustment module (3) realizes continuous torque adjustment of 0-200Nm through the innovative combination of magneto-rheological damper and harmonic reducer, with a response time less than 50ms. The magneto-rheological damper uses an intelligent control algorithm to dynamically adjust the damping coefficient according to real-time data, optimizing the mechanical properties of rehabilitation training.

5. The intelligent knee rehabilitation system based on the Internet of Things according to claim 1, characterized in that: The cloud-edge collaborative decision module (4) supports dual-mode communication and can realize efficient data transmission in low-power and high-bandwidth scenarios. 6.The smart knee rehabilitation system based on the Internet of Things according to claim 1, wherein: The edge computing module (2) cooperates with the cloud to realize privacy protection and collaborative optimization of multi-institution data through federated learning. The adaptive rehabilitation strategy engine (5) optimizes training intensity parameters in real time through the improved TD3 reinforcement learning algorithm, combined with a PID controller to realize dynamic adjustment. 7.The smart knee rehabilitation system based on the Internet of Things according to claim 1, wherein: The three-dimensional visual rehabilitation module (6) is developed based on Unity3D engine, supporting three-dimensional visualization and remote guidance of real-time data, and providing a doctor terminal interactive interface to support real-time monitoring and emergency braking mechanism of rehabilitation progress. 8.The smart knee rehabilitation system based on Internet of Things according to claim 1, wherein: The adaptive rehabilitation strategy engine (5) further includes a multi-source data fusion algorithm and an attention mechanism, establishes a six-degree-of-freedom kinematics model of the knee joint, combines an attention mechanism-based sensor data fusion framework, and improves the accuracy and efficiency of data processing. The TD3 algorithm can adjust the training scheme according to the real-time state of the patient through the training of the six-degree-of-freedom kinematics model of the knee joint, improve the rehabilitation effect, and the attention mechanism can dynamically allocate weights according to the importance of sensor data to optimize the data fusion effect. 9.The smart knee rehabilitation system based on the Internet of Things according to claim 1, wherein: The edge computing module (2) further includes a 5G+ edge computing collaborative architecture for realizing low-delay transmission and processing of real-time data and supporting a remote rehabilitation guidance mode. The 5G network and edge computing node cooperate to ensure the real-time and reliability of data processing. 10.The smart knee rehabilitation system based on the Internet of Things according to claim 1, wherein: The system further includes an emergency braking mechanism module (7) mainly used for automatically triggering a safety circuit when abnormal motion is detected to ensure patient safety.

Images