Lower limb artery occlusion assessment and rehabilitation system based on functional behaviors

Abstract

The invention relates to the technical field of disease rehabilitation assessment, and discloses a functional behavior-based assessment and rehabilitation system for lower limb artery occlusion, which is characterized in that gait data, muscle force data, blood flow change data and electromyographic signal data of a patient are acquired by arranging a data acquisition module, and a rehabilitation assessment report of the patient is generated by using a multi-modal fusion algorithm; and the artery occlusion degree, the lower limb blood circulation condition and the motion function state of the patient are comprehensively evaluated. On the basis, the system generates a personalized exercise scheme through a reinforcement learning algorithm, and dynamically adjusts the exercise scheme according to the real-time exercise data feedback of the patient, so as to ensure the effectiveness and safety of the rehabilitation scheme. Meanwhile, the system provides a guide feedback module, the patient is guided to complete the correct motion posture through voice, vision or vibration prompts, and secondary injury caused by improper motion posture is avoided. By dynamically optimizing the exercise scheme and guiding feedback in real time, personalized rehabilitation guidance of the patient is achieved, and the rehabilitation effect and exercise compliance are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of rehabilitation assessment, and particularly to an assessment and rehabilitation system for lower extremity arterial occlusive disease based on functional behavior. Background Art

[0002] Lower extremity arterial occlusive disease (LEAOD) is a common peripheral vascular disease. Its main cause is arterial stenosis or occlusion caused by atherosclerosis, resulting in lower extremity blood circulation disorders. In severe cases, it can lead to tissue necrosis and even amputation. The treatment options for this disease mainly include drug treatment, interventional treatment, and surgical operation. However, clinical studies have shown that rehabilitation training, especially exercise therapy, plays an important role in improving the lower extremity blood circulation of patients, relieving symptoms, and reducing the risk of restenosis. Therefore, it is crucial to reasonably formulate personalized exercise programs and dynamically adjust the exercise intensity and frequency to promote the rehabilitation of patients.

[0003] However, current rehabilitation systems often adopt fixed exercise programs, lack real-time collection and analysis of patients' functional behavior data, and cannot be dynamically adjusted according to the individualized rehabilitation needs of patients, resulting in poor rehabilitation effects. In addition, existing systems lack a guiding and feedback mechanism for patients' exercise postures and exercise effects, and problems such as improper exercise postures, excessive or insufficient exercise intensity are likely to occur, affecting the rehabilitation process of patients. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide an assessment and rehabilitation system for lower extremity arterial occlusive disease based on functional behavior, so as to solve the problems that when formulating exercise programs for patients with lower extremity arterial occlusive disease, it is impossible to dynamically adjust according to the individualized rehabilitation needs of patients, lack a guiding and feedback mechanism for patients' exercise postures and exercise effects, and result in poor rehabilitation effects.

[0005] The present invention discloses an assessment and rehabilitation system for lower extremity arterial occlusive disease based on functional behavior. The system includes a data collection module, an assessment module, a guiding and feedback module, a rehabilitation guidance module, and a data storage module; wherein,

[0006] The data collection module is used to collect the functional behavior data of patients; the functional behavior data includes gait data, muscle strength data, blood flow change data, and electromyogram signal data;

[0007] The assessment module is used to generate a rehabilitation assessment report for patients based on the collected functional behavior data by using a multi-modal fusion algorithm;

[0008] The guiding feedback module is used to guide the patient to correctly complete the movement postures in the movement plan through voice, visual or vibration prompts when the patient is exercising according to the initial movement plan, and record the movement data of the patient during the exercise process;

[0009] The rehabilitation guidance module is used to generate a personalized movement plan based on the rehabilitation assessment report through an artificial intelligence algorithm, adjust the movement plan according to the patient's movement data, and use the adjusted movement plan as the patient's next movement plan;

[0010] The data storage module is used to record the patient's movement data and rehabilitation assessment report for the doctor to view and adjust the rehabilitation plan.

[0011] Furthermore, the data acquisition module includes a gait sensor, a muscle strength sensor, a blood flow sensor, and an electromyogram sensor; among them,

[0012] The gait sensor is installed at the sole and / or ankle position of the patient and is used to collect the patient's gait data;

[0013] The muscle strength sensor is installed at the thigh and / or calf part of the patient and is used to measure the lower limb muscle strength of the patient;

[0014] The blood flow sensor is installed at the position of the patient's lower limb artery and is used to monitor the blood flow velocity and blood oxygen saturation of the patient's lower limbs;

[0015] The electromyogram sensor is installed around the patient's knee joint and is used to detect the lower limb electromyogram signal of the patient.

[0016] Furthermore, the initial movement plan is formulated by the doctor according to the patient's basic physical parameters.

[0017] Furthermore, the evaluation module generates the patient's rehabilitation assessment report based on the collected functional behavior data using a multimodal fusion algorithm, specifically including:

[0018] Perform normalization processing on the functional behavior data, and perform feature extraction operations through a multimodal fusion algorithm after normalization processing to obtain a comprehensive feature vector;

[0019] Input the comprehensive feature vector into the rehabilitation assessment model, and generate a rehabilitation assessment report based on the output result of the rehabilitation assessment model.

[0020] Furthermore, the feature extraction operation performed through the multimodal fusion algorithm includes:

[0021] Use the long short-term memory network LSTM to perform time-dynamic feature extraction operations on gait data, blood flow change data, and electromyogram signal data to generate a time series feature vector;

[0022] Perform Fourier transform on the electromyogram signal data, extract the main frequency components, spectral energy distribution, and frequency bandwidth, and generate a frequency-domain feature vector;

[0023] Perform static feature extraction on the muscle strength data through a convolutional neural network CNN to generate a spatial feature vector;

[0024] Calculate the modal weights based on the importance of each modality data, and fuse the time-series feature vector, frequency-domain feature vector, and spatial feature vector through a fully connected network according to the calculated modal weights to generate a comprehensive feature vector.

[0025] Further, the rehabilitation evaluation model consists of a multi-layer perceptron MLP network and a decision tree ensemble model;

[0026] The multi-layer perceptron network is used to perform high-dimensional feature extraction and continuous variable prediction based on the comprehensive feature vector, and output the patient's motor ability score and lower limb blood circulation condition score;

[0027] The decision tree ensemble model is used to perform classification decisions based on the output results of the multi-layer perceptron network and generate different indicators in the rehabilitation evaluation report. Further, the output results of the rehabilitation evaluation model include the arterial occlusion degree score, lower limb blood circulation condition score, and motor function limitation level; the rehabilitation evaluation report includes the output results of the rehabilitation evaluation model.

[0028] Further, the generation of a personalized exercise plan based on the rehabilitation evaluation report through an artificial intelligence algorithm includes:

[0029] Determine the patient's basic physical parameters; the basic physical parameters include age, height, weight, gender, and past medical history;

[0030] Use the output results of the rehabilitation evaluation model and the patient's basic physical parameters as input data and input them into the exercise plan generation model, and output the exercise plan through the exercise plan generation model;

[0031] The exercise plan generation model uses a reinforcement learning algorithm.

[0032] Further, the reinforcement learning algorithm includes the definition of a state space, an action space, and a reward function; among them,

[0033] Define the output results of the rehabilitation evaluation model and the patient's basic physical parameters as the state space;

[0034] Define the action space as different combinations of exercise plans. The exercise plans include exercise types, exercise intensities, and exercise durations; the exercise types include Berg exercise, gait training, resistance training, and dynamic stretching;

[0035] Design a reward function based on the exercise data feedback by the patient; the exercise data includes functional behavior data, physiological parameter data, and exercise posture data.

[0036] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0037] In the present invention, a data acquisition module is provided to collect the gait data, muscle strength data, blood flow change data, and electromyogram signal data of the patient, and a multi-modal fusion algorithm is used to generate a rehabilitation assessment report of the patient, comprehensively evaluating the degree of arterial occlusion, the lower limb blood circulation condition, and the motor function state of the patient. On this basis, the system generates a personalized exercise plan through a reinforcement learning algorithm and dynamically adjusts the exercise plan according to the real-time exercise data feedback of the patient to ensure the effectiveness and safety of the rehabilitation plan. At the same time, the system provides a guiding feedback module to guide the patient to complete the correct exercise posture through voice, vision, or vibration prompts, avoiding secondary injuries caused by improper exercise postures. The present invention realizes personalized rehabilitation guidance for patients and improves the rehabilitation effect and exercise compliance by dynamically optimizing the exercise plan and providing real-time guiding feedback. Description of the Drawings

[0038] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, form a part of the present application, and do not limit the embodiments of the present invention. In the drawings:

[0039] Figure 1 is a schematic structural diagram of an evaluation and rehabilitation system for lower limb arterial occlusion based on functional behavior disclosed in another embodiment of the present invention. Detailed Embodiments

[0040] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments.

[0041] Embodiment 1

[0042] The present invention discloses an evaluation and rehabilitation system for lower limb arterial occlusion based on functional behavior. Please refer to Figure 1 , Figure 1 is a schematic structural diagram of an evaluation and rehabilitation system for lower limb arterial occlusion based on functional behavior disclosed in an embodiment of the present invention. The system includes a data acquisition module, an evaluation module, a guiding feedback module, a rehabilitation guidance module, and a data storage module; wherein,

[0043] The data acquisition module is used to collect the functional behavior data of the patient; the functional behavior data includes, but is not limited to, gait data, muscle strength data, blood flow change data, and electromyogram signal data.

[0044] The evaluation module is used to generate a rehabilitation evaluation report for the patient by using a multimodal fusion algorithm based on the collected functional behavior data;

[0045] The guiding feedback module is used to, when the patient performs exercises according to the initial exercise plan, guide the patient to correctly complete the exercise postures in the exercise plan through voice, visual or vibration prompts, and record the exercise data of the patient during the exercise process;

[0046] The rehabilitation guidance module is used to generate a personalized exercise plan based on the rehabilitation evaluation report through an artificial intelligence algorithm, adjust the exercise plan according to the exercise data of the patient, and use the adjusted exercise plan as the patient's next exercise plan;

[0047] The data storage module is used to record the exercise data and rehabilitation evaluation report of the patient for the doctor to view and adjust the rehabilitation plan.

[0048] Furthermore, the data acquisition module includes but is not limited to a gait sensor, a muscle strength sensor, a blood flow sensor, and an electromyography sensor; among them,

[0049] The gait sensor is installed at the sole and / or ankle position of the patient and is used to collect the gait data of the patient;

[0050] The muscle strength sensor is installed at the thigh and / or calf part of the patient and is used to measure the lower limb muscle strength of the patient;

[0051] The blood flow sensor is installed at the position of the patient's lower limb artery and is used to monitor the blood flow velocity and blood oxygen saturation of the patient's lower limbs, that is, the blood flow change data includes blood flow velocity and blood oxygen saturation data.

[0052] The electromyography sensor is installed around the patient's knee joint and is used to detect the lower limb electromyography signal of the patient.

[0053] Furthermore, the initial exercise plan is formulated by the doctor according to the basic physical parameters of the patient.

[0054] Furthermore, the evaluation module generates a rehabilitation evaluation report for the patient by using a multimodal fusion algorithm based on the collected functional behavior data, specifically including:

[0055] Normalize the functional behavior data, and perform feature extraction operations through a multimodal fusion algorithm after normalization to obtain a comprehensive feature vector;

[0056] Input the comprehensive feature vector into the rehabilitation evaluation model, and generate a rehabilitation evaluation report based on the output result of the rehabilitation evaluation model.

[0057] Furthermore, performing feature extraction operations through a multimodal fusion algorithm includes:

[0058] Use the long short-term memory network (LSTM) to perform time-dynamic feature extraction operations on gait data, blood flow change data, and electromyogram (EMG) signal data to generate time-series feature vectors;

[0059] Perform Fourier transform on the EMG signal data to extract the main frequency components, spectral energy distribution, and frequency bandwidth, and generate frequency-domain feature vectors;

[0060] Perform static feature extraction operations on the muscle strength data through a convolutional neural network (CNN) to generate spatial feature vectors;

[0061] Calculate the modal weights based on the importance of each type of modal data, and fuse the time-series feature vectors, frequency-domain feature vectors, and spatial feature vectors through a fully connected network according to the calculated modal weights to generate comprehensive feature vectors.

[0062] It can be understood that the EMG signal data mainly reflects the changes in electrical activities generated by muscles during movement, and is an important basis for analyzing the muscle function status of patients.

[0063] Since the EMG signal data has time-series characteristics, and its intensity and fluctuation frequency change over time, the LSTM is used to perform time-dynamic feature extraction on the EMG signal data. The LSTM network can capture the change trends of EMG signals at different time steps and extract time-dynamic features such as muscle activation patterns and fatigue development processes. These features can help evaluate the patient's continuous muscle contraction ability during movement and the fatigue state during exercise, providing data support for dynamically adjusting the exercise plan.

[0064] In addition, since the EMG signal data also contains frequency-domain information, that is, the energy distribution of the signal in different frequency ranges. By performing Fourier transform on the EMG signal data to convert the EMG signal from the time domain to the frequency domain, the main frequency components, spectral energy distribution, and frequency bandwidth of the EMG signal can be extracted. These features can reflect the muscle contraction frequency, muscle activity intensity, and muscle fatigue state of the patient. For example, an increase in the energy of the low-frequency component may indicate muscle fatigue, while the energy distribution of the high-frequency component reflects the rapid muscle contraction ability.

[0065] Through the dual analysis of time-dynamic feature extraction and frequency-domain feature extraction of the EMG signal data, the present invention can more comprehensively evaluate the muscle function status of patients.

[0066] Modalities refer to different types of data sources and feature representations, including time - dynamic modality, frequency - domain modality, and spatial modality. In the embodiments of the present invention, by introducing a multi - modality fusion algorithm, it is possible to comprehensively analyze the time - dynamic features, frequency - domain features, and spatial features of patients, and achieve in - depth mining and fusion of patients' functional behavior data. At the same time, through the dynamic adjustment of modality weights, it is possible to reasonably allocate the importance of different modality data according to the actual situation of patients, and generate a more individualized comprehensive feature vector.

[0067] Further, the rehabilitation assessment model is composed of a multi - layer perceptron (MLP) network and a decision - tree ensemble model.

[0068] Among them, the multi - layer perceptron network is used for high - dimensional feature extraction and continuous variable prediction based on the comprehensive feature vector, and outputs the motor ability score and the lower - limb blood circulation situation score of the patient.

[0069] The decision - tree ensemble model is used to perform classification decisions based on the output results of the multi - layer perceptron network, and generate different indicators in the rehabilitation assessment report.

[0070] The multi - layer perceptron network is a feed - forward neural network, which consists of multiple fully - connected layers and non - linear activation functions. The high - dimensional features of patients extracted by the MLP network in the embodiments of the present invention include, but are not limited to: the stride, step frequency, gait change trend of gait data, the force change distribution of muscle strength data, the fluctuation amplitude and average velocity of blood flow change data, and the main frequency component and spectral energy of electromyogram signal data.

[0071] The decision - tree ensemble model adopts the gradient - boosting decision - tree method to perform classification decisions based on the output results of the MLP network, including, but not limited to, the classification of arterial occlusion degree, the classification of lower - limb blood circulation situation, and the classification of motor function status.

[0072] Preferably, the continuous variable prediction formula of the MLP network includes:

[0073]

[0074] where y score is the predicted continuous variable score, including the motor ability score and the lower - limb blood circulation situation score; L is the number of network layers; h (l) is the feature vector of the l - th layer; W (l) is the weight matrix of the l - th layer; b l is the bias vector of the l - th layer; σ is the activation function; is the weight adjustment coefficient of the feature vector of each layer; γ is the weight adjustment sensitivity parameter; ΔF l is the change in the influence amplitude of the current modality data on the prediction result, and one - layer network corresponds to one modality data vector.

[0075] In an embodiment of the present invention, by setting the rehabilitation evaluation model to be composed of a multi-layer perceptron network and a decision tree ensemble model, high-dimensional feature extraction and classification decision can be performed on the functional behavior data of patients, thereby improving the accuracy of patient rehabilitation evaluation.

[0076] Further, the output results of the rehabilitation evaluation model include the arterial occlusion degree score, the lower limb blood circulation status score, and the motor function limitation level; the rehabilitation evaluation report includes, but is not limited to, the output results of the rehabilitation evaluation model.

[0077] Further, generating a personalized exercise plan based on the rehabilitation evaluation report through an artificial intelligence algorithm includes:

[0078] Determine the basic physical parameters of the patient; the basic physical parameters include age, height, weight, gender, and medical history;

[0079] Input the output results of the rehabilitation evaluation model and the basic physical parameters of the patient as input data into the exercise plan generation model, and output the exercise plan through the exercise plan generation model;

[0080] The exercise plan generation model uses a reinforcement learning algorithm.

[0081] Further, the reinforcement learning algorithm includes the definition of a state space, an action space, and a reward function; wherein,

[0082] Define the output results of the rehabilitation evaluation model and the basic physical parameters of the patient as the state space;

[0083] Define the action space as different combinations of exercise plans, and the exercise plans include exercise types, exercise intensities, and exercise durations; the exercise types include Berg exercise, gait training, resistance training, and dynamic stretching;

[0084] Design a reward function according to the exercise data feedback by the patient; the exercise data includes functional behavior data, physiological parameter data, and exercise posture data.

[0085] Specifically, the physiological parameter data includes, but is not limited to, the patient's heart rate, blood pressure, etc., and is monitored by corresponding physiological parameter sensors. The exercise posture data is recorded by a camera or an inertial sensor arranged in the guidance feedback module.

[0086] Further, the reward function R t The formula is:

[0087] R t =α·ΔF t -β·E t

[0088] Wherein, ΔF tThe improvement amplitude of the rehabilitation evaluation index for the patient at time step t; E t The energy consumption during the patient's movement at time step t; α and β are weight parameters.

[0089] Specifically, the improvement amplitude of the rehabilitation evaluation index is determined by comparing the initial motion data and the real-time motion data. The energy consumption during the movement is calculated based on data such as gait data, electromyogram signal data, heart rate, blood pressure, etc., to evaluate whether the patient's exercise load is reasonable.

[0090] In the embodiment of the present invention, a personalized exercise plan can be dynamically generated according to the patient's rehabilitation evaluation report and real-time motion data, and the patient can be guided to correctly complete the exercise through the guidance feedback module to avoid exercise injuries. Compared with the existing fixed exercise plan, the system of the present invention can adjust the exercise type, intensity and duration in real time, realizing precise rehabilitation guidance and greatly improving the patient's rehabilitation effect and compliance.

[0091] Finally, it should be noted that: An evaluation and rehabilitation system for lower extremity arterial occlusion based on functional behavior disclosed in the embodiments of the present invention only discloses the preferred embodiments of the present invention, and is only used to illustrate the technical solutions of the present invention, rather than to limit it; Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A functional behavior-based assessment and rehabilitation system for lower extremity arterial occlusive disease, characterized in that: The system includes a data acquisition module, an evaluation module, a guidance feedback module, a rehabilitation guidance module and a data storage module; wherein, The data acquisition module is used to collect the patient's functional behavior data; the functional behavior data includes gait data, muscle strength data, blood flow change data and electromyographic signal data; The evaluation module is used to generate a rehabilitation evaluation report for the patient based on the collected functional behavior data using a multimodal fusion algorithm; The guidance feedback module is used to guide the patient to correctly complete the exercise posture in the exercise plan through voice, visual or vibration prompts when the patient exercises according to the initial exercise plan, and record the patient's exercise data during the exercise process; The rehabilitation guidance module is used to generate a personalized exercise plan based on the rehabilitation assessment report through an artificial intelligence algorithm, and adjust the exercise plan according to the patient's exercise data, and use the adjusted exercise plan as the patient's next exercise plan; The data storage module is used to record the patient's movement data and rehabilitation assessment report for doctors to review and adjust the rehabilitation plan.

2. The functional behavior-based lower extremity arterial occlusive disease assessment and rehabilitation system according to claim 1, characterized in that: The data acquisition module includes a gait sensor, a muscle force sensor, a blood flow sensor and an electromyography sensor; wherein, The gait sensor is installed on the sole and / or ankle of the patient to collect the gait data of the patient; The muscle force sensor is installed on the patient's thigh and / or calf to measure the patient's lower limb muscle strength; The blood flow sensor is installed at the patient's lower limb artery to monitor the patient's lower limb blood flow velocity and blood oxygen saturation; The myoelectric sensor is installed around the patient's knee joint and is used to detect the patient's lower limb myoelectric signals.

3. The functional behavior-based lower extremity arterial occlusive disease assessment and rehabilitation system according to claim 1, characterized in that: The initial exercise program is formulated by a doctor based on the patient's basic physical parameters.

4. The functional behavior-based lower extremity arterial occlusive disease assessment and rehabilitation system according to any one of claims 1 to 3, characterized in that: The evaluation module generates a rehabilitation evaluation report for the patient based on the collected functional behavior data using a multimodal fusion algorithm, specifically including: Normalizing the functional behavior data, and performing feature extraction operations through a multimodal fusion algorithm after normalization to obtain a comprehensive feature vector; The comprehensive feature vector is input into the rehabilitation assessment model, and a rehabilitation assessment report is generated based on the output results of the rehabilitation assessment model.

5. The functional behavior-based lower extremity arterial occlusive disease assessment and rehabilitation system according to claim 4, characterized in that: The performing of feature extraction operation by the multimodal fusion algorithm comprises: Use the long short-term memory network LSTM to extract time dynamic features from gait data, blood flow change data, and electromyographic signal data to generate time series feature vectors; Perform Fourier transform on the electromyographic signal data, extract the main frequency component, spectrum energy distribution and frequency bandwidth, and generate frequency domain feature vector; The convolutional neural network (CNN) is used to extract static features from muscle strength data and generate spatial feature vectors. The modal weight is calculated based on the importance of each modal data, and the time series feature vector, frequency domain feature vector and spatial feature vector are fused through a fully connected network according to the calculated modal weight to generate a comprehensive feature vector.

6. The functional behavior-based lower extremity arterial occlusive disease assessment and rehabilitation system according to claim 5, characterized in that: The rehabilitation assessment model is composed of a multi-layer perceptron MLP network and a decision tree integration model; The multilayer perceptron network is used to extract high-dimensional features and predict continuous variables based on the comprehensive feature vector, and output the patient's exercise ability score and lower limb blood circulation score; The decision tree ensemble model is used to perform classification decisions based on the output results of the multi-layer perceptron network and generate different indicators in the rehabilitation assessment report.

7. The functional behavior-based lower extremity arterial occlusive disease assessment and rehabilitation system according to claim 6, characterized in that: The output results of the rehabilitation assessment model include the arterial occlusion degree score, the lower limb blood circulation status score, and the motor function limitation level; the rehabilitation assessment report includes the output results of the rehabilitation assessment model.

8. The functional behavior-based lower extremity arterial occlusive disease assessment and rehabilitation system according to claim 1, characterized in that: The generation of a personalized exercise program based on a rehabilitation assessment report by an artificial intelligence algorithm includes: Determine the patient's basic physical parameters; the basic physical parameters include age, height, weight, gender and past medical history; The output results of the rehabilitation assessment model and the basic physical parameters of the patient are input into the exercise program generation model as input data, and the exercise program generation model outputs the exercise program; The motion plan generation model adopts a reinforcement learning algorithm.

9. The functional behavior-based lower extremity arterial occlusive disease assessment and rehabilitation system according to claim 8, characterized in that: The reinforcement learning algorithm includes the definition of state space, action space and reward function; wherein, The output results of the rehabilitation assessment model and the basic physical parameters of the patient are defined as a state space; The action space is defined as a combination of different exercise programs, wherein the exercise program includes exercise type, exercise intensity and exercise duration; the exercise type includes Berger exercise, gait training, resistance training and dynamic stretching; A reward function is designed based on the motion data fed back by the patient; the motion data includes functional behavior data, physiological parameter data, and motion posture data.

Images