Intelligent orthopedic rehabilitation self-adaptive evaluation method based on multi-modal physiological feedback

Abstract

The invention discloses an intelligent orthopedic rehabilitation self-adaptive evaluation method and device based on multi-modal physiological feedback. The method comprises the steps that multi-modal physiological data including electromyographic signals, knee joint three-dimensional motion trails, plantar pressure distribution and cognitive input degree of a subject are synchronously collected; constructing a three-dimensional baseline vector according to the MMT muscle force test, knee joint ROM measurement and VAS pain score of the subject; inputting the multi-modal physiological data and the three-dimensional baseline vector into a personalized orthopedic rehabilitation model, and outputting a knee joint function index prediction value and personalized rehabilitation adjustment parameters; dynamically adjusting electrical stimulation parameters according to rehabilitation adjustment parameters output by the personalized orthopedic rehabilitation model; if the cognitive load index exceeds the early warning line, the electrical stimulation intensity is reduced; otherwise, activating the hippocampus body wave stimulation mode. Adjustment is carried out according to the real-time state of the subject, blindness and subjectivity in the rehabilitation process are avoided, and the rehabilitation efficiency and success rate are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of knee joint function reconstruction and rehabilitation, and particularly relates to an intelligent orthopedic rehabilitation adaptive evaluation method, device, and computing device based on multi-modal physiological feedback. Background Art

[0002] In the field of orthopedic rehabilitation, traditional rehabilitation evaluation and treatment methods mainly rely on doctors' experience, lack comprehensive analysis of multi-modal data, and the evaluation results are not comprehensive and accurate enough. Since most evaluation methods are static and cannot monitor and adjust the dynamic changes during the rehabilitation process in real time, the rehabilitation effect is not good. In addition, existing methods usually ignore the cognitive load and attention dispersion problems of patients, resulting in an affected rehabilitation effect.

[0003] To solve the above problems, the present invention proposes an intelligent orthopedic rehabilitation adaptive evaluation method based on multi-modal physiological feedback, which is adjusted according to the real-time state of the subject to avoid blindness and subjectivity during the rehabilitation process and improve the rehabilitation efficiency and success rate. Summary of the Invention

[0004] In view of the above problems, the present invention provides an intelligent orthopedic rehabilitation adaptive evaluation method, device, and computing device based on multi-modal physiological feedback.

[0005] According to one aspect of the present invention, there is provided an intelligent orthopedic rehabilitation adaptive evaluation method based on multi-modal physiological feedback, including:

[0006] Synchronously collecting multi-modal physiological data including the electromyogram signal, three-dimensional knee joint movement trajectory, plantar pressure distribution, and cognitive engagement of the subject; constructing a three-dimensional baseline vector according to the subject's MMT muscle strength test, knee joint ROM measurement, and VAS pain score;

[0007] Inputting the multi-modal physiological data and the three-dimensional baseline vector into a personalized orthopedic rehabilitation model to output a predicted value of the knee joint function index and personalized rehabilitation adjustment parameters; wherein, the rehabilitation adjustment parameters include electrical stimulation adjustment parameters;

[0008] Taking the predicted value of the knee joint function index, cognitive load index, and VAS pain score as the state space, and taking the electrical stimulation parameters as the action space, dynamically adjusting the electrical stimulation parameters according to the rehabilitation adjustment parameters output by the personalized orthopedic rehabilitation model;

[0009] Real-time monitoring the cognitive load index of the subject, wherein the cognitive load index is calculated according to the EEG band energy ratio; comparing the cognitive load index with a set attention dispersion warning line; if the cognitive load index exceeds the warning line, reducing the electrical stimulation intensity; otherwise, activating the hippocampal wave stimulation mode.

[0010] In an alternative approach, the method for synchronously collecting the multimodal physiological data further includes:

[0011] Collecting the raw electromyography signals of the quadriceps femoris of the subject through four-channel high-density surface electromyography electrodes;

[0012] Real-time calculating the three-dimensional motion trajectory of the knee joint through two IMU nodes on the thigh and the calf, wherein the three-dimensional motion trajectory of the knee joint includes flexion / extension angle, internal rotation / external rotation angle, and adduction / abduction angle;

[0013] Capturing the center-of-pressure offset trajectory during the stance phase / swing phase through a point-flexible pressure-sensing insole and constructing a plantar mechanical distribution heat map;

[0014] Monitoring the theta wave-alpha power ratio through prefrontal bipolar leads to quantify the cognitive engagement.

[0015] In an alternative approach, the method further includes:

[0016] Performing dynamic decoupling control on the electrical stimulation intensity, cognitive state, and pain level according to the three-dimensional state space transfer equation of the cognitive load-motion coupling coefficient; wherein, the three-dimensional state space transfer equation is:

[0017]

[0018] Wherein, H is the kinematic Hamiltonian of the knee joint; θ is the joint angle vector; CLIM is the cognitive load modulation index; ΔP stim is the adjustment amount of the electrical stimulation parameter; K P is the proportionality coefficient for adjusting the amplitude of the electrical stimulation adjustment; ROM is the range of motion of the joint; θ i+1 is the (i + 1)-th joint angle; is the integral of the reference power spectrum within the frequency range [α, β]; is the integral of the actual power spectrum within the frequency range [α, β]; VAS is the pain score; n is the dimension of the joint angle vector; ω is the independent variable of the power spectral density function, representing frequency.

[0019] In an alternative approach, the personalized orthopedic rehabilitation model includes a multimodal data preprocessing module, a biomechanical feature extraction module, a neurophysiological feature extraction module, a multimodal fusion module, and a prediction and optimization module;

[0020] Among them, the biomechanical feature extraction module includes a GCN network layer, a GAT network layer, and a ChebNets network layer, which are used to transform the knee joint biomechanical model into a graph structure to extract deep biomechanical features;

[0021] The neurophysiological feature extraction module includes a TCN network layer and an SE Blocks network block, and is used to capture the time dependence of EEG signals and EMG signals;

[0022] The prediction and optimization module includes a high-level policy layer, a low-level policy layer, an Actor network layer, a Critic network layer, an A2C network layer, and a MAML meta-learning module without relevance, and is used to learn the optimal electrical stimulation strategy.

[0023] In an optional manner, the cost function of the personalized orthopedic rehabilitation model is: J total =∫0 T [L biomechanical (t)+L neurophysiological (t)+L clinical (t)]dt

[0024] +Φ coupling

[0025] where L biomechanical (t) is the biomechanical integral term; L neurophysiological (t) is the neurophysiological integral term; L clinical (t) is the coupling term; θ pred (t) is the predicted knee joint movement trajectory; θ target (t) is the target knee joint movement trajectory; β is the biomechanical memory coefficient; τ is the time decay constant; H is the Hamiltonian; θ(s) is the knee joint movement trajectory at time s; θ α (t); θ α (t) is the power of the EEG signal in the alpha frequency band; θ β (t) is the power of the EEG signal in the beta frequency band; γ ref is the reference EEG frequency band ratio; μ is the cumulative change weight of the electromyogram signal; EMG env (s) is the electromyogram signal envelope at time s; VAS(t) is the visual analogue pain score at time t; ROM baseline is the baseline joint range of motion; CLI(s) is the cognitive load index at time s; CLI th is the threshold of the cognitive load index; ΔP stim (t) is the change amount of the electrical stimulation parameter at time t; K P is the proportional gain coefficient; CLIM(t) is the cognitive load modulation index at time t.

[0027] In an alternative approach, the personalized orthopedic rehabilitation model further includes an adaptive meta - learning module, which realizes the rapid adaptation of patient - specific parameters through the MAML algorithm;

[0028] Among them, in the initial training stage, the general kinetic characteristics of the rehabilitation process are extracted according to cross - patient meta - learning;

[0029] For new patients, the secondary rapid adaptation method is activated and the prediction error of the personalized orthopedic rehabilitation model is reduced to less than 60% of the baseline level through 3 - 5 iterations.

[0030] In an alternative approach, the GCN network layer, GAT network layer, and ChebNets network layer respectively extract node features and edge features in the knee joint biomechanical graph structure through spectral convolution, attention network layer, and Chebyshev polynomial approximation network layer, and fuse them to characterize the mechanical transmission and interaction during the knee joint movement process.

[0031] In an alternative approach, the high - level policy layer is trained according to the A2C algorithm to learn long - term electrical stimulation strategies;

[0032] The low - level policy layer rapidly adapts to individual differences according to the MAML algorithm;

[0033] The Actor network layer and Critic network layer are used to estimate the state value function and policy function.

[0034] According to another aspect of the present invention, there is provided an intelligent orthopedic rehabilitation adaptive evaluation device based on multi - modal physiological feedback, including:

[0035] A multi - modal data acquisition and baseline construction module, which is used to synchronously collect multi - modal physiological data including the electromyogram signal, three - dimensional knee joint movement trajectory, plantar pressure distribution, and cognitive engagement of the subject; construct a three - dimensional baseline vector according to the subject's MMT muscle strength test, knee joint ROM measurement, and VAS pain score;

[0036] A personalized rehabilitation model prediction and adjustment module, which is used to input the multi - modal physiological data and the three - dimensional baseline vector into the personalized orthopedic rehabilitation model, and output the predicted value of the knee joint function index and personalized rehabilitation adjustment parameters; among them, the rehabilitation adjustment parameters include electrical stimulation adjustment parameters;

[0037] An adaptive electrical stimulation parameter adjustment module, which uses the predicted value of the knee joint function index, cognitive load index, and VAS pain score as the state space, and the electrical stimulation parameters as the action space, and dynamically adjusts the electrical stimulation parameters according to the rehabilitation adjustment parameters output by the personalized orthopedic rehabilitation model;

[0038] A cognitive load monitoring and hippocampal wave stimulation module is used to monitor the cognitive load index of a subject in real time. Among them, the cognitive load index is calculated based on the EEG band energy ratio; the cognitive load index is compared with a set attention dispersion warning line; if the cognitive load index exceeds the warning line, the electrical stimulation intensity is reduced; otherwise, the hippocampal wave stimulation mode is activated.

[0039] According to another aspect of the present invention, a computing device is provided, including: a processor, a memory, a communication interface, and a communication bus. The processor, the memory, and the communication interface complete communication with each other through the communication bus;

[0040] The memory is used to store at least one executable instruction, and the executable instruction causes the processor to execute the operations corresponding to the above-mentioned intelligent orthopedic rehabilitation adaptive evaluation method based on multimodal physiological feedback.

[0041] The solution provided by the present invention includes: synchronously collecting multimodal physiological data including the electromyogram signal, three-dimensional knee joint movement trajectory, plantar pressure distribution, and cognitive engagement of the subject; constructing a three-dimensional baseline vector based on the subject's MMT muscle strength test, knee joint ROM measurement, and VAS pain score; inputting the multimodal physiological data and the three-dimensional baseline vector into a personalized orthopedic rehabilitation model to output a predicted value of the knee joint function index and personalized rehabilitation adjustment parameters; among them, the rehabilitation adjustment parameters include electrical stimulation adjustment parameters; using the predicted value of the knee joint function index, cognitive load index, and VAS pain score as the state space, and the electrical stimulation parameters as the action space, dynamically adjusting the electrical stimulation parameters according to the rehabilitation adjustment parameters output by the personalized orthopedic rehabilitation model; real-time monitoring the cognitive load index of the subject, where the cognitive load index is calculated based on the EEG band energy ratio; comparing the cognitive load index with a set attention dispersion warning line; if the cognitive load index exceeds the warning line, the electrical stimulation intensity is reduced; otherwise, the hippocampal wave stimulation mode is activated. The present invention adjusts according to the real-time state of the subject, avoids blindness and subjectivity in the rehabilitation process, and improves the rehabilitation efficiency and success rate.

[0042] The above description is only an overview of the technical solution of the present invention. In order to be able to understand the technical means of the present invention more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features, and advantages of the present invention more obvious and understandable, the following specifically describes the specific embodiments of the present invention. Brief Description of the Drawings

[0043] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the following detailed description of the preferred embodiments. The drawings are only for the purpose of showing the preferred embodiments and are not considered to be a limitation of the present invention. Also, throughout the drawings, the same reference numerals are used to represent the same components. In the drawings:

[0044] Figure 1 A schematic flowchart of an intelligent orthopedic rehabilitation adaptive evaluation method based on multi-modal physiological feedback according to an embodiment of the present invention is shown;

[0045] Figure 2 A schematic diagram of a knee joint rehabilitation device according to an embodiment of the present invention is shown Figure 1 ;

[0046] Figure 3 A schematic diagram of a knee joint rehabilitation device according to an embodiment of the present invention is shown Figure 2 ;

[0047] Figure 4 A schematic diagram of the knee joint rehabilitation effect according to an embodiment of the present invention is shown Figure 1 ;

[0048] Figure 5 A schematic diagram of the knee joint rehabilitation effect according to an embodiment of the present invention is shown Figure 2 ;

[0049] Figure 6 A schematic framework diagram of an intelligent orthopedic rehabilitation adaptive evaluation device based on multi-modal physiological feedback according to an embodiment of the present invention is shown;

[0050] Figure 7 A schematic structural diagram of a computing device according to an embodiment of the present invention is shown. Detailed Embodiments

[0051] The exemplary embodiments of the present invention will be described in more detail below with reference to the drawings. Although the exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present invention can be more thoroughly understood and the scope of the present invention can be completely conveyed to those skilled in the art.

[0052] Figure 1 A schematic flowchart of an intelligent orthopedic rehabilitation adaptive evaluation method based on multi-modal physiological feedback according to an embodiment of the present invention is shown. Specifically, as Figure 1 shown, the method includes the following steps:

[0053] Step S101, synchronously collect multi-modal physiological data including the myoelectric signals, three-dimensional knee joint movement trajectories, plantar pressure distributions, and cognitive engagement levels of the subject; construct a three-dimensional baseline vector based on the subject's MMT muscle strength test, knee joint ROM measurement, and VAS pain score.

[0054] In this embodiment, by synchronously collecting multi-modal physiological data including myoelectric signals, three-dimensional knee joint movement trajectories, plantar pressure distributions, and cognitive engagement levels, the limitations of single-modal data are avoided. Constructing a three-dimensional baseline vector based on the subject's MMT muscle strength test, knee joint ROM measurement, and VAS pain score provides a personalized rehabilitation evaluation basis for each subject, improving the pertinence of the rehabilitation plan. And by constructing a three-dimensional baseline vector to monitor and dynamically adjust various parameters during the rehabilitation process in real time, the scientific nature of the rehabilitation evaluation is improved.

[0055] Specifically, as Figure 2 、 Figure 3 shown, collect the raw myoelectric signals of the quadriceps femoris of the subject through four-channel high-density surface myoelectric electrodes to monitor muscle activity. Real-time solve the three-dimensional knee joint movement trajectory (including flexion / extension angle, internal rotation / external rotation angle, and adduction / abduction angle) through two IMU nodes on the thigh and calf, and then reflect the movement state of the knee joint. Capture the pressure center offset trajectory during the stance phase / swing phase through a point flexible pressure sensing insole, and construct a plantar mechanical distribution heat map to evaluate the gait and mechanical distribution of the subject. Monitor the ratio of theta wave to alpha power through prefrontal bipolar leads to quantify the cognitive engagement level of the subject and monitor the cognitive state during the rehabilitation process. Then, evaluate the muscle strength of the subject through a muscle strength test, evaluate the knee joint range of motion of the subject through a joint range of motion measurement, evaluate the pain level of the subject through a pain score, and integrate the above data (i.e., the subject's MMT muscle strength test, knee joint ROM measurement, and VAS pain score) to construct a three-dimensional baseline vector.

[0056] In an alternative manner, the method for synchronously collecting the multi-modal physiological data further includes:

[0057] Collect the raw myoelectric signals of the quadriceps femoris of the subject through four-channel high-density surface myoelectric electrodes;

[0058] Real-time solve the three-dimensional knee joint movement trajectory through two IMU nodes on the thigh and calf, where the three-dimensional knee joint movement trajectory includes flexion / extension angle, internal rotation / external rotation angle, and adduction / abduction angle;

[0059] Capture the pressure center offset trajectory during the stance phase / swing phase through a point flexible pressure sensing insole and construct a plantar mechanical distribution heat map;

[0060] Monitor the ratio of theta wave to alpha power through prefrontal bipolar leads to quantify the cognitive engagement level.

[0061] In this embodiment, the electromyography signal reflects the excitatory level of the muscle, the motion trajectory reflects the motion range and angle of the joint, the plantar pressure distribution reflects the body balance and support condition, and the cognitive engagement reflects the degree of the patient's concentration. The data of each of the above modalities complement each other to more accurately evaluate the patient's rehabilitation status. The four-channel high-density surface electromyography electrode can more precisely capture the muscle activity pattern of the quadriceps femoris and distinguish the contributions of different muscles. The IMU (Inertial Measurement Unit) calculates the three-dimensional motion trajectory of the knee joint in real time, avoiding the limitations of the traditional optical motion capture system and enabling measurement in a natural environment. The dot flexible pressure sensing insole precisely captures the plantar pressure distribution. The prefrontal bipolar lead monitors the theta wave to alpha power ratio as an objective way to quantify cognitive engagement, which can avoid subjective deviation. All sensors are non-invasive, causing no pain or discomfort to the patient, improving the patient's acceptance and compliance. The above lightweight devices such as the IMU, pressure sensing insole, and surface electromyography electrode are convenient to carry and use and can be measured in various environments such as clinics, homes, and communities.

[0062] Specifically, place the four-channel high-density surface electromyography electrode (including electrode patches, wires, and electromyography signal collector) on the quadriceps femoris and fix it according to the standard electrode placement positions to ensure good contact between the electrode and the skin. Use straps or tapes to fix the IMU nodes on the thigh and calf respectively, ensuring that the sensors do not slide or move and perform initial calibration. Place the pressure sensing insole inside the shoe and cover the entire sole of the foot. Place the bipolar lead electroencephalogram electrode at the prefrontal position and fix it according to the standard electrode placement positions. Use a synchronous trigger signal or software synchronization mechanism to make all sensors start collecting data at the same time point, and set an appropriate sampling frequency to capture sufficient signal details (for example, the sampling frequency of the electromyography signal is usually 1000 Hz, the sampling frequency of the IMU is usually 100 Hz, and the sampling frequency of the electroencephalogram signal is usually 250 Hz).

[0063] Step S102: Input the multi-modal physiological data and the three-dimensional baseline vector into the personalized orthopedic rehabilitation model, and output the predicted value of the knee joint function index and the personalized rehabilitation adjustment parameters; wherein, the rehabilitation adjustment parameters include the electrical stimulation adjustment parameters.

[0064] In this embodiment, based on the multi-modal physiological data of the subject (electromyogram signal, three-dimensional knee joint movement trajectory, plantar pressure distribution, cognitive engagement) and the three-dimensional baseline vectors (MMT muscle strength test, knee joint ROM measurement, VAS pain score), the rehabilitation parameters can be dynamically adjusted according to the patient's real-time physiological feedback and clinical evaluation data, realizing adaptive rehabilitation, more precisely controlling the rehabilitation process, and improving the rehabilitation effect. For example, the preprocessed multi-modal physiological data and three-dimensional baseline vectors are input into the personalized orthopedic rehabilitation model to predict the patient's knee joint function index (60 points) and personalized electrical stimulation adjustment parameters (electrical stimulation intensity is 10 mA, frequency is 20 Hz, pulse width is 200 μs). According to the electrical stimulation adjustment parameters output by the model, the parameters of the electrical stimulation device are set to guide the patient to perform quadriceps femoris electrical stimulation training and corresponding rehabilitation exercises. The multi-modal physiological data of the patient is collected again and a clinical evaluation is conducted. The model predicts that the patient's knee joint function index is 65 points, indicating that the rehabilitation effect has improved. According to the newly output electrical stimulation adjustment parameters again (electrical stimulation intensity is 12 mA, frequency is 22 Hz, pulse width is 210 μs), the parameters of the electrical stimulation device are adjusted and the rehabilitation training continues.

[0065] In an alternative approach, the personalized orthopedic rehabilitation model includes a multi-modal data preprocessing module, a biomechanical feature extraction module, a neurophysiological feature extraction module, a multi-modal fusion module, and a prediction and optimization module;

[0066] Among them, the biomechanical feature extraction module includes a GCN network layer, a GAT network layer, and a ChebNets network layer, which are used to convert the knee joint biomechanical model into a graph structure to extract deep biomechanical features;

[0067] The neurophysiological feature extraction module includes a TCN network layer and an SE Blocks network block, which are used to capture the time dependence of EEG signals and EMG signals;

[0068] The prediction and optimization module includes a high-level policy layer, a low-level policy layer, an Actor network layer, a Critic network layer, an A2C network layer, and a MAML meta-learning module without relevance, which are used to learn the optimal electrical stimulation strategy.

[0069] In this embodiment, the EEG signal reflects the patient's neural activities, the EMG signal reflects the muscle activities, the joint angle is the range of flexion and extension angles of the knee joint, the gait data includes walking speed, step length, walking frequency, etc., the pain score is the degree of pain evaluated by the Visual Analogue Scale (VAS), and the subjective feedback of the patient during the rehabilitation process includes fatigue level and comfort level, etc. The multi-modal data preprocessing module preprocesses the EEG signal, EMG signal, joint angle, and gait data. The biomechanical feature extraction module converts the knee joint biomechanical model into a graph structure, and uses GCN, GAT, and ChebNets network layers to extract deep biomechanical features. The neurophysiological feature extraction module uses the TCN network layer to capture the time dependence of the EEG signal and EMG signal, and uses the SE Blocks network block to enhance the representation ability of important features. The multi-modal fusion module fuses the biomechanical features and neurophysiological features to generate a comprehensive feature representation. Through the high-level policy layer, low-level policy layer, Actor network layer, Critic network layer, A2C network layer, and MAML meta-learning module independent of MAML, the optimal electrical stimulation strategy shown in Table 1 is learned.

[0070] Table 1

[0071]

[0072] In an optional manner, the high-level policy layer is trained according to the A2C algorithm to learn the long-term electrical stimulation strategy;

[0073] The low-level policy layer quickly adapts to individual differences according to the MAML algorithm;

[0074] The Actor network layer and the Critic network layer are used to estimate the state value function and the policy function.

[0075] In this embodiment, by combining the Actor (policy) and Critic (value) networks, the A2C algorithm reduces the variance of policy updates by using the advantage function, making the learning process more stable and easier to learn long-term dependencies, which is crucial for the electrical stimulation policy because the correct stimulation needs to be coordinated in time to achieve the desired effect. Fast adaptation to individual differences (MAML, Model-Agnostic Meta-Learning) is a meta-learning algorithm that can quickly adapt the model to new tasks during training. In the context of the electrical stimulation policy, it can quickly adapt to the physiological differences, disease states, etc. of different patients without having to train from scratch, significantly reducing the time required for personalized adjustment for each patient. The Actor-Critic architecture simultaneously learns the optimal policy (through the Actor network) and the state value function (through the Critic network). Among them, the Critic network evaluates the quality of the current policy and feeds back the information to the Actor network, thus guiding the improvement of the policy, and the learning process is more efficient because the update of the policy is based on the estimation of the value function rather than simply relying on the reward signal.

[0076] Specifically, the A2C high-level policy layer takes the patient's current state as input and outputs a series of goals or instructions for low-level policies. For example, the high-level policy may decide to increase the muscle strength output by 20% within the next 10 seconds. It is trained using the A2C algorithm, where the reward signal is based on long-term goals (such as improving motor ability, reducing pain, etc.). The Actor network outputs the probability distribution of actions, and the Critic network outputs the value estimate of the state. The MAML low-level policy layer takes the goal instructions from the high-level policy, the patient's current state, and local observations (such as muscle activity, electrophysiological signals, etc.) as input and outputs specific electrical stimulation parameters (such as current intensity, pulse width, frequency). It is trained using the MAML algorithm. The training process of MAML includes two loops: the inner loop and the outer loop. In the inner loop (Inner Loop), a small amount of data from specific patients is used to fine-tune the low-level policy. In the outer loop (Outer Loop), the initial parameters of the model are updated so that it can quickly adapt to new patients in the inner loop. The Actor-Critic network layer learns the policy function and outputs actions (electrical stimulation parameters) according to the current state (input). The Critic network evaluates the state value function and outputs an estimate of the goodness or badness of the state according to the current state (input). For example, for a patient with a slow walking speed after a stroke, the high-level policy may instruct the low-level policy to increase the activity level of the thigh muscles during walking, especially during the swing phase. The low-level policy stimulates the thigh muscles according to the patient's electromyogram signals with current intensity and frequency to help the patient lift the leg. For a patient with gait asymmetry after a stroke, the high-level policy may instruct the low-level policy to adjust the stimulation strategy to balance the activity levels of the muscles on both sides. The low-level policy improves gait symmetry according to the patient's gait data and electromyogram signals.

[0077] In an alternative approach, the cost function of the personalized orthopedic rehabilitation model is: J total =∫0 T [L biomechanical (t)+L neurophysiological (t)+L clinical (t)]dt

[0078] +Φ coupling

[0079] where L biomechanical (t) is the biomechanical integral term; L neurophysiological (t) is the neurophysiological integral term; L clinical (t) is the coupling term; θ pred (t) is the predicted knee joint movement trajectory; θ target(t) is the target knee joint movement trajectory; β is the biomechanical memory coefficient; τ is the time decay constant; H is the Hamiltonian; θ(s) is the knee joint movement trajectory at time s; θ α (t); θ α (t) is the power of the EEG signal in the alpha frequency band; θ β (t) is the power of the EEG signal in the beta frequency band; γ ref is the reference EEG frequency band ratio; μ is the cumulative change weight of the electromyogram signal; EMG env (s) is the envelope of the electromyogram signal at time s; VAS(t) is the visual analogue pain score at time t; ROM baseline is the range of joint motion at baseline; CLI(s) is the cognitive load index at time s; CLI th is the threshold of the cognitive load index; ΔP stim (t) is the change amount of the electrical stimulation parameter at time t; K P is the proportional gain coefficient; CLIM(t) is the cognitive load modulation index at time t.

[0081] In this embodiment, rehabilitation is a dynamic process, and past treatment effects will affect future outcomes. By considering the cumulative effect and time dependence of the rehabilitation process in integral form, the time-dependent relationship can be captured. For example, the "memory effect" in biomechanics is considered, that is, the influence of past forces on the current joint movement trajectory will decay over time, but there is still a certain "memory". Therefore, it is more in line with the real situation of human movement (the rehabilitation effect is as Figure 4 、 shown). Neurophysiological feedback uses the frequency band ratio of electroencephalogram signals to reflect the state of neural activity. Since neural plasticity is very important during the rehabilitation process, the cognitive state and neuromuscular control ability of patients are evaluated by monitoring electroencephalogram signals, so as to adjust the treatment plan.

[0082] In an optional manner, the GCN network layer, the GAT network layer, and the ChebNets network layer respectively extract the node features and edge features in the knee joint biomechanical graph structure through spectral convolution, the attention network layer, and the Chebyshev polynomial approximation network layer, and fuse them to characterize the mechanical transmission and interaction during the knee joint movement process.

[0083] In this embodiment, the knee joint biomechanics is modeled as a graph structure, which can naturally express the connection relationships and mechanical transmission paths among the components of the knee joint (for example, bones, muscles, ligaments). Among them, the nodes of the graph represent the components of the knee joint (femur, tibia, patella, quadriceps femoris, hamstring, gastrocnemius, anterior cruciate ligament (ACL), posterior cruciate ligament (PCL), medial collateral ligament (MCL), lateral collateral ligament (LCL), articular cartilage), and the edges of the graph represent the connection relationships and mechanical transmission paths among the nodes (joint connections between bones, attachment points between muscles and bones, bones connected by ligaments, mechanical interactions between nodes). However, it is very difficult for traditional machine learning methods to effectively process the above complex structured data. Further, the Chebyshev polynomial is used to approximate the spectral graph convolution, avoiding complex Fourier transforms, thereby improving the computational efficiency and enabling faster training and inference.

[0084] In an alternative manner, the personalized orthopedic rehabilitation model further includes an adaptive meta-learning module, and the adaptive meta-learning module realizes the rapid adaptation of patient-specific parameters through the MAML algorithm;

[0085] Among them, in the initial training stage, general kinetic features of the rehabilitation process are extracted according to cross-patient meta-learning;

[0086] For new patients, activate the secondary rapid adaptation method and reduce the prediction error of the personalized orthopedic rehabilitation model to less than 60% of the baseline level through 3-5 iterations.

[0087] In this embodiment, the MAML (Model-Agnostic Meta-Learning) algorithm can quickly adapt to new tasks, and can transfer the general knowledge learned from a large amount of patient data to new patients through a small number of iterations. The model does not need to train from scratch for each patient. In addition, the MAM algorithm itself is model-agnostic, so it can be applied to various different model architectures, and the basic model most suitable for the orthopedic rehabilitation task can be flexibly selected.

[0088] Step S103: Using the predicted value of the knee joint function index, the cognitive load index, and the VAS pain score as the state space, and the electrical stimulation parameters as the action space, dynamically adjust the electrical stimulation parameters according to the rehabilitation adjustment parameters output by the personalized orthopedic rehabilitation model.

[0089] In this embodiment, based on the personalized orthopedic rehabilitation model, the electrical stimulation parameters are dynamically adjusted according to the knee joint function, cognitive load, and pain degree status of the patient, realizing true personalized treatment, while traditional electrical stimulation treatments usually use fixed parameters, ignoring the differences between patients.

[0090] Specifically, a state space is defined. The knee joint function index uses a scale such as WOMAC (Western Ontario and McMaster Universities Osteoarthritis Index) or other similar scales to evaluate the functional status of the knee joint and converts it into a numerical value. The cognitive load index refers to the mental effort that the patient needs to exert during the rehabilitation process, which can be evaluated using the NASA-TLX (Task Load Index) method. NASA-TLX is a subjective evaluation scale that measures cognitive load by evaluating mental demand, physical demand, temporal demand, performance, effort, and frustration levels. The VAS pain score uses a Visual Analog Scale (VAS) to let the patient subjectively rate the pain level (0 - 10). Then, an action space is defined, and the value ranges of each electrical stimulation parameter are combined into an action vector (frequency, pulse width, current intensity, duty cycle, electrode position).

[0091] In an alternative approach, the method further includes:

[0092] Performing dynamic decoupling control on the electrical stimulation intensity, cognitive state, and pain level according to the three-dimensional state space transfer equation of the cognitive load - motion coupling coefficient; wherein, the three-dimensional state space transfer equation is:

[0093]

[0094] Wherein, H is the kinematic Hamiltonian of the knee joint; θ is the joint angle vector; CLIM is the cognitive load modulation index; ΔP stim is the adjustment amount of the electrical stimulation parameter; K P is the proportionality coefficient used to adjust the amplitude of the electrical stimulation adjustment; ROM is the range of joint motion; θ i+1 is the (i + 1)-th joint angle; is the integral of the reference power spectrum within the frequency range [α, β]; is the integral of the actual power spectrum within the frequency range [α, β]; VAS is the pain score; n is the dimension of the joint angle vector; ω is the independent variable of the power spectral density function, representing frequency.

[0095] In this embodiment, the adjustment amount of electrical stimulation is closely related to knee kinematics, cognitive load, and pain level through a three-dimensional state space transfer equation, realizing the refined regulation of electrical stimulation parameters, avoiding the one-size-fits-all parameter setting in traditional methods, and being more in line with the individual differences of patients. Considering the impact of cognitive load on the rehabilitation effect, the electrical stimulation intensity and cognitive state are dynamically decoupled to avoid the increase in cognitive burden caused by excessive stimulation, thereby affecting the compliance and rehabilitation progress of patients. The Hamiltonian of knee kinematics is used to more comprehensively describe the motion state of the knee joint, making the adjustment of electrical stimulation parameters more biomechanically based.

[0096] Step S104: Real-time monitor the cognitive load index of the subject, where the cognitive load index is calculated based on the EEG band energy ratio; compare the cognitive load index with a set attention dispersion warning line; if the cognitive load index exceeds the warning line, reduce the electrical stimulation intensity; otherwise, activate the hippocampal wave stimulation mode.

[0097] In this embodiment, by real-time monitoring EEG data and calculating the cognitive load index, the cognitive state of the subject is tracked in real time, and the intervention measures are more accurate. The activation of the hippocampal wave stimulation mode promotes memory consolidation and the improvement of cognitive function. This embodiment is easy to combine with other physiological indicators (such as eye movement tracking, heart rate variability, etc.) to further improve the accuracy of cognitive load assessment.

[0098] The solution provided by the present invention includes: synchronously collecting multi-modal physiological data including the electromyogram signal, three-dimensional knee joint movement trajectory, plantar pressure distribution, and cognitive engagement of the subject; constructing a three-dimensional baseline vector based on the MMT muscle strength test, knee joint ROM measurement, and VAS pain score of the subject; inputting the multi-modal physiological data and the three-dimensional baseline vector into a personalized orthopedic rehabilitation model to output the predicted value of the knee joint function index and personalized rehabilitation adjustment parameters; where the rehabilitation adjustment parameters include electrical stimulation adjustment parameters; taking the predicted value of the knee joint function index, cognitive load index, and VAS pain score as the state space and the electrical stimulation parameters as the action space, dynamically adjusting the electrical stimulation parameters according to the rehabilitation adjustment parameters output by the personalized orthopedic rehabilitation model; real-time monitoring the cognitive load index of the subject, where the cognitive load index is calculated based on the EEG band energy ratio; compare the cognitive load index with a set attention dispersion warning line; if the cognitive load index exceeds the warning line, reduce the electrical stimulation intensity; otherwise, activate the hippocampal wave stimulation mode. The present invention adjusts according to the real-time state of the subject, avoiding blindness and subjectivity in the rehabilitation process, and improving the rehabilitation efficiency and success rate.

[0099] Figure 5The schematic framework diagram of the intelligent orthopedic rehabilitation adaptive evaluation device based on multimodal physiological feedback according to an embodiment of the present invention is shown. The intelligent orthopedic rehabilitation adaptive evaluation device based on multimodal physiological feedback includes:

[0100] A multimodal data acquisition and baseline construction module 610, configured to synchronously acquire multimodal physiological data including the myoelectric signal, three-dimensional motion trajectory of the knee joint, plantar pressure distribution, and cognitive engagement of the subject; construct a three-dimensional baseline vector according to the MMT muscle strength test, knee joint ROM measurement, and VAS pain score of the subject;

[0101] A personalized rehabilitation model prediction and adjustment module 620, configured to input the multimodal physiological data and the three-dimensional baseline vector into a personalized orthopedic rehabilitation model, and output a predicted value of the knee joint function index and personalized rehabilitation adjustment parameters; wherein, the rehabilitation adjustment parameters include electrical stimulation adjustment parameters;

[0102] An adaptive electrical stimulation parameter adjustment module 630, configured to use the predicted value of the knee joint function index, cognitive load index, and VAS pain score as the state space, and the electrical stimulation parameters as the action space, and dynamically adjust the electrical stimulation parameters according to the rehabilitation adjustment parameters output by the personalized orthopedic rehabilitation model;

[0103] A cognitive load monitoring and hippocampal wave stimulation module 640, configured to monitor the cognitive load index of the subject in real time, wherein the cognitive load index is calculated according to the EEG band energy ratio; compare the cognitive load index with a set attention distraction warning line; if the cognitive load index exceeds the warning line, reduce the electrical stimulation intensity; otherwise, activate the hippocampal wave stimulation mode.

[0104] Figure 6 The schematic structural diagram of an embodiment of the computing device according to the present invention is shown. The specific embodiments of the present invention do not limit the specific implementation of the computing device.

[0105] As Figure 7 Figure 7 shown, the computing device may include: a processor 702, a communication interface 504, a memory 706, and a communication bus 708.

[0106] Wherein: the processor 702, the communication interface 704, and the memory 706 communicate with each other through the communication bus 708. The communication interface 704 is used to communicate with network elements of other devices such as clients or other servers. The processor 702 is configured to execute a program 710, and specifically may execute the relevant steps in the above-mentioned embodiment of the intelligent orthopedic rehabilitation adaptive evaluation method based on multimodal physiological feedback.

[0107] Specifically, the program 710 may include program code that includes computer operation instructions.

[0108] The processor 702 may be a central processing unit (CPU), or a specific integrated circuit (ASIC) (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of the present invention. One or more processors included in the computing device may be of the same type of processor, such as one or more CPUs; or may be of different types of processors, such as one or more CPUs and one or more ASICs.

[0109] The memory 706 is used to store the program 710. The memory 706 may include high-speed RAM memory and may also include non-volatile memory, such as at least one disk memory.

[0110] The solution provided by the present invention includes: synchronously collecting multi-modal physiological data including the myoelectric signals, three-dimensional motion trajectories of the knee joint, plantar pressure distribution, and cognitive engagement of the subject; constructing a three-dimensional baseline vector according to the subject's MMT muscle strength test, knee joint ROM measurement, and VAS pain score; inputting the multi-modal physiological data and the three-dimensional baseline vector into a personalized orthopedic rehabilitation model to output a predicted value of the knee joint function index and personalized rehabilitation adjustment parameters; wherein, the rehabilitation adjustment parameters include electrical stimulation adjustment parameters; taking the predicted value of the knee joint function index, the cognitive load index, and the VAS pain score as the state space, taking the electrical stimulation parameters as the action space, and dynamically adjusting the electrical stimulation parameters according to the rehabilitation adjustment parameters output by the personalized orthopedic rehabilitation model; real-time monitoring the cognitive load index of the subject, wherein the cognitive load index is calculated according to the EEG band energy ratio; comparing the cognitive load index with a set attention distraction warning line; if the cognitive load index exceeds the warning line, reducing the electrical stimulation intensity; otherwise, activating the hippocampal wave stimulation mode. The present invention adjusts according to the real-time state of the subject, avoids blindness and subjectivity in the rehabilitation process, and improves the rehabilitation efficiency and success rate.

[0111] Those skilled in the art can understand that the modules in the devices in the embodiments can be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or components in the embodiments can be combined into one module or unit or component, and in addition, they can be divided into multiple sub-modules or sub-units or sub-components. Except that at least some of such features and / or processes or units are mutually exclusive, any combination can be used to combine all the features disclosed in this specification (including the accompanying claims, abstract, and drawings) and all the processes or units of any method or device so disclosed. Unless otherwise explicitly stated, each feature disclosed in this specification (including the accompanying claims, abstract, and drawings) can be replaced by an alternative feature that provides the same, equivalent, or similar purpose. In addition, those skilled in the art can understand that although some of the embodiments herein include certain features included in other embodiments rather than other features, the combination of the features of different embodiments means that it is within the scope of the present invention and forms different embodiments. For example, in the following claims, any one of the claimed embodiments can be used in any combination. The present invention can be implemented by means of hardware including several different elements and by means of a properly programmed computer. In the unit claims listing several devices, several of these devices can be embodied by the same hardware item. The steps in the above embodiments, unless otherwise specified, should not be construed as a limitation on the execution order.

Claims

1. An intelligent orthopedic rehabilitation adaptive assessment method based on multimodal physiological feedback, used for assessing and optimizing knee joint function reconstruction, characterized in that, Including: Synchronously collect multimodal physiological data including the subject's electromyogram signal, three-dimensional knee joint movement trajectory, plantar pressure distribution, and cognitive engagement; construct a three-dimensional baseline vector based on the subject's MMT muscle strength test, knee joint ROM measurement, and VAS pain score; Input the multimodal physiological data and the three-dimensional baseline vector into a personalized orthopedic rehabilitation model, and output the predicted value of the knee joint function index and personalized rehabilitation adjustment parameters; wherein, the rehabilitation adjustment parameters include electrical stimulation adjustment parameters; Taking the predicted value of the knee joint function index, the cognitive load index, and the VAS pain score as the state space, and the electrical stimulation parameters as the action space, dynamically adjust the electrical stimulation parameters according to the rehabilitation adjustment parameters output by the personalized orthopedic rehabilitation model; Real-time monitor the subject's cognitive load index, wherein the cognitive load index is calculated based on the EEG band energy ratio; compare the cognitive load index with a set attention distraction warning line; if the cognitive load index exceeds the warning line, reduce the electrical stimulation intensity; otherwise, activate the hippocampal wave stimulation mode.

2. The intelligent orthopedic rehabilitation adaptive evaluation method based on multimodal physiological feedback according to claim 1, wherein The method for synchronously collecting the multimodal physiological data further includes: Collect the raw electromyogram signal of the quadriceps femoris of the subject through a four-channel high-density surface electromyogram electrode; Real-time calculate the three-dimensional knee joint movement trajectory through two IMU nodes on the thigh and calf, wherein the three-dimensional knee joint movement trajectory includes flexion / extension angle, internal rotation / external rotation angle, and adduction / abduction angle; Capture the pressure center offset trajectory during the stance phase / swing phase through a point-flexible pressure sensing insole and construct a plantar mechanical distribution heat map; Monitor the ratio of theta wave to alpha power through prefrontal bipolar leads to quantify cognitive engagement.

3. The intelligent orthopedic rehabilitation adaptive evaluation method based on multimodal physiological feedback according to claim 1, wherein The method further includes: Perform dynamic decoupling control on the electrical stimulation intensity, cognitive state, and pain level according to the three-dimensional state space transfer equation of the cognitive load-movement coupling coefficient; wherein, the three-dimensional state space transfer equation is: Among them, H is the Hamiltonian of knee joint kinematics; θ is the joint angle vector; CLIM is the cognitive load modulation index; ΔP stim is the adjustment amount of the electrical stimulation parameter; K P is the proportionality coefficient used to adjust the amplitude of the electrical stimulation adjustment; ROM is the range of motion of the joint; θ i+1 is the (i + 1)-th joint angle; is the integral of the reference power spectrum within the frequency range [α, β]; is the integral of the actual power spectrum within the frequency range [α, β]; VAS is the pain score; n is the dimension of the joint angle vector; ω is the independent variable of the power spectral density function, representing frequency.

4. The intelligent orthopedic rehabilitation adaptive evaluation method based on multimodal physiological feedback according to claim 1, wherein The personalized orthopedic rehabilitation model includes a multimodal data preprocessing module, a biomechanical feature extraction module, a neurophysiological feature extraction module, a multimodal fusion module, and a prediction and optimization module; Wherein, the biomechanical feature extraction module includes a GCN network layer, a GAT network layer, and a ChebNets network layer, and is used to transform the knee joint biomechanical model into a graph structure to extract deep biomechanical features; The neurophysiological feature extraction module includes a TCN network layer and an SE Blocks network block, and is used to capture the time dependence of EEG signals and EMG signals; The prediction and optimization module includes a high-level policy layer, a low-level policy layer, an Actor network layer, a Critic network layer, an A2C network layer, and a MAML meta-learning module independent of the model, and is used to learn the optimal electrical stimulation strategy.

5. The intelligent orthopedic rehabilitation adaptive evaluation method based on multi-modal physiological feedback according to claim 1, characterized in that The cost function of the personalized orthopedic rehabilitation model is: Among them, L biomechanical (t) is a biomechanical integral term; L neurophysiological (t) is a neurophysiological integral term; L clinical (t) is a coupling term; θ pred (t) is the predicted knee joint movement trajectory; θ target (t) is the target knee joint movement trajectory; β is the biomechanical memory coefficient; τ is the time decay constant; H is the Hamiltonian; θ(s) is the knee joint movement trajectory at time s; θ α (t); θ α (t) is the power of the EEG signal in the alpha frequency band; θ β (t) is the power of the EEG signal in the beta frequency band; γ ref is the reference EEG frequency band ratio; μ is the cumulative change weight of the electromyogram signal; EMG env (s) is the envelope of the electromyogram signal at time s; VAS(t) is the visual analogue pain score at time t; ROM baseline is the baseline range of joint motion; CLI(s) is the cognitive load index at time s; CLI th is the threshold of the cognitive load index; ΔP stim (t) is the change amount of the electrical stimulation parameter at time t; K P is the proportional gain coefficient; CLIM(t) is the cognitive load modulation index at time t.

6. The intelligent orthopedic rehabilitation adaptive evaluation method based on multimodal physiological feedback according to claim 1, wherein The personalized orthopedic rehabilitation model further includes an adaptive meta-learning module, and the adaptive meta-learning module realizes the rapid adaptation of patient-specific parameters through the MAML algorithm; Wherein, in the initial training stage, general dynamic features of the rehabilitation process are extracted according to cross-patient meta-learning; For new patients, activate the secondary rapid adaptation method and reduce the prediction error of the personalized orthopedic rehabilitation model to less than 60% of the baseline level through 3-5 iterations.

7. The intelligent orthopedic rehabilitation adaptive assessment method based on multimodal physiological feedback according to claim 4, wherein The GCN network layer, GAT network layer, and ChebNets network layer extract and fuse node features and edge features in the knee joint biomechanical graph structure through spectral convolution, attention network layer, and Chebyshev polynomial approximation network layer, respectively, to characterize the mechanical transmission and interaction during knee joint movement.

8. The intelligent orthopedic rehabilitation adaptive assessment method based on multimodal physiological feedback according to claim 4 or 5, characterized in that, The high-level policy layer is trained according to the A2C algorithm to learn long-term electrical stimulation strategies; The low-level policy layer quickly adapts to individual differences according to the MAML algorithm; The Actor network layer and Critic network layer are used to estimate the state value function and policy function.

9. An intelligent orthopedic rehabilitation adaptive evaluation device based on multimodal physiological feedback, characterized in that, Including: A multimodal data acquisition and baseline construction module, which is used to synchronously acquire multimodal physiological data including the electromyogram signal, three-dimensional knee joint movement trajectory, plantar pressure distribution, and cognitive engagement of the subject; construct a three-dimensional baseline vector based on the subject's MMT muscle strength test, knee joint ROM measurement, and VAS pain score; A personalized rehabilitation model prediction and adjustment module, which is used to input the multimodal physiological data and the three-dimensional baseline vector into the personalized orthopedic rehabilitation model, and output the predicted value of the knee joint function index and personalized rehabilitation adjustment parameters; among them, the rehabilitation adjustment parameters include electrical stimulation adjustment parameters; An adaptive electrical stimulation parameter adjustment module, which uses the predicted value of the knee joint function index, cognitive load index, and VAS pain score as the state space, and the electrical stimulation parameters as the action space, and dynamically adjusts the electrical stimulation parameters according to the rehabilitation adjustment parameters output by the personalized orthopedic rehabilitation model; A cognitive load monitoring and hippocampal wave stimulation module, which is used to monitor the cognitive load index of the subject in real time, where the cognitive load index is calculated based on the EEG band energy ratio; compare the cognitive load index with the set attention distraction warning line; if the cognitive load index exceeds the warning line, reduce the electrical stimulation intensity; otherwise, activate the hippocampal wave stimulation mode.

10. A computing device, comprising: A processor, a memory, a communication interface, and a communication bus, through which the processor, the memory, and the communication interface complete communication with each other; The memory is used to store at least one executable instruction, and the executable instruction causes the processor to execute the operations corresponding to the above-mentioned intelligent orthopedic rehabilitation adaptive evaluation method based on multimodal physiological feedback.

Images