An intelligent scoring method and system for joint movement

Abstract

The present application provides an intelligent scoring method and system for joint movement. Among them, inertial motion data with six degrees of freedom and surface electromyogram signals are collected, and a motion sequence topological structure including spatio-temporal benchmarks is established based on a dynamic sampling strategy; then the inertial data is decomposed into rigid body and physiological tremor components to generate a three-dimensional biomechanical model containing joint rotation center drift, multi-plane coupling error and trajectory curvature; then a time-varying mapping relationship between surface electromyogram signals and motion load is constructed, the signals are segmented based on phase mutation points and the phase synchronism of agonist and antagonist muscle groups is analyzed to extract electromyogram co-contraction patterns; finally, through technologies such as spatio-temporal coupling and dynamic transfer functions, motion feature tensors are generated in a four-dimensional spatio-temporal coordinate system, and their evolution patterns are analyzed in real time to construct a cluster of scoring functions, and a composite score value is generated. The technical solution provided by the present application significantly improves the accuracy and reliability of joint movement quality assessment.

Description

Technical Field

[0001] This application relates to the technical field of intelligent scoring, and particularly to an intelligent scoring method and system for joint movement. Background Art

[0002] The diagnosis and rehabilitation assessment of joint diseases require accurate measurement of the movement state of joints and related muscle activities. To provide personalized treatment plans, doctors and technicians need a method that can accurately quantify the quality of joint movement in order to timely adjust the rehabilitation training plan.

[0003] Currently, joint movement analysis mainly relies on an optical motion capture system combined with surface electromyography (sEMG). These methods usually collect data using a fixed sampling frequency and analyze the movement characteristics of joints through standard biomechanical models. In addition, most of the existing scoring systems are based on fixed parameters or simple threshold comparisons, lacking detailed consideration of the dynamic changes during joint movement.

[0004] The existing technologies have limitations in dealing with complex joint movements. Especially, the recognition of subtle movement components such as physiological tremors is not accurate enough, and it is difficult to reflect the specific changes in each stage of joint movement in real time. At the same time, traditional methods have deficiencies in integrating surface electromyography signals and joint kinematic parameters, and cannot effectively capture the dynamic changes in muscle co - action and phase synchronization, thus affecting the accuracy of joint movement quality assessment. Summary of the Invention

[0005] This application provides an intelligent scoring method and system for joint movement to solve the problem in the prior art that it is difficult to reflect the specific changes in each stage of joint movement in real time.

[0006] In the first aspect, this application provides an intelligent scoring method for joint movement, including:

[0007] Collecting inertial motion data and surface electromyography signals of six degrees of freedom of the target joint movement chain, triggering a dynamic sampling strategy based on the joint movement phase, and establishing a motion sequence topological structure including spatio - temporal reference markers;

[0008] Decomposing the inertial motion data into rigid body motion components and physiological tremor components, and generating a three - dimensional biomechanical model including the drift amount of the instantaneous rotation center of the joint, multi - plane coupling error, and the curvature of the motion trajectory through dynamic inverse solution;

[0009] Constructing a time - varying mapping relationship between surface electromyography signals and motion load, asymmetric window - length segmentation of surface electromyography signals based on the phase mutation points of kinematic parameters, establishing a phase synchronization quantification model for the contraction waveforms of agonist muscles and the inhibition waveforms of antagonist muscles within the segmentation interval, and extracting the myoelectric co - action pattern across the motion cycle through the analysis of the energy gradient field of surface electromyography signals;

[0010] Spatially and temporally couple the joint instantaneous rotation center drift amount in the three-dimensional biomechanical model with the electromyogram co-contraction pattern, establish a differential constraint relationship between joint pose changes and electromyogram activation intensity through the kinematic chain dynamics transfer function, and generate a motion feature tensor in a four-dimensional spatio-temporal coordinate system;

[0011] Real-time analyze the evolution pattern of the motion feature tensor, and autonomously construct a cluster of scoring functions with a time-varying topological structure according to the coupling strength differences in the motion acceleration period, stable period, and deceleration period. Each sub-function is non-linearly superimposed through the kinematic chain energy transfer efficiency and the neural control stability index to generate a composite scoring value with the function of diagnosing motion quality.

[0012] Optionally, the spatially and temporally coupling the joint instantaneous rotation center drift amount in the three-dimensional biomechanical model with the electromyogram co-contraction pattern, establishing a differential constraint relationship between joint pose changes and electromyogram activation intensity through the kinematic chain dynamics transfer function, and generating a motion feature tensor in a four-dimensional spatio-temporal coordinate system includes:

[0013] Based on the kinematic chain dynamics transfer function, spatially and temporally align the instantaneous angular velocity covariant derivative of the joint instantaneous rotation center drift amount in the three-dimensional biomechanical model with the phase gradient of the electromyogram co-contraction pattern to generate a dynamic coupling equation; wherein, the constraint conditions of the dynamic coupling equation are jointly calibrated by the joint ligament tension distribution model and the time-domain attenuation rate of the surface electromyogram signal suppression waveform to ensure the physical dimension consistency of the kinematic parameters and electromyogram parameters;

[0014] In a four-dimensional spatio-temporal coordinate system, perform a tensor product operation on the output of the dynamic coupling equation and the propagation path of the joint multi-plane coupling error to generate an initial feature tensor field; wherein, the dimension of the tensor product operation is dynamically adjusted by the curvature change rate of the joint rotation center drift trajectory, so that the initial tensor field simultaneously encodes the kinematic chain energy propagation direction and the electromyogram phase synchrony;

[0015] Calculate the viscoelastic coefficient matrix of the kinematic chain by using the asymmetric distribution characteristic of the surface electromyogram signal energy gradient, and adaptively smooth-filter the initial tensor field based on the viscoelastic coefficient matrix to generate a motion feature tensor including the physiological hysteresis effect.

[0016] Optionally, the calculating the viscoelastic coefficient matrix of the kinematic chain by using the asymmetric distribution characteristic of the surface electromyogram signal energy gradient includes:

[0017] Construct an energy gradient covariance tensor based on the phase gradient distribution of the electromyogram co-contraction pattern, and perform anisotropic weighting on the covariance tensor through the curvature change rate of the joint rotation center drift trajectory in the three-dimensional biomechanical model to generate an initial set of viscoelastic parameters with the characteristics of the kinematic chain energy propagation direction;

[0018] Input the set of initial viscoelastic parameters into a convolution kernel function constructed based on the time-domain attenuation rate of the surface electromyogram signal suppression waveform, extract the frequency-domain response features corresponding to the joint ligament tension distribution model through a multi-scale sliding window, and perform parameter normalization processing in combination with the constraint conditions of the dynamic coupling equation to form a frequency-domain mapping relationship of the viscoelastic coefficient matrix;

[0019] Perform adaptive smoothing filtering on the initial tensor field based on the viscoelastic coefficient matrix to generate a motion feature tensor including physiological hysteresis effects, including:

[0020] Establish a bandwidth adjustment factor for the tensor field filter according to the frequency-domain mapping relationship, dynamically match the coupling weight between the energy propagation direction of the motion chain and the electromyogram phase synchrony using the asymmetric distribution characteristic of the surface electromyogram signal energy gradient, and generate an adaptive filtering kernel with a physiological hysteresis compensation effect;

[0021] Perform multi-channel parallel convolution operations on the initial feature tensor field through the adaptive filtering kernel, synchronously fuse the propagation path curvature parameters of the joint multi-plane coupling error, and iteratively update the energy distribution topological structure of the motion feature tensor to generate a motion feature tensor including physiological hysteresis effects.

[0022] Optionally, analyze the evolution mode of the motion feature tensor in real time, and autonomously construct a cluster of scoring functions with a time-varying topological structure according to the coupling strength differences in the motion acceleration period, stable period, and deceleration period. Each sub-function is non-linearly superimposed through the motion chain energy transfer efficiency and the neural control stability index to generate a composite scoring value with a motion quality diagnosis function, including:

[0023] In the motion feature tensor, according to the phase boundary markers output by the motion stage recognition module, extract the distribution of manifold curvature extreme points along the energy propagation path, and calculate the covariance matrix of the manifold curvature change rate and the electromyogram phase synchronization stability within each motion stage to generate an energy dissipation spatio-temporal map; wherein, the detection rule of the manifold curvature extreme points is dynamically defined by the amplitude-frequency characteristics of the joint multi-plane coupling error.

[0024] In the neural control time-series feature space, map the abnormal regions in the energy dissipation spatio-temporal map to the time-frequency overlap interval of the electromyogram suppression waveform, and generate a set of sub-functions with variable-dimensional weights based on the hysteresis effect parameters output by the motion chain viscoelastic correction model; wherein, the topological structure of the sub-functions is dynamically constructed by the Hausdorff dimension of the joint rotation center drift trajectory, and the coupling weight between the sub-functions is calculated by the mutual information entropy of the electromyogram phase delay and the motion chain energy propagation efficiency.

[0025] Integrate the output values of the sub - function set along the energy propagation path of the kinematic chain through a topological integration algorithm to generate a composite score value reflecting joint movement coordination; at the same time, compare the composite score value with the physiological consistency threshold of the motion feature tensor, and dynamically adjust the iteration step size of the dynamic calibration factor through a back - propagation mechanism to achieve closed - loop optimization.

[0026] Optionally, in the neuro - control time - series feature space, map the abnormal regions in the energy - dissipation spatio - temporal atlas to the time - frequency overlapping interval of the electromyogram suppression waveform, and generate a sub - function set with variable - dimension weights based on the hysteresis effect parameters output by the kinematic - chain visco - elastic correction model, including:

[0027] Construct a time - frequency mask matrix based on the boundary features of the abnormal regions in the energy - dissipation spatio - temporal atlas, and perform phase - delay compensation on the mask matrix through the hysteresis effect parameters of the kinematic - chain visco - elastic correction model to generate a time - frequency overlapping feature vector with constraints on the energy propagation direction of the kinematic chain;

[0028] Input the time - frequency overlapping feature vector into a topological mapping network constructed based on the mutual information entropy between electromyogram phase delay and kinematic - chain energy propagation efficiency, and dynamically adjust the connection weights between network layers using the Hausdorff dimension of the joint rotation center drift trajectory to output an initial weight set of sub - functions containing variable - dimension parameters;

[0029] The generation of the sub - function set with variable - dimension weights includes:

[0030] Construct a multi - dimensional weight assignment function based on the initial weight set of sub - functions and the time - frequency decay characteristics of the electromyogram suppression waveform, and perform dynamic truncation filtering on the assignment function through the physiological consistency threshold of the motion feature tensor to generate a sub - function dimension adjustment factor that satisfies the kinematic - chain visco - elastic constraints;

[0031] Use the dimension adjustment factor to perform non - linear interpolation reconstruction on the output of the topological mapping network, and synchronously fuse the covariance matrix features of the manifold curvature change rate of the energy - dissipation spatio - temporal atlas to iteratively generate a sub - function set with a time - varying topological structure.

[0032] Optionally, it further includes:

[0033] Extract the topological invariance index of the energy propagation path in the motion feature tensor, perform non - linear mapping with the phase synchronization stability of the electromyogram co - pattern to generate a dynamic calibration factor;

[0034] Inject the dynamic calibration factor feedback into the constraint conditions of the dynamic coupling equation through an iterative optimization algorithm until the energy propagation path of the motion feature tensor and the time - frequency distribution of the electromyogram suppression waveform reach a preset physiological consistency threshold.

[0035] Optionally, for constructing the time-varying mapping relationship between the surface electromyogram signal and the exercise load, the surface electromyogram signal is segmented with an asymmetric window length based on the phase mutation points of the kinematic parameters. A phase synchronization quantization model of the contraction waveform of the active muscle group and the inhibition waveform of the antagonist muscle group is established within the segmented interval, and the electromyogram coordination pattern across the exercise cycle is extracted through the analysis of the energy gradient field of the surface electromyogram signal, including:

[0036] Construct a dynamic window length adjustment factor based on the detection result of the phase mutation point of the kinematic parameter. Extract the time-frequency energy mutation characteristics of the surface electromyogram signal at the phase mutation point through the wavelet transform modulus maximum detection algorithm, and generate a segmented index of the surface electromyogram signal with an asymmetric window length segmentation boundary;

[0037] Perform dynamic window division on the original surface electromyogram signal according to the segmented index of the surface electromyogram signal, and perform asymmetric time-domain alignment on the signal segments of the active muscle group and the antagonist muscle group by using the dynamic window length adjustment factor to generate a segmented interval containing the phase synchronization analysis reference point;

[0038] Within the segmented interval, use the Hilbert-Huang transform to extract the instantaneous phase difference sequences of the surface electromyogram signals of the active muscle group and the antagonist muscle group respectively. Combine the dynamic window length adjustment factor to construct a phase difference mutual information entropy matrix, and perform dynamic weighting on the entropy matrix through the curvature parameter of the energy propagation path of the motion chain to generate a phase synchronization index reflecting the neural control stability;

[0039] Construct an energy gradient covariance tensor based on the segmented index of the surface electromyogram signal within the segmented interval. Perform principal component decomposition on the phase synchronization index along the energy propagation direction of the motion chain, extract the principal component vector characterizing the cross-cycle coordination characteristics, and perform motion chain energy normalization processing on the principal component vector through the amplitude-frequency characteristics of the multi-plane coupling error in the three-dimensional biomechanical model to generate a feature set corresponding to the neuromuscular coordination pattern.

[0040] In a second aspect, the present application provides a joint motion intelligent scoring system, including:

[0041] An establishment module, configured to collect inertial motion data and surface electromyogram signals of six degrees of freedom of the target joint motion chain, and establish a motion sequence topological structure including spatio-temporal reference markers based on the joint motion phase to trigger a dynamic sampling strategy;

[0042] A generation module, configured to decompose the inertial motion data into a rigid body motion component and a physiological tremor component, and generate a three-dimensional biomechanical model including the instantaneous rotation center drift amount of the joint, the multi-plane coupling error, and the curvature of the motion trajectory through dynamic inverse solution;

[0043] A processing module, configured to construct a time-varying mapping relationship between surface electromyography signals and exercise loads, perform asymmetric window length segmentation on the surface electromyography signals based on phase mutation points of kinematic parameters, establish a phase synchronization quantization model for the contraction waveforms of agonist muscle groups and the inhibition waveforms of antagonist muscle groups within the segmented intervals, and extract electromyographic coordination patterns across exercise cycles through surface electromyography signal energy gradient field analysis;

[0044] The generation module is further configured to perform spatio-temporal coupling on the joint instantaneous rotation center drift amount in the three-dimensional biomechanical model and the electromyographic coordination pattern, establish a differential constraint relationship between joint pose changes and electromyographic activation intensity through a kinematic chain dynamics transfer function, and generate a motion feature tensor in a four-dimensional spatio-temporal coordinate system; analyze the evolution pattern of the motion feature tensor in real time, and autonomously construct a cluster of scoring functions with a time-varying topological structure according to the coupling intensity differences in the motion acceleration period, stable period, and deceleration period, where each sub-function is non-linearly superimposed through the kinematic chain energy transfer efficiency and the neural control stability index to generate a composite score value with a motion quality diagnosis function.

[0045] In a third aspect, the present application provides a computing device, including a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a joint motion intelligent scoring method as described in the first aspect above.

[0046] In a fourth aspect, the present application provides a computer storage medium storing a computer program, and when the computer program is executed by a computer, it implements a joint motion intelligent scoring method as described in the first aspect.

[0047] In this application, inertial motion data and surface electromyography (sEMG) signals of six degrees of freedom of the target joint kinematic chain are collected. Based on the joint motion phase-triggered dynamic sampling strategy, a motion sequence topological structure containing spatio-temporal reference markers is established. The inertial motion data is decomposed into rigid body motion components and physiological tremor components, and a three-dimensional biomechanical model containing the drift amount of the instantaneous joint rotation center, multi-plane coupling error, and motion trajectory curvature is generated through dynamic inverse solution. A time-varying mapping relationship between the sEMG signal and the motion load is constructed. Based on the phase mutation points of the kinematic parameters, the sEMG signal is segmented with an asymmetric window length. A phase synchronization quantification model of the contraction waveform of the agonist muscle group and the inhibition waveform of the antagonist muscle group is established within the segmented interval, and the myoelectric synergy pattern across the motion cycle is extracted through the analysis of the sEMG signal energy gradient field. The drift amount of the instantaneous joint rotation center in the three-dimensional biomechanical model is spatiotemporally coupled with the myoelectric synergy pattern, and a differential constraint relationship between the joint pose change and the myoelectric activation intensity is established through the kinematic chain dynamics transfer function, generating a motion feature tensor in the four-dimensional spatio-temporal coordinate system. The evolution pattern of the motion feature tensor is analyzed in real time. According to the coupling strength differences in the motion acceleration period, stable period, and deceleration period, a cluster of scoring functions with a time-varying topological structure is autonomously constructed. Each sub-function is non-linearly superimposed through the motion chain energy transfer efficiency and the neural control stability index to generate a composite score value with the function of diagnosing the motion quality.

[0048] The technical solution of this application has the following beneficial effects:

[0049] By combining inertial motion data of six degrees of freedom and sEMG signals, not only the mechanical motion characteristics of the joint are considered, but also the electrophysiological characteristics of muscle activities are deeply analyzed, realizing the detailed quantification of the complex dynamic behavior during joint movement. Using dynamic inverse solution technology and a three-dimensional biomechanical model, the subtle changes in joint movement can be accurately captured. In addition, through the asymmetric window length segmentation and phase synchronization quantification of the sEMG signal, the muscle synergy can be more realistically reflected, thereby improving the scoring accuracy. Finally, the composite score value generated based on these advanced analysis methods helps to achieve the accurate diagnosis of joint motion quality.

[0050] Furthermore, the method of the present invention involves spatio-temporally coupling the amount of instantaneous rotation center drift of a joint in a three-dimensional biomechanical model with an electromyographic co-contraction pattern through a kinematic chain dynamics transfer function to generate a dynamic coupling equation. The constraint conditions of this equation are jointly determined by the joint ligament tension distribution and the time-domain attenuation rate of the surface electromyographic signal suppression waveform to ensure parameter consistency. In a four-dimensional spatio-temporal coordinate system, the output of the joint multi-plane coupling error propagation path is combined with the dynamic coupling equation through tensor product operation to generate an initial eigen-tensor field. The dimension of this field is adjusted according to the curvature change rate of the joint rotation center drift trajectory, while encoding the energy propagation direction and electromyographic phase synchronization. Further, the viscoelastic coefficient matrix is calculated using the asymmetric distribution characteristic of the surface electromyographic signal energy gradient, and the initial tensor field is adaptively smoothed and filtered based on this matrix to finally generate a motion eigen-tensor field that includes physiological hysteresis effects. This method solves the problems in traditional technologies of being difficult to accurately capture the subtle joint movements and muscle co-contraction through precise spatio-temporal coupling and dynamic adjustment mechanisms, effectively improving the accuracy and reliability of joint movement quality assessment, especially being able to more realistically reflect the energy propagation characteristics and electromyographic phase synchronization of the kinematic chain during complex joint movements.

[0051] Furthermore, the method of the present invention utilizes the asymmetric distribution characteristic of the surface electromyographic signal energy gradient. First, an energy gradient covariance tensor is constructed based on the phase gradient distribution of the electromyographic co-contraction pattern and anisotropically weighted by the curvature change rate of the joint rotation center drift trajectory to generate an initial set of viscoelastic parameters with the characteristics of the kinematic chain energy propagation direction. Then, this parameter set is input into a convolution kernel function constructed by the time-domain attenuation rate of the surface electromyographic signal suppression waveform. The frequency-domain response characteristics are extracted through a multi-scale sliding window and parameter normalization processing to form the frequency-domain mapping relationship of the viscoelastic coefficient matrix. Then, a bandwidth adjustment factor is established according to this frequency-domain mapping relationship, and the coupling weights are dynamically matched in combination with the asymmetric distribution characteristic of the surface electromyographic signal energy gradient to generate an adaptive filtering kernel. Finally, this filtering kernel is used to perform multi-channel parallel convolution operations on the initial eigen-tensor field, synchronously fusing the curvature parameters of the joint multi-plane coupling error propagation path, and iteratively updating the energy distribution topological structure of the motion eigen-tensor field to generate a motion eigen-tensor field that includes physiological hysteresis effects. This method effectively compensates for the physiological hysteresis effects ignored in traditional analysis methods during complex joint movement processing through precise calculation and application of the viscoelastic coefficient matrix, combined with adaptive smoothing filtering technology, significantly improving the capture accuracy of subtle joint movements and muscle co-contraction, and thus enhancing the authenticity and reliability of joint movement quality assessment.

[0052] These aspects or other aspects of the present application will be more clearly understood in the following description of the embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] To more clearly illustrate the technical solutions in the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0054] Figure 1 The flowchart of an intelligent joint movement scoring method provided by the present application is shown;

[0055] Figure 2 The structural schematic diagram of an intelligent joint movement scoring system provided by the present application is shown;

[0056] Figure 3 The structural schematic diagram of a computing device provided by the present application is shown. Detailed implementation manners

[0057] In order to enable those skilled in the art of the present technology to better understand the solutions of the present application, the following will clearly and completely describe the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application.

[0058] In some processes described in the specification and claims of the present application and the above-mentioned drawings, a plurality of operations that appear in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order in which they appear herein or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions such as "first" and "second" in this article are used to distinguish different messages, devices, modules, etc., and do not represent a sequence, nor do they limit that "first" and "second" are of different types.

[0059] The following will clearly and completely describe the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, rather than all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present application.

[0060] Figure 1 A flowchart of an intelligent joint movement scoring method is provided for the embodiments of the present application. As Figure 1 shown, the method includes:

[0061] 101. Collect the six-degree-of-freedom inertial motion data and surface electromyography signals of the target joint kinematic chain, trigger a dynamic sampling strategy based on the joint motion phase, and establish a motion sequence topological structure including spatio-temporal reference markers.

[0062] In this step, collect the six-degree-of-freedom inertial motion data and surface electromyography signals of the target joint kinematic chain. These data include information such as position, velocity, and acceleration, which are used to analyze the motion state of the joint and muscle activity, and trigger a dynamic sampling strategy based on the joint motion phase to establish a motion sequence topological structure including spatio-temporal reference markers.

[0063] In the embodiment of the present application, capture the six-degree-of-freedom motion data of the target joint and the related surface electromyography signals through sensors, and use a dynamic sampling strategy to trigger data collection at different frequencies according to different stages of joint motion, so as to construct a spatio-temporal marker topological structure that accurately reflects the entire process of joint motion.

[0064] Suppose it is necessary to evaluate the knee rehabilitation of an athlete. First, install inertial measurement units (IMUs) and surface electromyography (sEMG) sensors around the athlete's knee. When the athlete performs a series of specified actions, collect relevant data in real time and record the key spatio-temporal markers at each action stage for subsequent analysis.

[0065] 102. Decompose the inertial motion data into rigid body motion components and physiological tremor components, and generate a three-dimensional biomechanical model including the drift amount of the instantaneous rotation center of the joint, multi-plane coupling error, and motion trajectory curvature through dynamic inverse solution.

[0066] In this step, decompose the collected inertial motion data into rigid body motion components and physiological tremor components. The former describes the basic motion mode of the joint, and the latter reflects the subtle involuntary motion. Generate a three-dimensional biomechanical model including the drift amount of the instantaneous rotation center of the joint, multi-plane coupling error, and motion trajectory curvature through dynamic inverse solution technology.

[0067] In the embodiment of the present application, use a dynamic inverse solution method to separate the rigid body motion and physiological tremor components from the original data, and then calculate the change of the instantaneous rotation center of the joint during motion, the error generated by the interaction between planes, and the bending degree of the motion trajectory to form a detailed three-dimensional biomechanical model.

[0068] Continuing the above example, by processing the collected data, it is possible to separate the normal motion part of the athlete's knee during motion and the minor vibrations caused by incomplete rehabilitation, and accordingly establish a biomechanical model to quantify the recovery status.

[0069] 103. Construct the time-varying mapping relationship between surface electromyography (sEMG) signals and exercise loads. Segment the sEMG signals with asymmetric window lengths based on the phase mutation points of kinematic parameters. Establish a quantization model for the phase synchronization between the contraction waveforms of agonist muscles and the inhibition waveforms of antagonist muscles within the segmented intervals, and extract the myoelectric coordination patterns across exercise cycles through the analysis of the energy gradient field of sEMG signals.

[0070] In this step, constructing the time-varying mapping relationship between sEMG signals and exercise loads aims to identify the muscle activation patterns at different exercise stages. Segment the sEMG signals with asymmetric window lengths based on the phase mutation points of kinematic parameters to distinguish the contraction waveforms of agonist muscles and the inhibition waveforms of antagonist muscles, establish a quantization model for phase synchronization, and simultaneously extract the myoelectric coordination patterns across exercise cycles.

[0071] In the embodiments of this application, a specific algorithm is used to identify the phase mutation points in sEMG signals. Based on this, the window length is dynamically adjusted for signal segmentation. The behavior patterns of agonist muscles and antagonist muscles are analyzed separately, a phase synchronization index between them is constructed, and further research is conducted on how muscles coordinate during the entire exercise process.

[0072] In the above case, it can be determined which muscles are mainly working and which are assisting or inhibiting when an athlete performs a specific action. This helps to understand the cooperation efficiency between muscles and provides a basis for formulating more effective rehabilitation training.

[0073] 104. Spatially and temporally couple the joint instantaneous rotation center drift amount in the three-dimensional biomechanical model with the myoelectric coordination pattern. Establish a differential constraint relationship between joint pose changes and myoelectric activation intensity through the kinematic chain dynamics transfer function, and generate a motion feature tensor in a four-dimensional spatio-temporal coordinate system.

[0074] In this step, combine the joint instantaneous rotation center drift amount in the three-dimensional biomechanical model with the myoelectric coordination pattern. Establish a differential constraint relationship between joint pose changes and myoelectric activation intensity through the kinematic chain dynamics transfer function, and finally generate a motion feature tensor in a four-dimensional spatio-temporal coordinate system that comprehensively describes the joint motion characteristics.

[0075] In the embodiments of this application, utilize the existing three-dimensional biomechanical model and myoelectric coordination pattern, correlate the two through mathematical modeling, create a dynamic system that can accurately represent the relationship between joint position changes and muscle activity intensity, and express it in the form of a tensor in a four-dimensional spatio-temporal coordinate system for subsequent analysis.

[0076] In the example of an athlete, combine the motion characteristics of the knee with the muscle activity intensity to generate a comprehensive motion feature tensor, which enables the functional recovery of the knee to be observed from multiple dimensions.

[0077] 105. Analyze the evolution pattern of the motion feature tensor in real time, and autonomously construct a cluster of scoring functions with a time-varying topological structure according to the coupling strength differences among the motion acceleration period, the stable period, and the deceleration period. Each sub-function is non-linearly superimposed through the motion chain energy transfer efficiency and the neural control stability index to generate a composite scoring value with the function of diagnosing motion quality.

[0078] In this step, analyze the evolution pattern of the motion feature tensor in real time, and autonomously construct a cluster of scoring functions according to the different characteristics of the motion acceleration period, the stable period, and the deceleration period. Each sub-function realizes non-linear superposition by considering the motion chain energy transfer efficiency and the neural control stability, and finally generates a composite scoring value with the diagnostic function.

[0079] In the embodiments of the present application, through continuous monitoring and analysis of the motion feature tensor, the characteristics of different motion stages are identified, and the scoring criteria are automatically adjusted to ensure accurate evaluation of each stage. By integrating multiple evaluation indicators, a scoring system that comprehensively reflects the joint motion quality is formed.

[0080] For athletes, this process helps to give specific scores according to their performance in different motion stages, guiding their rehabilitation training to be more scientific and effective.

[0081] In summary, steps 101 to 105 cover a complete process from data collection, processing to analysis, aiming to provide a comprehensive and accurate intelligent joint motion scoring system to meet the needs of personalized medicine and rehabilitation, and improve the treatment effect and quality of life.

[0082] To solve the problem that traditional methods are difficult to accurately capture the physiological hysteresis effect during the joint motion process, in some embodiments, in step 104, the drift amount of the instantaneous rotation center of the joint in the three-dimensional biomechanical model is spatiotemporally coupled with the electromyography co-pattern, and a differential constraint relationship between the joint pose change and the electromyography activation intensity is established through the motion chain dynamics transfer function to generate a motion feature tensor in the four-dimensional spacetime coordinate system, including:

[0083] Based on the kinetic transfer function of the kinematic chain, the spatio-temporal alignment is performed between the instantaneous angular velocity covariant derivative of the instantaneous rotation center drift of the joint in the three-dimensional biomechanical model and the phase gradient of the electromyography co-contraction pattern to generate a dynamic coupling equation; wherein, the constraint conditions of the dynamic coupling equation are jointly calibrated by the joint ligament tension distribution model and the time-domain attenuation rate of the surface electromyography signal suppression waveform to ensure the physical dimension consistency of the kinematic parameters and the electromyography parameters; in the four-dimensional spatio-temporal coordinate system, the output of the dynamic coupling equation is subjected to a tensor product operation with the propagation path of the joint multi-planar coupling error to generate an initial eigen-tensor field; wherein, the dimension of the tensor product operation is dynamically adjusted by the curvature change rate of the joint rotation center drift trajectory, so that the initial tensor field encodes both the energy propagation direction of the kinematic chain and the electromyography phase synchronization; the viscoelastic coefficient matrix of the kinematic chain is calculated by using the asymmetric distribution characteristic of the surface electromyography signal energy gradient, and the initial tensor field is adaptively smoothed and filtered based on the viscoelastic coefficient matrix to generate a motion feature tensor including the physiological hysteresis effect.

[0084] In this embodiment, firstly, the concept of the kinetic transfer function of the kinematic chain is introduced, which is used to describe the mechanical behavior during joint movement and its relationship with muscle activity; secondly, a dynamic coupling equation is generated by combining the instantaneous rotation center drift of the joint and the electromyography co-contraction pattern, and this equation can reflect the change of the joint movement state; thirdly, a tensor product operation is adopted to combine the result of the dynamic coupling equation with the joint multi-planar coupling error to form an initial eigen-tensor field in the four-dimensional spatio-temporal coordinate system that can express the joint movement characteristics; finally, by analyzing the energy gradient of the surface electromyography signal, a viscoelastic coefficient matrix is constructed and applied to the adaptive smoothing filter to compensate for the physiological hysteresis effect.

[0085] In the embodiment of the present application, firstly, sensors are used to obtain joint movement data and surface electromyography signals; secondly, the rigid body movement and physiological tremor components are separated by dynamic inverse solution technology, and a three-dimensional biomechanical model is established; thirdly, the sampling strategy is adjusted according to different stages of joint movement to identify the behavior patterns of the agonist and antagonist muscle groups; then, in combination with the kinetic transfer function of the kinematic chain, the joint movement is associated with muscle activity to form a dynamic coupling equation; next, these information is integrated through a tensor product operation to generate a tensor field describing the joint movement characteristics; finally, the initial tensor field is processed by applying adaptive filtering technology to ensure that it can accurately reflect the physiological hysteresis effect.

[0086] The following is a specific example:

[0087] Suppose it is necessary to evaluate the recovery of a patient after knee surgery. First, an inertial measurement unit (IMU) and a surface electromyography (sEMG) sensor are installed around the patient's knee; the collected data is used to distinguish rigid body motion and physiological tremors, and a three-dimensional biomechanical model is established; subsequently, according to the patient's motion pattern, key kinematic parameter phase mutation points are identified, and the signal window length is adjusted for segmentation; next, the above information is used to generate a dynamic coupling equation, and its results are combined with the joint multi-plane coupling error to form an initial feature tensor field in a four-dimensional space-time coordinate system; finally, by analyzing the energy gradient of the surface electromyography signal, a viscoelastic coefficient matrix is calculated and applied to adaptive smoothing filtering to generate a motion feature tensor containing physiological hysteresis effects. Through the above steps, not only can the recovery status of the patient's knee be understood in detail, but also personalized rehabilitation suggestions can be provided for the patient, thus accelerating the rehabilitation process.

[0088] In order to further improve the capture accuracy of physiological hysteresis effects in joint motion analysis, in some embodiments, calculating the viscoelastic coefficient matrix of the motion chain using the asymmetric distribution characteristics of the surface electromyography signal energy gradient includes:

[0089] Constructing an energy gradient covariance tensor based on the phase gradient distribution of the myoelectric synergy pattern, anisotropically weighting the covariance tensor through the curvature change rate of the joint rotation center drift trajectory in the three-dimensional biomechanical model to generate an initial set of viscoelastic parameters with the characteristics of the energy propagation direction of the motion chain; inputting the initial set of viscoelastic parameters into a convolution kernel function constructed based on the time-domain decay rate of the surface electromyography signal suppression waveform, extracting the frequency-domain response characteristics corresponding to the joint ligament tension distribution model through a multi-scale sliding window, and performing parameter normalization processing in combination with the constraint conditions of the dynamic coupling equation to form a frequency-domain mapping relationship of the viscoelastic coefficient matrix; the adaptive smoothing filtering of the initial tensor field based on the viscoelastic coefficient matrix to generate a motion feature tensor containing physiological hysteresis effects includes: establishing a bandwidth adjustment factor for the tensor field filter according to the frequency-domain mapping relationship, dynamically matching the coupling weight of the energy propagation direction of the motion chain and the myoelectric phase synchronization using the asymmetric distribution characteristics of the surface electromyography signal energy gradient to generate an adaptive filtering kernel with a physiological hysteresis compensation effect; performing multi-channel parallel convolution operations on the initial feature tensor field through the adaptive filtering kernel, synchronously fusing the propagation path curvature parameters of the joint multi-plane coupling error, and iteratively updating the energy distribution topological structure of the motion feature tensor to generate a motion feature tensor containing physiological hysteresis effects.

[0090] In this embodiment, the concept of the energy gradient covariance tensor is first introduced, which is used to describe the energy change and its spatial distribution in the electromyographic activity; secondly, by considering the curvature change rate of the drift trajectory of the joint rotation center, the above tensor is anisotropically weighted to obtain an initial viscoelastic parameter set reflecting the characteristics of the energy propagation direction of the kinematic chain; thirdly, the convolution kernel function is designed using the time-domain decay rate of the suppressed waveform of the surface electromyographic signal, and the corresponding frequency-domain features are extracted by the multi-scale sliding window technique; finally, the adaptive filtering method is adopted to adjust the bandwidth of the tensor field filter based on the above information to achieve the precise compensation of the physiological hysteresis effect.

[0091] In the embodiment of the present application, first, the phase gradient distribution of the electromyographic coactivation pattern is used to construct the energy gradient covariance tensor; secondly, based on the curvature change rate of the drift trajectory of the joint rotation center, the tensor is anisotropically weighted to generate an initial viscoelastic parameter set; thirdly, these parameters are input into a convolution kernel function constructed by the time-domain decay rate of the suppressed waveform of the surface electromyographic signal, and the frequency-domain response features are extracted by the multi-scale sliding window and the parameter normalization is performed in combination with the constraint conditions of the dynamic coupling equation; then, the bandwidth adjustment factor of the tensor field filter is set according to the obtained frequency-domain mapping relationship; next, the filtering kernel is adjusted using the asymmetric distribution characteristic of the energy gradient of the surface electromyographic signal so that it can match the coupling weight of the energy propagation direction of the kinematic chain and the electromyographic phase synchrony; finally, the adaptive filter kernel is applied to perform multi-channel parallel convolution operations on the initial feature tensor field, and the curvature parameters of the propagation path of the multi-plane coupling error of the joint are synchronously integrated to generate the final motion feature tensor.

[0092] The following is a specific example:

[0093] Suppose it is necessary to evaluate the knee rehabilitation of an athlete. First, an inertial measurement unit (IMU) and a surface electromyogram (sEMG) sensor are installed around the athlete's knee; after collecting data, an energy gradient covariance tensor is constructed based on the phase gradient distribution of the myoelectric synergy pattern; subsequently, the covariance tensor is anisotropically weighted according to the curvature change rate of the drift trajectory of the joint rotation center to generate an initial viscoelastic parameter set; then, these parameters are input into a convolution kernel function constructed by the time-domain decay rate of the surface electromyogram signal suppression waveform, and frequency-domain response features are extracted through a multi-scale sliding window and parameter normalization processing; after that, a bandwidth adjustment factor of the tensor field filter is set according to the obtained frequency-domain mapping relationship; the filter kernel is adjusted using the asymmetric distribution characteristic of the surface electromyogram signal energy gradient so that it can match the coupling weight of the energy propagation direction of the motion chain and the myoelectric phase synchronization; finally, an adaptive filter kernel is used to perform multi-channel parallel convolution operations on the initial feature tensor field, and the curvature parameters of the propagation paths of the multi-plane coupling errors of the joint are synchronously integrated. Through the above steps, not only can the recovery status of the athlete's knee be understood in detail, but also personalized rehabilitation suggestions can be provided for the athlete, thus accelerating the rehabilitation process.

[0094] In order to further improve the accuracy and personalization of joint motion quality assessment, in some embodiments, the evolution mode of the motion feature tensor is parsed in real time, and a cluster of scoring functions with a time-varying topological structure is autonomously constructed according to the coupling strength differences in the motion acceleration period, stable period, and deceleration period, where each sub-function is non-linearly superimposed through the motion chain energy transfer efficiency and the neural control stability index to generate a composite score value with a motion quality diagnosis function, including:

[0095] In the motion feature tensor, according to the phase boundary markers output by the motion phase recognition module, the distribution of manifold curvature extreme points is extracted along the energy propagation path, and the covariance matrix of the manifold curvature change rate and the electromyogram phase synchronization stability within each motion phase is calculated to generate an energy dissipation spatio-temporal map; wherein, the detection rule of the manifold curvature extreme points is dynamically defined by the amplitude-frequency characteristics of the joint multi-plane coupling error; in the neural control time series feature space, the abnormal regions in the energy dissipation spatio-temporal map are mapped to the time-frequency overlapping interval of the electromyogram suppression waveform, and based on the hysteresis effect parameters output by the motion chain viscoelastic correction model, a set of sub-functions with variable dimensional weights is generated; wherein, the topological structure of the sub-functions is dynamically constructed by the Hausdorff dimension of the joint rotation center drift trajectory, and the coupling weights between the sub-functions are calculated by the mutual information entropy of the electromyogram phase delay and the motion chain energy propagation efficiency; the output values of the set of sub-functions are integrated along the motion chain energy propagation path through a topological integration algorithm to generate a composite score value reflecting the joint motion coordination; meanwhile, the composite score value is compared with the physiological consistency threshold of the motion feature tensor, and the iteration step size of the dynamic calibration factor is dynamically adjusted through a backpropagation mechanism to achieve closed-loop optimization.

[0096] In this embodiment, the concept of the energy dissipation spatio-temporal map is first introduced, which is used to describe the change of energy distribution during joint motion and its relationship with muscle activity synchronization; secondly, the amplitude-frequency characteristics of the joint multi-plane coupling error are used to dynamically define the detection rule of the manifold curvature extreme points, ensuring the accurate capture of fine joint motions; thirdly, by analyzing the relationship between the electromyogram suppression waveform and the time-frequency overlapping interval, combined with the motion chain viscoelastic correction model, the effective identification of motion abnormal regions is realized; finally, the above information is processed by a topological integration algorithm to generate a composite score value that can comprehensively reflect the joint motion coordination.

[0097] In the embodiment of the present application, first, the phase boundary markers of different motion phases (acceleration, stability, deceleration) are identified from the motion feature tensor; secondly, the distribution of manifold curvature extreme points is extracted along the energy propagation path, and the covariance matrix of its change rate and the electromyogram phase synchronization stability is calculated to form an energy dissipation spatio-temporal map; thirdly, in the neural control time series feature space, the abnormal regions in the energy dissipation spatio-temporal map are mapped to the time-frequency overlapping interval of the electromyogram suppression waveform, and based on the hysteresis effect parameters output by the motion chain viscoelastic correction model, a set of sub-functions with variable dimensional weights is generated; then, the coupling weights between the sub-functions are dynamically adjusted according to the Hausdorff dimension of the joint rotation center drift trajectory; finally, the outputs of these sub-functions are integrated through a topological integration algorithm to generate a composite score value reflecting the joint motion coordination, and closed-loop optimization is performed through a backpropagation mechanism.

[0098] The following is a specific example:

[0099] Suppose it is necessary to evaluate the recovery of an athlete after knee surgery. First, an inertial measurement unit (IMU) and a surface electromyogram (sEMG) sensor are installed around the knee; after collecting data, the phase boundary markers of different motion stages are identified; then, the distribution of the extreme points of the manifold curvature is extracted along the energy propagation path, and the covariance matrix of the manifold curvature change rate and the myoelectric phase synchronization stability within each motion stage is calculated to generate an energy dissipation spatio-temporal map; subsequently, in the neural control time series feature space, the abnormal region in the energy dissipation spatio-temporal map is mapped to the time-frequency overlapping interval of the myoelectric inhibition waveform, and based on the hysteresis effect parameters output by the viscoelastic correction model of the kinematic chain, a set of sub-functions with variable dimensional weights is generated; then, the coupling weights between the sub-functions are dynamically adjusted according to the Hausdorff dimension of the joint rotation center drift trajectory; finally, the outputs of these sub-functions are integrated through a topological integration algorithm to generate a composite score value reflecting the joint motion coordination, and compared with the physiological consistency threshold of the motion feature tensor, and the iteration step size of the dynamic calibration factor is dynamically adjusted through a backpropagation mechanism. Through the above steps, not only can the recovery status of the athlete's knee be understood in detail, but also personalized rehabilitation suggestions can be provided for him, thus accelerating his rehabilitation process.

[0100] In order to further improve the accuracy of identifying abnormal regions during joint movement and generate a set of sub-functions that can adapt to different dimensional weights, in some embodiments, the mapping of the abnormal region in the energy dissipation spatio-temporal map to the time-frequency overlapping interval of the myoelectric inhibition waveform in the neural control time series feature space and generating a set of sub-functions with variable dimensional weights based on the hysteresis effect parameters output by the viscoelastic correction model of the kinematic chain includes:

[0101] Construct a time-frequency mask matrix based on the boundary characteristics of the abnormal region in the energy dissipation spatio-temporal map. Compensate the phase delay of the mask matrix through the hysteresis effect parameter of the viscoelastic correction model of the motion chain to generate a time-frequency overlapping feature vector with the constraint of the energy propagation direction of the motion chain. Input the time-frequency overlapping feature vector into a topological mapping network constructed based on the mutual information entropy between the myoelectric phase delay and the energy propagation efficiency of the motion chain. Dynamically adjust the connection weights between network layers using the Hausdorff dimension of the drift trajectory of the joint rotation center, and output an initial weight set of sub-functions containing variable dimension parameters. The generation of the sub-function set with variable dimension weights includes: constructing a multi-dimensional weight distribution function according to the initial weight set of the sub-functions and the time-frequency attenuation characteristics of the myoelectric suppression waveform, and dynamically truncating and filtering the distribution function through the physiological consistency threshold of the motion feature tensor to generate a sub-function dimension adjustment factor that satisfies the viscoelastic constraint of the motion chain. Use the dimension adjustment factor to perform non-linear interpolation and reconstruction on the output of the topological mapping network, synchronously fuse the covariance matrix features of the manifold curvature change rate of the energy dissipation spatio-temporal map, and iteratively generate a sub-function set with a time-varying topological structure.

[0102] In this embodiment, first, the concept of the time-frequency mask matrix is introduced, which is used to mark and process the abnormal regions in the energy dissipation spatio-temporal map. Second, the phase delay of these abnormal regions is adjusted by combining the hysteresis effect parameter of the viscoelastic correction model of the motion chain to ensure that it can accurately reflect the energy propagation direction of the motion chain. Third, the topological mapping network is used to dynamically adjust the connection weights between network layers to adapt to different motion modes. Finally, a multi-dimensional weight distribution function and non-linear interpolation and reconstruction techniques are used to generate a sub-function set with adaptive dimension weights.

[0103] In the embodiment of the present application, first, construct a time-frequency mask matrix according to the boundary characteristics of the abnormal region in the energy dissipation spatio-temporal map. Second, use the hysteresis effect parameter of the viscoelastic correction model of the motion chain to compensate the phase delay of the mask matrix to generate a time-frequency overlapping feature vector with the constraint of the energy propagation direction of the motion chain. Third, input these feature vectors into a topological mapping network constructed based on the mutual information entropy between the myoelectric phase delay and the energy propagation efficiency of the motion chain, and dynamically adjust the connection weights between network layers using the Hausdorff dimension of the drift trajectory of the joint rotation center. Fourth, construct a multi-dimensional weight distribution function according to the initial weight set of the sub-functions and the time-frequency attenuation characteristics of the myoelectric suppression waveform, and perform dynamic truncation and filtering on it through the physiological consistency threshold of the motion feature tensor. Finally, use the generated dimension adjustment factor to perform non-linear interpolation and reconstruction on the output of the topological mapping network, synchronously fuse the covariance matrix features of the manifold curvature change rate of the energy dissipation spatio-temporal map, and iteratively generate a sub-function set with a time-varying topological structure.

[0104] The following is a specific example:

[0105] Suppose it is necessary to evaluate the recovery of a patient after knee surgery. First, an inertial measurement unit (IMU) and a surface electromyogram (sEMG) sensor are installed around the patient's knee; after collecting data, a time-frequency mask matrix is constructed based on the boundary characteristics of the abnormal region of the energy dissipation spatio-temporal atlas; then, the phase delay compensation is performed on the mask matrix using the hysteresis effect parameter of the viscoelastic correction model of the kinematic chain to generate a time-frequency overlapping feature vector with the constraint of the energy propagation direction of the kinematic chain; subsequently, these feature vectors are input into a topological mapping network constructed based on the mutual information entropy between the electromyogram phase delay and the energy propagation efficiency of the kinematic chain, and the connection weights between network layers are dynamically adjusted using the Hausdorff dimension of the joint rotation center drift trajectory; then, a multi-dimensional weight distribution function is constructed according to the initial weight set of the sub-function and the time-frequency decay characteristics of the electromyogram suppression waveform, and it is dynamically truncated and filtered by the physiological consistency threshold of the motion feature tensor; finally, the output of the topological mapping network is non-linearly interpolated and reconstructed using the generated dimension adjustment factor, and the covariance matrix feature of the manifold curvature change rate of the energy dissipation spatio-temporal atlas is synchronously fused to iteratively generate a set of sub-functions with a time-varying topological structure. Through the above steps, not only can the knee recovery status of the patient be understood in detail, but also personalized rehabilitation training suggestions can be provided for the patient, thereby accelerating the rehabilitation process.

[0106] To further improve the accuracy and physiological consistency of joint motion analysis, in some embodiments, it further includes:

[0107] Extracting the topological invariance index of the energy propagation path in the motion feature tensor, performing a non-linear mapping with the phase synchronization stability of the electromyogram co-pattern to generate a dynamic calibration factor; injecting the dynamic calibration factor back into the constraint conditions of the dynamic coupling equation through an iterative optimization algorithm until the energy propagation path of the motion feature tensor and the time-frequency distribution of the electromyogram suppression waveform reach a preset physiological consistency threshold.

[0108] In this embodiment, the concept of the topological invariance index is first introduced, which is used to describe the stable characteristics of the energy propagation path during motion; secondly, these topological invariance indexes are combined with the phase synchronization stability of the electromyogram co-pattern through a non-linear mapping method to generate a dynamic calibration factor that can reflect the relationship between the two; finally, the iterative optimization algorithm is used to continuously adjust the dynamic calibration factor and feedback it into the dynamic coupling equation to ensure that the final result meets the preset physiological consistency standard.

[0109] In the embodiments of the present application, first, topological invariance indexes of the energy propagation path are extracted from the motion feature tensor; second, based on the phase synchronization stability between these indexes and the myoelectric synergy patterns, a non-linear mapping is performed to generate a dynamic calibration factor; third, an iterative optimization algorithm is used to feedback and inject the dynamic calibration factor into the constraint conditions of the dynamic coupling equation; then, the parameters are gradually adjusted to make the energy propagation path of the motion feature tensor gradually approach a preset physiological consistency threshold with the time-frequency distribution of the myoelectric inhibition waveform; finally, when the preset conditions are met, the iteration is stopped to complete the entire calibration process.

[0110] The following is a specific example:

[0111] Suppose it is necessary to evaluate the recovery of an athlete after knee surgery. First, an inertial measurement unit (IMU) and a surface electromyogram (sEMG) sensor are installed around the athlete's knee; after collecting data, topological invariance indexes of the energy propagation path are extracted from the motion feature tensor; then, based on the phase synchronization stability between these indexes and the myoelectric synergy patterns, a non-linear mapping is performed to generate a dynamic calibration factor; subsequently, an iterative optimization algorithm is used to feedback and inject the dynamic calibration factor into the constraint conditions of the dynamic coupling equation; then, the parameters are gradually adjusted to make the energy propagation path of the motion feature tensor gradually approach a preset physiological consistency threshold with the time-frequency distribution of the myoelectric inhibition waveform; during this process, continuous monitoring and adjustment are carried out until the difference between the two is minimized; finally, when all conditions meet the preset criteria, the iteration is ended to obtain the final calibration result. Through the above steps, not only can the recovery status of the athlete's knee be understood in detail, but also personalized rehabilitation training suggestions can be provided for the athlete, thereby accelerating the rehabilitation process.

[0112] In order to further improve the understanding of the relationship between surface electromyogram signals and exercise loads and accurately extract myoelectric synergy patterns, in some embodiments, the construction of the time-varying mapping relationship between surface electromyogram signals and exercise loads, the surface electromyogram signals are asymmetrically windowed based on the phase mutation points of kinematic parameters, a phase synchronization quantification model of the contraction waveform of the active muscle group and the inhibition waveform of the antagonist muscle group is established within the segmented intervals, and the myoelectric synergy patterns across exercise cycles are extracted through the analysis of the energy gradient field of the surface electromyogram signals, including:

[0113] Construct a dynamic window length adjustment factor based on the detection result of the phase mutation point of the kinematic parameter. Extract the time-frequency energy mutation characteristics of the surface electromyogram signal at the phase mutation point through the wavelet transform modulus maximum detection algorithm, and generate a segmented index of the surface electromyogram signal with an asymmetric window length segmentation boundary. Dynamically divide the original surface electromyogram signal according to the segmented index of the surface electromyogram signal, and use the dynamic window length adjustment factor to perform asymmetric time-domain alignment on the signal segments of the agonist muscle group and the antagonist muscle group to generate a segmentation interval containing the phase synchronization analysis reference point. In the segmentation interval, use the Hilbert-Huang transform to extract the instantaneous phase difference sequences of the surface electromyogram signals of the agonist muscle group and the antagonist muscle group respectively, construct a phase difference mutual information entropy matrix in combination with the dynamic window length adjustment factor, and perform dynamic weighting on the entropy matrix through the curvature parameter of the energy propagation path of the motion chain to generate a phase synchronization index reflecting the neural control stability. Based on the segmented index of the surface electromyogram signal in the segmentation interval, construct an energy gradient covariance tensor, perform principal component decomposition on the phase synchronization index along the energy propagation direction of the motion chain, extract the principal component vector characterizing the cross-cycle cooperation characteristics, and perform motion chain energy normalization processing on the principal component vector through the amplitude-frequency characteristics of the multi-plane coupling error in the three-dimensional biomechanical model to generate a feature set corresponding to the neuromyoelectric cooperation mode.

[0114] In this embodiment, the concept of the dynamic window length adjustment factor is first introduced, which is used to adjust the segmentation window length of the surface electromyogram signal according to the phase mutation point of the kinematic parameter. Secondly, the wavelet transform modulus maximum detection algorithm is used to identify the key time points in the surface electromyogram signal to generate an asymmetric window length segmentation boundary. Thirdly, the original signal is segmented and asymmetric time-domain alignment is performed using these boundaries to ensure that the signal segments of the agonist muscle group and the antagonist muscle group can be correctly matched. Finally, the instantaneous phase differences of each muscle group are analyzed through the Hilbert-Huang transform, and the cross-cycle electromyoelectric cooperation mode is extracted in combination with the energy gradient covariance tensor.

[0115] In the embodiments of the present application, first, a dynamic window length adjustment factor is constructed according to the detection result of the phase mutation point of the kinematic parameters; second, the wavelet transform modulus maximum detection algorithm is used to identify the key time points in the surface electromyogram signal, and a segmentation index of the surface electromyogram signal with an asymmetric window length segmentation boundary is generated; third, the original surface electromyogram signal is dynamically windowed according to these indexes, and the signal segments of the agonist and antagonist muscle groups are asymmetrically time-aligned by using the dynamic window length adjustment factor; then, in each segmentation interval, the Hilbert-Huang transform is used to extract the instantaneous phase difference sequence of the surface electromyogram signals of the agonist and antagonist muscle groups, and a phase difference mutual information entropy matrix is constructed in combination with the dynamic window length adjustment factor; next, the entropy matrix is dynamically weighted by the curvature parameter of the energy propagation path of the motion chain to generate a phase synchronization index reflecting the neural control stability; finally, an energy gradient covariance tensor is constructed based on the segmentation index of the surface electromyogram signal in the segmentation interval, principal component analysis is performed along the energy propagation direction of the motion chain to extract the principal component vectors characterizing the cross-cycle cooperation characteristics, and the motion chain energy is normalized by the amplitude-frequency characteristics of the multi-plane coupling error in the three-dimensional biomechanical model to generate a feature set corresponding to the neuromyoelectric cooperation mode.

[0116] The following is a specific example:

[0117] Suppose it is necessary to evaluate the recovery of an athlete after knee surgery. First, inertial measurement units (IMUs) and surface electromyogram (sEMG) sensors are installed around the athlete's knee; after collecting the data, a dynamic window length adjustment factor is constructed according to the detection result of the phase mutation point of the kinematic parameters; then, the wavelet transform modulus maximum detection algorithm is used to identify the key time points in the surface electromyogram signal, and a segmentation index of the surface electromyogram signal with an asymmetric window length segmentation boundary is generated; subsequently, the original surface electromyogram signal is dynamically windowed according to these indexes, and the signal segments of the agonist and antagonist muscle groups are asymmetrically time-aligned by using the dynamic window length adjustment factor; then, in each segmentation interval, the Hilbert-Huang transform is used to extract the instantaneous phase difference sequence of the surface electromyogram signals of the agonist and antagonist muscle groups, and a phase difference mutual information entropy matrix is constructed in combination with the dynamic window length adjustment factor; next, the entropy matrix is dynamically weighted by the curvature parameter of the energy propagation path of the motion chain to generate a phase synchronization index reflecting the neural control stability; on this basis, an energy gradient covariance tensor is constructed based on the segmentation index of the surface electromyogram signal in the segmentation interval, principal component analysis is performed along the energy propagation direction of the motion chain to extract the principal component vectors characterizing the cross-cycle cooperation characteristics, and the motion chain energy is normalized by the amplitude-frequency characteristics of the multi-plane coupling error in the three-dimensional biomechanical model to generate a feature set corresponding to the neuromyoelectric cooperation mode. Through the above steps, not only can the recovery status of the athlete's knee be understood in detail, but also personalized rehabilitation training suggestions can be provided for the athlete, thus accelerating the rehabilitation process.

[0118] Figure 2 The present application provides a schematic structural diagram of an intelligent joint motion scoring system. As Figure 2 shown, the device includes:

[0119] A building module 21, configured to collect inertial motion data and surface electromyogram signals of six degrees of freedom of a target joint motion chain, trigger a dynamic sampling strategy based on the joint motion phase, and establish a motion sequence topological structure including spatio-temporal reference markers;

[0120] A generating module 22, configured to decompose the inertial motion data into a rigid body motion component and a physiological tremor component, and generate a three-dimensional biomechanical model including the drift amount of the instantaneous rotation center of the joint, the multi-plane coupling error, and the curvature of the motion trajectory through dynamic inverse solution;

[0121] A processing module 23, configured to construct a time-varying mapping relationship between the surface electromyogram signal and the motion load, perform asymmetric window length segmentation on the surface electromyogram signal based on the phase mutation points of the kinematic parameters, establish a phase synchronization quantization model of the contraction waveform of the active muscle group and the inhibition waveform of the antagonist muscle group within the segmentation interval, and extract the electromyogram co-contraction pattern across the motion cycle through the analysis of the surface electromyogram signal energy gradient field;

[0122] The generating module is further configured to perform spatio-temporal coupling on the drift amount of the instantaneous rotation center of the joint in the three-dimensional biomechanical model and the electromyogram co-contraction pattern, establish a differential constraint relationship between the joint pose change and the electromyogram activation intensity through the motion chain dynamics transfer function, and generate a motion feature tensor in a four-dimensional spatio-temporal coordinate system; analyze the evolution pattern of the motion feature tensor in real time, and autonomously construct a cluster of scoring functions with a time-varying topological structure according to the coupling intensity differences in the motion acceleration period, the stable period, and the deceleration period, where each sub-function is non-linearly superimposed through the motion chain energy transfer efficiency and the neural control stability index to generate a composite scoring value with a motion quality diagnosis function.

[0123] Figure 2 The described intelligent joint motion scoring device can execute Figure 1 the intelligent joint motion scoring method described in the embodiments shown. The implementation principle and technical effects will not be elaborated further. For the intelligent joint motion scoring device in the above embodiments, the specific manners in which each module and unit perform operations have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0124] In a possible design, Figure 2 the intelligent joint motion scoring device in the embodiments shown can be implemented as a computing device. As Figure 3 shown, the computing device can include a storage component 31 and a processing component 32;

[0125] The storage component 31 stores one or more computer instructions, and the one or more computer instructions are called and executed by the processing component 32.

[0126] The processing component 32 is used for the above Figure 1 joint motion intelligent scoring method of the above embodiment.

[0127] Among them, the processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component may also be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components for executing the above method.

[0128] The storage component 31 is configured to store various types of data to support the operation of the terminal. The storage component can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disc.

[0129] Of course, the computing device may also necessarily include other components, such as input / output interfaces, display components, communication components, etc.

[0130] The input / output interface provides an interface between the processing component and the peripheral interface module, and the above peripheral interface module may be an output device, an input device, etc.

[0131] The communication component is configured to facilitate communication between the computing device and other devices in a wired or wireless manner, etc.

[0132] Among them, the computing device may be a physical device or an elastic computing host provided by a cloud computing platform, etc. At this time, the computing device may refer to a cloud server, and the above processing component, storage component, etc. may be basic server resources leased or purchased from the cloud computing platform.

[0133] The embodiment of the present application also provides a computer storage medium storing a computer program, and when the computer program is executed by a computer, it can implement the above Figure 1 joint motion intelligent scoring method of the shown embodiment.

[0134] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0135] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0136] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the essence of the above technical solution or the part that contributes to the prior art can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0137] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and are not intended to limit them. Although the present application 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 recorded in the foregoing embodiments or equivalently replace some of the technical features. 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 application.

Claims

1. An intelligent scoring method for joint movement, characterized in that Including: Collecting six-degree-of-freedom inertial motion data and surface electromyography signals of the target joint kinematic chain, establishing a motion sequence topological structure including spatio-temporal reference markers based on a dynamic sampling strategy triggered by joint motion phases; Decomposing the inertial motion data into rigid body motion components and physiological tremor components, and generating a three-dimensional biomechanical model including joint instantaneous rotation center drift, multi-plane coupling error, and motion trajectory curvature through dynamic inverse solution; Constructing a time-varying mapping relationship between surface electromyography signals and motion loads, asymmetric window length segmentation of surface electromyography signals based on phase mutation points of kinematic parameters, establishing a phase synchronization quantification model for active muscle group contraction waveforms and antagonist muscle group inhibition waveforms within the segmentation interval, and extracting electromyographic coordination patterns across motion cycles through surface electromyography signal energy gradient field analysis; Performing spatio-temporal coupling of the joint instantaneous rotation center drift in the three-dimensional biomechanical model and the electromyographic coordination pattern, establishing a differential constraint relationship between joint pose changes and electromyographic activation intensity through the kinematic chain dynamics transfer function, and generating a motion feature tensor in a four-dimensional spatio-temporal coordinate system; Real-time analyzing the evolution pattern of the motion feature tensor, autonomously constructing a cluster of scoring functions with a time-varying topological structure according to the coupling strength differences in the motion acceleration period, stable period, and deceleration period, where each sub-function is non-linearly superimposed through the kinematic chain energy transfer efficiency and neural control stability indicators to generate a composite scoring value with a motion quality diagnosis function.

2. The intelligent joint movement scoring method according to claim 1, wherein, The performing spatio-temporal coupling of the joint instantaneous rotation center drift in the three-dimensional biomechanical model and the electromyographic coordination pattern, establishing a differential constraint relationship between joint pose changes and electromyographic activation intensity through the kinematic chain dynamics transfer function, and generating a motion feature tensor in a four-dimensional spatio-temporal coordinate system includes: Based on the kinematic chain dynamics transfer function, spatio-temporally aligning the instantaneous angular velocity covariant derivative of the joint instantaneous rotation center drift in the three-dimensional biomechanical model and the phase gradient of the electromyographic coordination pattern to generate a dynamic coupling equation; wherein, the constraint conditions of the dynamic coupling equation are jointly calibrated by the joint ligament tension distribution model and the time-domain attenuation rate of the surface electromyography signal inhibition waveform to ensure the physical dimension consistency of kinematic parameters and electromyographic parameters; In a four-dimensional spatio-temporal coordinate system, performing a tensor product operation on the output of the dynamic coupling equation and the propagation path of the joint multi-plane coupling error to generate an initial feature tensor field; wherein, the dimension of the tensor product operation is dynamically adjusted by the curvature change rate of the joint rotation center drift trajectory, so that the initial tensor field encodes both the kinematic chain energy propagation direction and the electromyographic phase synchronization; Calculating the viscoelastic coefficient matrix of the kinematic chain using the asymmetric distribution characteristic of the surface electromyography signal energy gradient, and adaptively smoothing and filtering the initial tensor field based on the viscoelastic coefficient matrix to generate a motion feature tensor including physiological hysteresis effects.

3. The method according to claim 2, wherein The calculating the viscoelastic coefficient matrix of the kinematic chain using the asymmetric distribution characteristic of the surface electromyography signal energy gradient includes: Construct an energy gradient covariance tensor based on the phase gradient distribution of the myoelectric synergy pattern, and perform anisotropic weighting on the covariance tensor through the curvature change rate of the joint rotation center drift trajectory in the three-dimensional biomechanical model to generate an initial viscoelastic parameter set with the characteristics of the energy propagation direction of the kinematic chain; Input the initial viscoelastic parameter set into the convolution kernel function constructed based on suppressing the time-domain decay rate of the surface electromyogram signal, extract the frequency-domain response characteristics corresponding to the joint ligament tension distribution model through a multi-scale sliding window, and perform parameter normalization processing in combination with the constraint conditions of the dynamic coupling equation to form a frequency-domain mapping relationship of the viscoelastic coefficient matrix; Perform adaptive smoothing filtering on the initial tensor field based on the viscoelastic coefficient matrix to generate a motion feature tensor including physiological hysteresis effects, including: Establish a bandwidth adjustment factor for the tensor field filter according to the frequency-domain mapping relationship, and dynamically match the coupling weight between the energy propagation direction of the kinematic chain and the myoelectric phase synchrony using the asymmetric distribution characteristic of the surface electromyogram signal energy gradient to generate an adaptive filtering kernel with a physiological hysteresis compensation effect; Perform multi-channel parallel convolution operations on the initial feature tensor field through the adaptive filtering kernel, synchronously fuse the propagation path curvature parameters of the joint multi-plane coupling error, and iteratively update the energy distribution topological structure of the motion feature tensor to generate a motion feature tensor including physiological hysteresis effects.

4. The intelligent joint movement scoring method according to claim 1, wherein Analyze the evolution mode of the motion feature tensor in real time, and autonomously construct a cluster of scoring functions with a time-varying topological structure according to the coupling strength differences in the motion acceleration period, stable period, and deceleration period. Among them, each sub-function performs non-linear superposition through the energy transfer efficiency of the kinematic chain and the neural control stability index to generate a composite scoring value with a motion quality diagnosis function, including: In the motion feature tensor, according to the phase boundary markers output by the motion stage recognition module, extract the distribution of manifold curvature extreme points along the energy propagation path, and calculate the covariance matrix of the manifold curvature change rate and the myoelectric phase synchronization stability within each motion stage to generate an energy dissipation spatio-temporal map; among them, the detection rule of the manifold curvature extreme points is dynamically defined by the amplitude-frequency characteristics of the joint multi-plane coupling error. In the neural control time-series feature space, map the abnormal region in the energy dissipation spatio-temporal map to the time-frequency overlapping interval of the myoelectric suppression waveform, and generate a set of sub-functions with variable-dimensional weights based on the hysteresis effect parameters output by the kinematic chain viscoelastic correction model; among them, the topological structure of the sub-functions is dynamically constructed by the Hausdorff dimension of the joint rotation center drift trajectory, and the coupling weight between the sub-functions is calculated by the mutual information entropy of the myoelectric phase delay and the kinematic chain energy propagation efficiency. Perform integral operations on the output values of the set of sub-functions along the energy propagation path of the kinematic chain through the topological integration algorithm to generate a composite scoring value reflecting joint motion coordination; at the same time, compare the composite scoring value with the physiological consistency threshold of the motion feature tensor, and dynamically adjust the iteration step size of the dynamic calibration factor through the backpropagation mechanism to achieve closed-loop optimization.

5. The method according to claim 4, wherein In the neuro-control time-sequence feature space, map the abnormal region in the energy dissipation spatio-temporal atlas to the time-frequency overlapping interval of the electromyogram inhibition waveform, and generate a set of sub-functions with variable-dimensional weights based on the hysteresis effect parameters output by the kinematic chain viscoelastic correction model, including: Construct a time-frequency mask matrix based on the boundary features of the abnormal region of the energy dissipation spatio-temporal atlas, and perform phase delay compensation on the mask matrix through the hysteresis effect parameters of the kinematic chain viscoelastic correction model to generate a time-frequency overlapping feature vector with the constraint of the energy propagation direction of the kinematic chain; Input the time-frequency overlapping feature vector into a topological mapping network constructed based on the mutual information entropy between the electromyogram phase delay and the kinematic chain energy propagation efficiency, and dynamically adjust the connection weights between network layers using the Hausdorff dimension of the joint rotation center drift trajectory to output an initial weight set of sub-functions containing variable-dimensional parameters; The generation of a set of sub-functions with variable-dimensional weights includes: Construct a multi-dimensional weight allocation function according to the initial weight set of the sub-functions and the time-frequency attenuation characteristics of the electromyogram inhibition waveform, and perform dynamic truncation filtering on the allocation function through the physiological consistency threshold of the motion feature tensor to generate a sub-function dimension adjustment factor that satisfies the kinematic chain viscoelastic constraint; Use the dimension adjustment factor to perform non-linear interpolation reconstruction on the output of the topological mapping network, and synchronously fuse the covariance matrix features of the manifold curvature change rate of the energy dissipation spatio-temporal atlas to iteratively generate a set of sub-functions with a time-varying topological structure.

6. The method according to claim 2, wherein It also includes: Extract the topological invariance index of the energy propagation path in the motion feature tensor, and perform non-linear mapping with the phase synchronization stability of the electromyogram co-pattern to generate a dynamic calibration factor; Inject the dynamic calibration factor into the constraint conditions of the dynamic coupling equation through an iterative optimization algorithm until the energy propagation path of the motion feature tensor and the time-frequency distribution of the electromyogram inhibition waveform reach a preset physiological consistency threshold.

7. The method according to claim 1, wherein The construction of the time-varying mapping relationship between the surface electromyogram signal and the motion load, based on the phase mutation points of the kinematic parameters, asymmetric window length segmentation of the surface electromyogram signal, establishment of a phase synchronization quantification model for the contraction waveform of the active muscle group and the inhibition waveform of the antagonist muscle group within the segmentation interval, and extraction of the electromyogram co-pattern across the motion cycle through the analysis of the energy gradient field of the surface electromyogram signal, including: Construct a dynamic window length adjustment factor based on the detection results of the phase mutation points of the kinematic parameters, extract the time-frequency energy mutation characteristics of the surface electromyogram signal at the phase mutation points through the wavelet transform modulus maximum detection algorithm, and generate a segmentation index of the surface electromyogram signal with an asymmetric window length segmentation boundary; Perform dynamic window division on the original surface electromyogram signal according to the segmentation index of the surface electromyogram signal, and perform asymmetric time-domain alignment on the signal segments of the active muscle group and the antagonist muscle group using the dynamic window length adjustment factor to generate a segmentation interval containing the phase synchronization analysis reference point; In the said segmentation interval, the Hilbert-Huang transform is adopted to extract the instantaneous phase difference sequences of the surface electromyography signals of the agonist muscles and antagonist muscles respectively, a phase difference mutual information entropy matrix is constructed by combining the said dynamic window length adjustment factor, and the entropy matrix is dynamically weighted by the curvature parameter of the energy propagation path of the kinematic chain to generate a phase synchronization index reflecting the neural control stability; Based on the segmented indexing of the surface electromyography signals in the said segmentation interval, an energy gradient covariance tensor is constructed, principal component decomposition is performed on the said phase synchronization index along the energy propagation direction of the kinematic chain, a principal component vector characterizing the cross-cycle cooperation characteristics is extracted, and the principal component vector is subjected to kinematic chain energy normalization processing by the amplitude-frequency characteristics of the multi-plane coupling error in the said three-dimensional biomechanical model to generate a feature set corresponding to the neuromyoelectric cooperation mode.

8. An intelligent joint movement scoring system, characterized in that, It includes: An establishment module, configured to collect the inertial motion data and surface electromyography signals of six degrees of freedom of a target joint kinematic chain, and based on the joint motion phase, trigger a dynamic sampling strategy to establish a motion sequence topological structure including spatio-temporal reference marks; A generation module, configured to decompose the said inertial motion data into a rigid body motion component and a physiological tremor component, and generate a three-dimensional biomechanical model including the instantaneous rotation center drift amount of the joint, multi-plane coupling error and motion trajectory curvature through dynamic inverse solution; A processing module, configured to construct a time-varying mapping relationship between the surface electromyography signal and the motion load, perform asymmetric window length segmentation on the surface electromyography signal based on the phase mutation points of the kinematic parameters, establish a phase synchronization quantization model for the contraction waveform of the agonist muscles and the inhibition waveform of the antagonist muscles in the segmentation interval, and extract the electromyoelectric cooperation mode across motion cycles through the surface electromyography signal energy gradient field analysis; The said generation module is further configured to perform spatio-temporal coupling on the instantaneous rotation center drift amount of the joint in the said three-dimensional biomechanical model and the electromyoelectric cooperation mode, establish a differential constraint relationship between the joint pose change and the electromyoelectric activation intensity through the kinematic chain dynamics transfer function, and generate a motion feature tensor in a four-dimensional spatio-temporal coordinate system; the evolution mode of the said motion feature tensor is analyzed in real time, and according to the coupling intensity differences in the motion acceleration period, stable period and deceleration period, a cluster of scoring functions with a time-varying topological structure is autonomously constructed, wherein each sub-function is non-linearly superimposed through the kinematic chain energy transfer efficiency and the neural control stability index to generate a composite scoring value with a motion quality diagnosis function.

9. A computing device, characterized in that, It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement the joint motion intelligent scoring method as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that, A computer program is stored, and when the computer program is executed by the computer, the joint motion intelligent scoring method as described in any one of claims 1 to 7 is implemented.

Images