Discrete muscle loss detection and analysis integrated system, method, equipment and medium

Abstract

According to the discrete muscle loss detection and analysis integrated system, method and equipment and the medium provided by the invention, the state evaluation and detection of the muscle in the movement process can be deepened from simple in-vitro scale detection, and the evaluation of the muscle loss is more accurate; meanwhile, compared with other analysis technologies related to physiological parameters, the system directionally collects electrophysiological signals and motion posture data of thigh muscles of left and right legs in a power bicycle training scene through an integrated sensor of two-channel myoelectricity and an inertial navigation chip, so that multi-dimensional correlation analysis of a force exerting state and a motion track is realized; and local pertinence and data precision of leg muscle loss assessment are improved. The system is specially designed for a power bicycle training scene, the sensor is attached to the thigh muscle, a personalized and controllable motion form can be achieved, and therefore the muscle state can be better activated, and more accurate myoelectricity and muscle sound information can be obtained.

Description

Technical Field

[0001] This invention relates to the field of muscle function monitoring and motion analysis technology, and in particular to an integrated system, method, device and medium for discrete muscle loss detection and analysis. Background Technology

[0002] Chinese patent application CN118248281A, entitled "A Method for Screening Exercise Prescriptions for Elderly Sarcopenia Using Artificial Intelligence," discloses a method for screening exercise prescriptions based on a weighted matching of target patient characteristics and historical records using an elderly sarcopenia exercise prescription database. Its core principle is to dynamically adjust weights to optimize the matching results and assist doctors in developing personalized prescriptions. However, this patent application primarily focuses on matching patient characteristics and does not address the grading and quantification of muscle loss or the location of exertion points. This makes it difficult to provide precise guidance for adjusting rehabilitation training parameters and fails to form a closed loop of "detection-analysis-intervention." In contrast, this invention achieves full-process management of data collection, multimodal analysis, and rehabilitation guidance through a power cycling training scenario.

[0003] Chinese patent CN113488163B, entitled "A Method, Device, Equipment, and Medium for Sarcopenia Identification Based on Machine Vision," discloses a method for identifying sarcopenia by collecting walking videos of subjects, extracting key points, gait contour maps, and joint flexion data based on human posture recognition algorithms, and inputting these data into a gait recognition network. Its core is to use machine vision to analyze gait features (such as stride length, gait speed, and joint angles) to indirectly assess muscle function. However, this patent cannot deeply analyze muscle capacity; it only identifies sarcopenia and does not provide targeted rehabilitation training guidance. In contrast, this invention combines the assessment results to generate personalized training programs, forming a complete intervention loop.

[0004] Chinese patent application CN115376681A, entitled "An Automated Screening Method for Sarcopenia Based on Artificial Intelligence," discloses a method that collects L3-level images from abdominal CT scans, automatically segments muscle regions based on a U-net neural network, calculates muscle area, density, and skeletal muscle index (SMI), and performs sarcopenia screening with reference to Asian standards. However, this method requires expensive experimental equipment, making it unsuitable for large-scale community and home use, and only outputs screening results without linking them to rehabilitation strategies. In contrast, this invention automatically generates training parameter adjustment schemes based on the level of muscle loss, achieving closed-loop assessment.

[0005] Chinese patent application CN116898447A, entitled "Multi-channel Surface Electromyography (EMG) Signal Acquisition Electrode and System for Sarcopenia Detection," discloses the acquisition of surface EMG signals using a multi-channel EMG acquisition electrode (including a conductive circuit layer, a silicone support layer, and a conductive hydrogel), combined with wireless transmission and algorithm analysis (time-domain and frequency-domain characteristics) to assess sarcopenia. Its core principle is to indirectly predict muscle function indicators (such as grip strength and body composition) through EMG signal characteristics. However, this patent relies on the subject's voluntary movements, resulting in poor data consistency and limiting its application to large-scale sarcopenia analysis. In contrast, this invention, based on four standardized riding modes of a power bicycle, ensures the standardization and repeatability of data acquisition and utilizes neural networks to improve assessment accuracy and interpretability.

[0006] Existing technologies either rely on indirect indicators, are limited to single-modal data, or lack intervention loops, making it difficult to achieve accurate detection, quantitative assessment, and personalized intervention for muscle loss. This invention, through a comprehensive technical approach combining discrete multimodal sensing, standardized dynamic testing, multimodal fusion analysis, and rehabilitation closed-loop management, overcomes the shortcomings of existing technologies in terms of directness, comprehensiveness, controllability, and intervention, constructing a complete muscle loss management system covering "detection-analysis-guidance-reinforcement." Summary of the Invention

[0007] To achieve the above-mentioned objectives and other advantages of the present invention, a first objective of the present invention is to provide a discrete muscle loss detection and analysis method, comprising the following steps:

[0008] Acquire electromyography and myophonation of the left and right thighs under various cycling motion states, as well as cycling motion time series data;

[0009] For electromyography, myophone, and cycling motion data, perform time-frequency domain synchronous analysis to extract time-frequency multimodal features;

[0010] Calculate cross-modal correlation features and construct a multi-dimensional feature set;

[0011] A cycling network combining a multilayer perceptron and an LSTM network is used to process cycling motion time series data, capture the temporal fluctuation patterns of cycling actions, and generate multidimensional cycling modal spatiotemporal vectors.

[0012] An electromyography (EMG) network was constructed using a convolutional network incorporating an attention mechanism. Feature optimization and filtering were performed on the two-channel and bilateral contrast data of EMG, and a multidimensional EMG modal feature vector was output.

[0013] By optimizing the coherence graph structure of the myophone network through convolution, the coherence and frequency characteristics of the myophone signal are analyzed to obtain a multidimensional myophone modal feature vector.

[0014] The cycling modal spatiotemporal vector, the electromyographic modal feature vector, and the myophonic modal feature vector are weighted and fused through an attention mechanism to form a multimodal fusion feature set, which is then input into a fully connected classification training network to achieve multiple levels of muscle loss classification and the discrimination of core indicators.

[0015] By using the SHAP value feature weight interpretation mechanism, the contribution of key parameters in the fusion feature set is decomposed, and a detailed muscle state assessment report, the core strength difference of bilateral muscle force points, and muscle function compensation analysis based on cycling data are output.

[0016] Based on the detailed assessment results of muscle condition and combined with the pre-set rehabilitation training strategy model, personalized rehabilitation training advice and guidance are generated for the subjects.

[0017] Furthermore, the cycling actions include low-frequency high-resistance cycling mode, high-frequency zero-resistance cycling mode, constant-speed cycling mode, and incremental cycling mode; among which,

[0018] The low-frequency, high-resistance cycling mode is used to assess the maximum strength and explosive power of the subject's leg muscles;

[0019] The high-frequency zero-resistance cycling mode is used to assess the coordination of rapid muscle contraction and neural control, as well as the assessment of overall balance.

[0020] The constant speed riding mode is used to evaluate muscle steady-state force efficiency and fatigue resistance.

[0021] The incremental cycling mode is used to assess muscle maximum load tolerance and critical point function, and to detect the compensatory capacity and exhaustion characteristics of muscles under extreme conditions.

[0022] Furthermore, the time-frequency multimodal features include the root mean square amplitude, median frequency, average frequency, symmetry weight, muscle power slope, and signal entropy of the electromyographic signal; the frequency spectrum distribution and energy density of the muscle sound signal; and the trunk sway amplitude curve features and posture variation quantification values ​​corresponding to the cycling data.

[0023] Furthermore, the key parameters include specific electromyographic frequency characteristics, myophone coherence index, and cycling posture variability parameters.

[0024] Furthermore, the personalized rehabilitation training advice and guidance generated for the subjects includes targeted adjustments to cycling training parameters based on core force parameters, recommendations for auxiliary rehabilitation movements, and phased training cycle planning based on the degree of muscle loss.

[0025] The second objective of this invention is to provide an integrated system for discrete muscle loss detection and analysis, employing the aforementioned method, comprising a power bicycle module, a dual-mode sensor module, and a muscle analysis module; wherein,

[0026] The power bicycle module is used to set the bicycle riding resistance and realize real-time resistance changes in order to set various riding actions;

[0027] The dual-mode sensor module is attached to both sides of the subject's left and right thighs respectively to capture electromyography and myosalpingometry under various cycling motion states, and simultaneously obtains cycling motion data of both legs, and transmits the data to the muscle analysis module for analysis;

[0028] The muscle analysis module is used to preprocess the data, calculate cross-modal correlation features, construct a multi-dimensional feature set, and call a cycling network combining a multilayer perceptron and an LSTM network to process the temporal data of cycling movements, capture the temporal fluctuation patterns of cycling actions, and generate a multi-dimensional cycling modal spatiotemporal vector. It also constructs an electromyography (EMG) network using a convolutional network incorporating an attention mechanism, optimizes and filters features of the two-channel and bilateral contrast data of EMG, and outputs a multi-dimensional EMG modal feature vector. Furthermore, it uses a muscle sound network with a convolutionally optimized coherence graph structure to analyze the coherence and frequency characteristics of muscle sound signals, obtaining a multi-dimensional muscle sound modal feature vector. Finally, it integrates the cycling modal data with the relevant data. The spatiotemporal vector, the electromyographic modal feature vector, and the myophonic modal feature vector are weighted and fused through an attention mechanism to form a multimodal fusion feature set. This set is then input into a fully connected classification training network to achieve multiple levels of muscle loss classification and the discrimination of core indicators. Through the SHAP value feature weight interpretation mechanism, the contribution of key parameters in the fusion feature set is decomposed, and a detailed muscle state assessment report, the core strength difference of bilateral muscle exertion points, and a muscle function compensation analysis based on cycling data are output. Based on the detailed muscle state assessment results and combined with a preset rehabilitation training strategy model, personalized rehabilitation training advice and guidance are generated for the subjects.

[0029] Furthermore, the cycling actions include low-frequency high-resistance cycling mode, high-frequency zero-resistance cycling mode, constant-speed cycling mode, and incremental cycling mode; among which,

[0030] The low-frequency, high-resistance cycling mode is used to assess the maximum strength and explosive power of the subject's leg muscles;

[0031] The high-frequency zero-resistance cycling mode is used to assess the coordination of rapid muscle contraction and neural control, as well as the assessment of overall balance.

[0032] The constant speed riding mode is used to evaluate muscle steady-state force efficiency and fatigue resistance.

[0033] The incremental cycling mode is used to assess muscle maximum load tolerance and critical point function, and to detect the compensatory capacity and exhaustion characteristics of muscles under extreme conditions.

[0034] Furthermore, the time-frequency multimodal features include the root mean square amplitude, median frequency, average frequency, symmetry weight, muscle power slope, and signal entropy of the electromyographic signal; the frequency spectrum distribution and energy density of the muscle sound signal; and the trunk sway amplitude curve features and posture variation quantification values ​​corresponding to the cycling data.

[0035] The key parameters include specific electromyographic frequency characteristics, muscle-sound coherence index, and cycling posture variability parameters.

[0036] The personalized rehabilitation training advice and guidance generated for the subjects includes targeted adjustments to cycling training parameters based on core force parameters, recommendations for auxiliary rehabilitation movements, and phased training cycle planning based on the degree of muscle loss.

[0037] A third objective of the present invention is to provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method.

[0038] A fourth objective of the present invention is to provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described method.

[0039] Compared with the prior art, the beneficial effects of the present invention are:

[0040] This invention can move beyond simple in vitro scale testing to in-depth assessment of muscle status during exercise, resulting in a more accurate evaluation of muscle loss. Furthermore, compared to other physiological parameter analysis techniques, this system, through an integrated sensor combining two-channel electromyography (EMG) and an inertial navigation chip, can directionally collect electrophysiological signals and motion posture data of the left and right thigh muscles during power cycling training. This enables multi-dimensional correlation analysis of force exertion and movement trajectory, improving the local specificity and data accuracy of leg muscle loss assessment.

[0041] This system is designed specifically for power cycling training scenarios. The sensor is attached to the thigh muscle to enable personalized and controllable movement, thereby better activating muscle state and obtaining more accurate electromyography and myosalpingography information.

[0042] Furthermore, existing technologies often require the use of multiple independent devices (such as electromyography (EMG) devices and motion capture devices), which are complex to operate and difficult to synchronize data. This invention adopts an integrated sensor design of "two-channel EMG + inertial navigation chip". A single sensor can acquire multiple parameters, and the discrete arrangement (one on each leg) simplifies the system structure. It does not require complex wiring or large equipment support, and users can apply it themselves, reducing the operation threshold and usage cost.

[0043] This invention integrates electromyography (EMG) signals (reflecting muscle activation) with inertial navigation data (reflecting limb movement posture, speed, etc.) to comprehensively assess the force efficiency, symmetry, and fatigue trend of leg muscles. Compared with single-parameter detection (such as EMG or motion data only), it can more comprehensively determine the state of muscle loss and provide richer quantitative evidence for muscle function assessment.

[0044] Finally, the number of sensors can be adjusted according to needs (not limited to two), which can adapt to the needs of different detection accuracy or parts, enhancing the versatility of the system; at the same time, the miniaturized design of the sensors ensures that they do not interfere with power cycling training movements after being applied, improving user comfort and exercise safety.

[0045] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it according to the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Specific embodiments of the present invention are given in detail below with reference to the accompanying drawings. Attached Figure Description

[0046] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0047] Figure 1 Schematic diagram of an integrated system for discrete muscle loss detection and analysis;

[0048] Figure 2 Flowchart of a discrete muscle loss detection and analysis method;

[0049] Figure 3 A classification diagram of cycling movements;

[0050] Figure 4 Here is a flowchart of the time-frequency multimodal feature extraction process;

[0051] Figure 5 A flowchart classifying the six levels of muscle loss.

[0052] Figure 6 This is a flowchart illustrating the overall system's operational mode.

[0053] Figure 7 This is a schematic diagram of a computer device.

[0054] Figure 8 This is a schematic diagram of a computer-readable storage medium. Detailed Implementation

[0055] The present invention will now be further described with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be noted that, without conflict, the various embodiments or technical features described below can be arbitrarily combined to form new embodiments.

[0056] Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this invention.

[0057] The drawing numbers in this application are only used to distinguish the steps in the scheme and are not used to limit the execution order of the steps. The specific execution order is as described in the specification.

[0058] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0059] This invention provides a discrete sensing system suitable for power cycling training scenarios, enabling real-time acquisition and analysis of the exertion state of leg muscles and assessment of muscle loss risk. The specific solution is as follows:

[0060] Example 1

[0061] A discrete muscle loss detection and analysis integrated system, such as Figure 1 As shown, it includes a power bicycle module 1, a dual-mode sensor module 2, and a muscle analysis module 3; wherein,

[0062] The power bicycle module can automatically set the bicycle riding resistance and also realize real-time resistance changes. Based on this, four riding actions are set, such as... Figure 3 As shown, these represent low-frequency high-resistance riding mode, high-frequency zero-resistance riding mode, constant-speed riding mode, and incremental riding mode, respectively.

[0063] The low-frequency, high-resistance cycling mode can be used to assess the maximum strength and explosive power of the subject's leg muscles;

[0064] The high-frequency zero-resistance cycling mode can assess the coordination of rapid muscle contraction and neural control ability, as well as the overall balance.

[0065] The constant speed riding mode can assess muscle steady-state force efficiency and fatigue resistance;

[0066] Incremental cycling assesses the maximum load tolerance and critical point function of muscles, and detects the compensatory ability and exhaustion characteristics of muscles under extreme conditions.

[0067] Based on this, the dual-mode sensor modules are respectively attached to Figure 1 The electromyography and myosalpingography (EMG) of the subjects' left and right thighs were captured under these four cycling motion conditions. The cycling motion data of both legs was also obtained and transmitted to the muscle analysis module on the computer via WIFI for analysis.

[0068] It should be noted that the discrete sensing system based on human leg muscle monitoring includes two patch sensors, but is not limited to two; single, three, four, etc., are all within the scope of protection of this invention.

[0069] Each sensor integrates a two-channel electromyography (EMG) module and an inertial navigation chip into a single design, enabling simultaneous acquisition of leg muscle electrical signals and motion posture data, including but not limited to two-channel EMG, or other multi-channel EMG data.

[0070] For power cycling training scenarios, sensors are attached to the thigh muscles of both legs to enable directional monitoring of leg exertion.

[0071] like Figure 4 As shown, after the data enters the muscle analysis module, the module first preprocesses the data: for electromyography (EMG), myophone, and cycling motion data, it performs time-frequency domain synchronous analysis to extract time-frequency multimodal features, including but not limited to the root mean square amplitude, median frequency, average frequency, symmetry weight, muscle power slope, and signal entropy of the EMG signal, the frequency spectrum distribution and energy density of the myophone signal, and the trunk sway amplitude curve features and posture variation quantification values ​​corresponding to the cycling data; at the same time, based on the collected data from both sides of the left and right legs, it calculates cross-modal correlation features such as EMG-myophone coherence and constructs a multi-dimensional feature set. Subsequently, the muscle analysis module calls a cycling network combining a multilayer perceptron and an LSTM network to process the temporal data of cycling movements, capture the temporal fluctuation patterns of cycling actions, and generate a 16-dimensional spatiotemporal vector of cycling modality. It then constructs an electromyography (EMG) network using a convolutional network incorporating an attention mechanism to optimize and filter the features of the two-channel and bilateral contrast data of EMG, outputting a 32-dimensional EMG modal feature vector. Finally, it uses a muscle sound network with a convolutionally optimized coherence graph structure to analyze the coherence and frequency characteristics of the muscle sound signal, obtaining a 16-dimensional muscle sound modal feature vector.

[0072] The fusion analysis method of two-channel electromyography signals and inertial navigation data is used to determine the efficiency, symmetry and fatigue of leg muscles by correlating multi-dimensional data, and then to assess the state of muscle loss.

[0073] like Figure 5As shown, the muscle analysis module weights and fuses the three acquired modal feature vectors through an attention mechanism to form a multimodal fusion feature set. This set is then input into a fully connected classification training network to classify muscle loss into six levels (from A to E, indicating increasingly severe loss) and to determine core indicators such as the distribution of force point strength. Simultaneously, to enhance the interpretability and rehabilitation guidance value of the analysis results, the system introduces a SHAP value feature weight interpretation mechanism. This mechanism decomposes the contribution of key parameters in the fusion feature set (such as specific electromyographic frequency characteristics, muscle-phonemic coherence indices, and cycling posture variability parameters) and outputs a detailed muscle state assessment report, differences in core strength of bilateral muscle force points, and muscle function compensation analysis based on cycling data.

[0074] Ultimately, based on the detailed assessment results of muscle status and combined with the preset rehabilitation training strategy model, the muscle analysis module generates personalized rehabilitation training advice and guidance for the subjects. This includes targeted adjustments to cycling training parameters for core force exertion parameters (such as power bike resistance and cadence optimization), recommendations for auxiliary rehabilitation movements (targeted strengthening movements based on muscle activation characteristics), and phased training cycle planning based on the degree of muscle loss. This forms a complete technical process of "data collection - multimodal analysis - intelligent assessment - rehabilitation closed loop", and builds a muscle loss analysis and strengthening system covering detection, analysis, and intervention.

[0075] This invention generates personalized solutions based on evaluation results, including adjustments to cycling parameters, recommendations for assistive movements, and phased periodic planning. It also achieves a closed-loop intervention mechanism for dynamic optimization through data feedback during the training process, forming a personalized rehabilitation training closed-loop management system.

[0076] This invention obtains muscle limit data through pre-testing and performs preliminary classification, achieving adaptive matching between cycling motion parameters and the subject's basic ability, ensuring the accuracy of data collection and training safety, and providing a dynamic parameter adjustment mechanism adapted to individual abilities.

[0077] Specifically, such as Figure 6As shown, the overall system operation is divided into four main steps. First, pre-testing and basic grading are performed: subjects first undergo maximum resistance and maximum speed tests. Based on a power cycling training scenario, this test obtains basic data on leg muscles under extreme exercise conditions (such as power output, exercise limit threshold, etc.), which serves as the basis for subsequent grading. After the test, the system performs a preliminary rating (Level IV, the higher the level, the worse the muscle ability) based on the obtained data, providing an initial reference for switching power cycling training levels and ensuring that subsequent training and data collection are adapted to the subject's basic motor ability. Next, dynamic data acquisition is performed: During the movement execution phase, subjects divide the standard parameters by the preliminary muscle rating obtained in the previous step, using this as the initial parameters for their cycling movement. Simultaneously, discrete multimodal sensors applied to the thigh muscles of both legs are activated to simultaneously acquire electromyography (EMG) and myocardial phonation (MCM) data, capturing muscle electrophysiological signals and movement posture data to provide multimodal raw data for muscle function analysis.

[0078] Next, data processing and muscle assessment are performed: The system performs time-frequency domain synchronous parsing preprocessing on electromyography (EMG), myophone, and cycling multimodal data to extract time-frequency multimodal features. At the same time, it calculates cross-modal correlation features such as EMG and myophone coherence. After eliminating data noise and aligning the time dimension, a standardized muscle function dataset is constructed. The processed data is then used to perform muscle status rating (AE level) based on a fusion network algorithm, combined with muscle loss degree classification rules and force point strength discrimination logic. The output includes assessment results such as quantitative classification of muscle loss degree, comparison of bilateral muscle force point strength, and localization of muscle function defect nature.

[0079] Finally, a closed-loop system for muscle loss assessment is constructed: the muscle status analysis results obtained from the assessment are, on the one hand, intuitively fed back to the subject through a home-based device, clearly presenting the current state of muscle function and the distribution of problems; on the other hand, it serves as the core input for rehabilitation training strategies, planning personalized rehabilitation training programs for the subject (optimizing power cycling training parameters such as resistance, cadence, and unilateral force compensation training modes based on differences in the strength of force exertion points; recommending auxiliary rehabilitation movements based on electromyography and myophonation characteristics, matching muscle activation sequence, and strengthening weak muscle groups; and formulating phased training cycles based on AE loss levels, including monitoring nodes and reassessment cycles); when the subject performs training through the home-based device, dual-mode sensors synchronously collect training process data and transmit it back to the medical device via WIFI; the medical device calls the SHAP value feature weight interpretation mechanism to re-analyze the transmitted data, dynamically optimize the training program, and ultimately form a closed-loop muscle loss analysis and reinforcement system, realizing a closed-loop muscle loss management system of "detection-analysis-guidance-reinforcement".

[0080] This invention provides an intelligent muscle loss analysis system based on multimodal discrete muscle monitoring. The system dynamically activates leg muscles during power cycling training, and, in conjunction with discrete multimodal sensors applied to the left and right thighs, simultaneously collects muscle electrophysiological signals and movement posture data, thereby obtaining detailed information on overall and local muscle function changes. Then, using network algorithms to fuse and analyze the multidimensional data, it can not only perform multi-level classification assessments of muscle loss but also accurately locate the strength distribution of muscle exertion points, thus providing targeted guidance for the subject's subsequent rehabilitation training. Ultimately, this forms a closed-loop muscle loss management system encompassing "detection-analysis-guidance-reinforcement."

[0081] Example 2

[0082] A discrete muscle loss detection and analysis method is provided, based on the aforementioned integrated discrete muscle loss detection and analysis system. A detailed description of the integrated discrete muscle loss detection and analysis system can be found in the corresponding description in the above system embodiments, and will not be repeated here. Figure 2 As shown, the method includes the following steps:

[0083] Acquire electromyography and myophonation of the left and right thighs under various cycling motion states, as well as cycling motion time series data;

[0084] like Figure 1 As shown, this embodiment automatically sets the cycling resistance through a power bicycle module, and can also realize real-time resistance changes. Based on this, four cycling actions are set, such as... Figure 3 As shown, these represent low-frequency high-resistance riding mode, high-frequency zero-resistance riding mode, constant-speed riding mode, and incremental riding mode, respectively.

[0085] The low-frequency, high-resistance cycling mode can be used to assess the maximum strength and explosive power of the subject's leg muscles;

[0086] The high-frequency zero-resistance cycling mode can assess the coordination of rapid muscle contraction and neural control ability, as well as the overall balance.

[0087] The constant speed riding mode can assess muscle steady-state force efficiency and fatigue resistance;

[0088] Incremental cycling assesses the maximum load tolerance and critical point function of muscles, and detects the compensatory ability and exhaustion characteristics of muscles under extreme conditions.

[0089] Based on this, the dual-mode sensor modules are respectively attached to Figure 1 The electromyography and myophonation of the left and right thighs are captured under these four cycling motion states. At the same time, cycling motion data of both legs can be obtained and transmitted to the muscle analysis module on the computer via WIFI for analysis.

[0090] like Figure 4 As shown, the data is preprocessed: the electromyography, myophone, and cycling motion data are analyzed in the time and frequency domain simultaneously to extract time and frequency multimodal features, including but not limited to the root mean square amplitude, median frequency, average frequency, symmetry weight, muscle power slope, and signal entropy of the electromyography signal, the frequency spectrum distribution and energy density of the myophone signal, and the trunk sway amplitude curve features and posture variation quantification values ​​corresponding to the cycling data.

[0091] Meanwhile, based on the collected bilateral data of the left and right legs, cross-modal correlation features such as electromyography-muscle sound coherence are calculated to construct a multi-dimensional feature set;

[0092] A cycling network combining a multilayer perceptron and an LSTM network is used to process cycling motion time series data, capture the temporal fluctuation patterns of cycling actions, and generate a 16-dimensional cycling modal spatiotemporal vector.

[0093] An electromyography (EMG) network was constructed using a convolutional network incorporating an attention mechanism. Feature optimization and filtering were performed on the two-channel and bilateral contrast data of EMG, and a 32-dimensional EMG modal feature vector was output.

[0094] By optimizing the coherence graph structure of the myophone network through convolution, the coherence and frequency characteristics of the myophone signal are analyzed, resulting in a 16-dimensional myophone modal feature vector.

[0095] like Figure 5 As shown, the spatiotemporal vector of the cycling modality, the feature vector of the electromyography modality, and the feature vector of the myophonic modality are weighted and fused through an attention mechanism to form a multimodal fusion feature set, which is then input into a fully connected classification training network to achieve the discrimination of core indicators such as six levels of muscle loss (from A to E, the loss becomes increasingly severe) and the distribution of force point strength.

[0096] Meanwhile, to enhance the interpretability and rehabilitation guidance value of the analysis results, the system introduces the SHAP value feature weight interpretation mechanism, which decomposes the contribution of key parameters in the fusion feature set (such as specific electromyographic frequency characteristics, muscle sound coherence index, cycling posture variability parameters, etc.) and outputs detailed muscle state assessment reports, differences in core strength of bilateral muscle force points, and muscle function compensation analysis based on cycling data.

[0097] Ultimately, based on the detailed assessment results of muscle status and combined with the pre-set rehabilitation training strategy model, personalized rehabilitation training advice and guidance are generated for the subjects. This includes targeted adjustments to cycling training parameters for core force exertion parameters (such as power bike resistance and cadence optimization), recommendations for auxiliary rehabilitation movements (targeted strengthening movements based on muscle activation characteristics), and phased training cycle planning based on the degree of muscle loss. This forms a complete technical process of "data collection - multimodal analysis - intelligent assessment - rehabilitation closed loop", and constructs a muscle loss analysis and strengthening system covering detection, analysis, and intervention.

[0098] Example 3

[0099] A computer device 1, such as Figure 7 As shown, the system includes a memory 11, a processor 12, and a computer program 13 stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of a discrete muscle loss detection and analysis method. For a detailed description of the method, please refer to the corresponding description in the above method embodiments; it will not be repeated here.

[0100] Example 4

[0101] A computer-readable storage medium, such as Figure 8 As shown, a computer program is stored thereon. When executed by a processor, the computer program implements the steps of a discrete muscle loss detection and analysis method. For a detailed description of the method, please refer to the corresponding description in the above method embodiments, which will not be repeated here.

[0102] The number of devices and processing scale described herein are for the purpose of simplifying the description of the invention. Applications, modifications, and variations of the invention will be readily apparent to those skilled in the art.

[0103] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

[0104] The apparatus, computer device, and non-volatile computer storage medium and method provided in the embodiments of this specification are corresponding. Therefore, the apparatus, computer device, and non-volatile computer storage medium also have similar beneficial technical effects as the corresponding method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the corresponding apparatus, computer device, and non-volatile computer storage medium will not be repeated here.

[0105] Those skilled in the art will also know that, besides implementing the controller in the form of purely computer-readable program code, the same functions can be achieved by logically programming the method steps, making the controller take the form of logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers (PLCs), and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the devices included within it for implementing various functions can also be considered structures within that hardware component. Alternatively, the devices for implementing various functions can be considered as both software units implementing the method and structures within a hardware component.

[0106] The systems, apparatuses, or units described in the above embodiments can be implemented by computer chips or physical entities, or by products with certain functions. For ease of description, the above apparatuses are described separately as various units based on their functions. Of course, when implementing one or more embodiments of this specification, the functions of each unit can be implemented in one or more software and / or hardware.

[0107] Those skilled in the art will understand that the embodiments of this specification can be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the embodiments of this specification can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0108] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0109] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0110] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0111] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0112] This specification may be described in the general context of computer-executable instructions, such as program units, that are executed by a computer. Generally, program units include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This specification may also be practiced in distributed computing environments, where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program units may reside in local and remote computer storage media, including storage devices.

[0113] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0114] The above description is merely an embodiment of this specification and is not intended to limit the scope of one or more embodiments of this specification. Various modifications and variations can be made to one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of one or more embodiments of this specification.

Claims

1. A discrete muscle loss detection and analysis method, characterized in that, Includes the following steps: Acquire electromyography and myophonation of the left and right thighs under various cycling motion states, as well as cycling motion time series data; For electromyography, myophone, and cycling motion data, perform time-frequency domain synchronous analysis to extract time-frequency multimodal features; Calculate cross-modal correlation features and construct a multi-dimensional feature set; A cycling network combining a multilayer perceptron and an LSTM network is used to process cycling motion time series data, capture the temporal fluctuation patterns of cycling actions, and generate multidimensional cycling modal spatiotemporal vectors. An electromyography (EMG) network was constructed using a convolutional network incorporating an attention mechanism. Feature optimization and filtering were performed on the two-channel and bilateral contrast data of EMG, and a multidimensional EMG modal feature vector was output. By optimizing the coherence graph structure of the myophone network through convolution, the coherence and frequency characteristics of the myophone signal are analyzed to obtain a multidimensional myophone modal feature vector. The cycling modal spatiotemporal vector, the electromyographic modal feature vector, and the myophonic modal feature vector are weighted and fused through an attention mechanism to form a multimodal fusion feature set, which is then input into a fully connected classification training network to achieve multiple levels of muscle loss classification and the discrimination of core indicators. By using the SHAP value feature weight interpretation mechanism, the contribution of key parameters in the fusion feature set is decomposed, and a detailed muscle state assessment report, the core strength difference of bilateral muscle force points, and muscle function compensation analysis based on cycling data are output. Based on the detailed assessment results of muscle condition and combined with the pre-set rehabilitation training strategy model, personalized rehabilitation training advice and guidance are generated for the subjects.

2. The discrete muscle loss detection and analysis method as described in claim 1, characterized in that: The cycling actions include low-frequency high-resistance cycling mode, high-frequency zero-resistance cycling mode, constant-speed cycling mode, and incremental cycling mode; among them... The low-frequency, high-resistance cycling mode is used to assess the maximum strength and explosive power of the subject's leg muscles; The high-frequency zero-resistance cycling mode is used to assess the coordination of rapid muscle contraction and neural control, as well as the assessment of overall balance. The constant speed riding mode is used to evaluate muscle steady-state force efficiency and fatigue resistance. The incremental cycling mode is used to assess muscle maximum load tolerance and critical point function, and to detect the compensatory capacity and exhaustion characteristics of muscles under extreme conditions.

3. The discrete muscle loss detection and analysis method as described in claim 1, characterized in that: The time-frequency multimodal features include the root mean square amplitude, median frequency, average frequency, symmetry weight, muscle power slope, and signal entropy of the electromyographic signal; the frequency spectrum distribution and energy density of the muscle sound signal; and the trunk sway amplitude curve features and posture variation quantification values ​​corresponding to the cycling data.

4. The discrete muscle loss detection and analysis method as described in claim 1, characterized in that: The key parameters include specific electromyographic frequency characteristics, muscle sound coherence index, and cycling posture variability parameters.

5. The discrete muscle loss detection and analysis method as described in claim 1, characterized in that: The personalized rehabilitation training advice and guidance generated for the subjects includes targeted adjustments to cycling training parameters based on core force parameters, recommendations for auxiliary rehabilitation movements, and phased training cycle planning based on the degree of muscle loss.

6. A discrete muscle loss detection and analysis integrated system, using the method described in any one of claims 1 to 5, characterized in that: It includes a power bicycle module, a dual-mode sensor module, and a muscle analysis module; among which, The power bicycle module is used to set the bicycle riding resistance and realize real-time resistance changes in order to set various riding actions; The dual-mode sensor module is attached to both sides of the subject's left and right thighs respectively to capture electromyography and myosalpingometry under various cycling motion states, and simultaneously obtains cycling motion data of both legs, and transmits the data to the muscle analysis module for analysis; The muscle analysis module is used to preprocess the data, calculate cross-modal correlation features, construct a multi-dimensional feature set, and call a cycling network combining a multilayer perceptron and an LSTM network to process the temporal data of cycling movements, capture the temporal fluctuation patterns of cycling actions, and generate a multi-dimensional cycling modal spatiotemporal vector. It also constructs an electromyography (EMG) network using a convolutional network incorporating an attention mechanism, optimizes and filters features of the two-channel and bilateral contrast data of EMG, and outputs a multi-dimensional EMG modal feature vector. Furthermore, it uses a muscle sound network with a convolutionally optimized coherence graph structure to analyze the coherence and frequency characteristics of muscle sound signals, obtaining a multi-dimensional muscle sound modal feature vector. Finally, it integrates the cycling modal data with the relevant data. The spatiotemporal vector, the electromyographic modal feature vector, and the myophonic modal feature vector are weighted and fused through an attention mechanism to form a multimodal fusion feature set. This set is then input into a fully connected classification training network to achieve multiple levels of muscle loss classification and the discrimination of core indicators. Through the SHAP value feature weight interpretation mechanism, the contribution of key parameters in the fusion feature set is decomposed, and a detailed muscle state assessment report, the core strength difference of bilateral muscle exertion points, and a muscle function compensation analysis based on cycling data are output. Based on the detailed muscle state assessment results and combined with a preset rehabilitation training strategy model, personalized rehabilitation training advice and guidance are generated for the subjects.

7. The integrated system for discrete muscle loss detection and analysis as described in claim 6, characterized in that: The cycling actions include low-frequency high-resistance cycling mode, high-frequency zero-resistance cycling mode, constant-speed cycling mode, and incremental cycling mode; among them... The low-frequency, high-resistance cycling mode is used to assess the maximum strength and explosive power of the subject's leg muscles; The high-frequency zero-resistance cycling mode is used to assess the coordination of rapid muscle contraction and neural control, as well as the assessment of overall balance. The constant speed riding mode is used to evaluate muscle steady-state force efficiency and fatigue resistance. The incremental cycling mode is used to assess muscle maximum load tolerance and critical point function, and to detect the compensatory capacity and exhaustion characteristics of muscles under extreme conditions.

8. The integrated system for discrete muscle loss detection and analysis as described in claim 6, characterized in that: The time-frequency multimodal features include the root mean square amplitude, median frequency, average frequency, symmetry weight, muscle power slope, and signal entropy of the electromyographic signal; the frequency spectrum distribution and energy density of the muscle sound signal; and the trunk sway amplitude curve features and posture variation quantification values ​​corresponding to the cycling data. The key parameters include specific electromyographic frequency characteristics, muscle-sound coherence index, and cycling posture variability parameters. The personalized rehabilitation training advice and guidance generated for the subjects includes targeted adjustments to cycling training parameters based on core force parameters, recommendations for auxiliary rehabilitation movements, and phased training cycle planning based on the degree of muscle loss.

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 5.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 5.

Images