A cervical and lumbar vertebra rehabilitation training system integrating posture monitoring and movement guidance

Abstract

This invention relates to the field of cervical and lumbar spine rehabilitation technology, and more particularly to a cervical and lumbar spine rehabilitation training system integrating posture monitoring and motion guidance. The system includes the following modules: a data acquisition and preprocessing module, used to collect real-time three-dimensional posture data of the user's cervical, thoracic, and lumbar spine and related joints, and generate motion data after standardization processing; and a micro-deviation analysis module, which generates a personalized micro-deviation accumulation model based on the motion data, predicts micro-deviations in real time, and generates personalized training adjustment strategies based on the user's physiological characteristics, historical training data, and current movements. In this invention, the micro-deviation analysis module extracts joint position, angle, velocity, and acceleration features from the standardized motion data, and combines this with nonlinear system modeling, user physiological characteristics, and historical training data to realize a personalized micro-deviation accumulation model, generating real-time micro-deviation prediction and training adjustment strategies.

Description

Technical Field

[0001] This invention relates to the field of cervical and lumbar spine rehabilitation technology, and in particular to a cervical and lumbar spine rehabilitation training system that integrates posture monitoring and motion guidance. Background Technology

[0002] Rehabilitation training software is typically used to monitor and analyze the posture of a user's cervical, thoracic, and lumbar spine and related joints. It records and evaluates the range, angle, and sequence of movements by collecting three-dimensional posture data, and provides users with basic training guidance and feedback. This type of software can be combined with single movements or preset training plans to assist in the management of the user's rehabilitation training process, so as to facilitate the execution of sports rehabilitation and the tracking of its effects.

[0003] In existing technologies, rehabilitation training software can usually only perform simple analysis of single movements or macroscopic postures, lacking continuous tracking and cumulative modeling of minute movement deviations, and unable to make dynamic predictions in combination with individual physiological characteristics. This leads to the gradual accumulation of minute deviations by users during long-term training without being detected, which may result in potential chronic load or sports injury risks. Summary of the Invention

[0004] To overcome the above shortcomings, this invention provides a cervical and lumbar spine rehabilitation training system that integrates posture monitoring and motion guidance. It aims to improve the problem that existing rehabilitation software cannot track minute movement deviations and accumulate modeling, leading to potential chronic load or sports injury risks.

[0005] In a first aspect, the present invention provides the following technical solution: a cervical and lumbar spine rehabilitation training system integrating posture monitoring and motion guidance, comprising the following modules: The data acquisition and preprocessing module is used to collect the user's cervical spine, thoracic spine, lumbar spine and related joints in real time, and generate motion data after standardization processing; The micro-deviation analysis module generates a user-specific micro-deviation accumulation model based on motion data, predicts micro-deviations in real time, and generates personalized training adjustment strategies based on the user's physiological characteristics, historical training data, and current actions. The action adjustment module is used to predict the local load anomalies caused by the user's next action based on the training adjustment strategy, and generate optimization instructions to update the training action by adjusting the action amplitude, sequence and virtual auxiliary prompts. The collaborative optimization module identifies abnormal coupling of multiple body parts based on user action data, generates collaborative optimization strategies, and transmits optimization instructions to the action adjustment module. The model update module is used to dynamically update the micro-bias accumulation model and training adjustment strategy based on the real-time action data of each training session; The training strategy update module generates new training instructions based on the updated micro-bias accumulation model and training adjustment strategy, and adjusts the training actions and training process in real time.

[0006] By adopting the above technical solution, continuous tracking and cumulative modeling of users' minute movement deviations are achieved. Combined with individual physiological characteristics and historical training data, dynamic prediction is performed, and real-time training adjustment strategies are generated, thereby effectively intervening in potential chronic load or sports injury risks.

[0007] Preferably, the standardization process includes: The three-dimensional pose data of the user's cervical spine, thoracic spine, lumbar spine and related joints are converted into a unified reference coordinate system; The motion data acquired from multiple joints is time-aligned, and the motion data is processed using adaptive filtering or low-pass filtering methods. The motion amplitude of each joint is normalized. For lost or abnormally collected data, interpolation algorithms or prediction methods based on historical data are used to complete the data, and the processed data is integrated to generate standardized action data.

[0008] Preferably, the generated micro-bias accumulation model includes: Extract joint position, angle, motion speed, and acceleration features from standardized motion data; Time series modeling is performed on the extracted feature data to form a feature sequence for each action stage; Nonlinear system modeling methods are used to train the feature sequences to establish a micro-bias accumulation model; By combining model parameters with users' physiological characteristics and historical training data, a user-personalized micro-bias accumulation model is formed.

[0009] Preferably, the real-time prediction micro-bias includes: Receive standardized user action data and a micro-deviation accumulation model; Real-time motion data is matched with a micro-bias accumulation model, and feature sequences are compared for each motion stage. A recurrent neural network is used to calculate the real-time feature sequence and predict the micro-deviation value at the next time step. The prediction results are used to generate a sequence of action deviations, which are then used to generate personalized training and adjustment strategies.

[0010] Preferably, the personalized training adjustment strategy includes: Receive real-time predicted micro-deviation sequences and user physiological characteristic data; Analyze the micro-deviation sequence to identify deviations in the amplitude, angle, and sequence of movements; The training adjustment parameters are determined by matching the motion deviation with the user's historical training data; Generate training adjustment strategies, including adjusting the range of motion, the sequence of motions, and auxiliary prompts.

[0011] Preferably, the prediction of local load anomalies caused by the user's next action includes: Perform joint-by-joint and stage-by-stage analysis on motion data to calculate the load state of each joint in the next motion time step; The load prediction model is used to calculate the local stress of each joint to determine the possible location and magnitude of abnormal loads. Generate a sequence of local load anomalies to guide adjustments to the magnitude, sequence, and auxiliary prompts of actions.

[0012] Preferably, the identification of multi-part motion coupling anomalies includes: Receive real-time user action data and micro-deviation accumulation model; Correlation analysis is performed on the motion data of each joint to calculate the coupling relationship between joint angles, velocities, and accelerations; By comparing the inter-joint coupling relationship with historical movement patterns, abnormal coupling situations can be identified.

[0013] Preferably, the updated micro-bias accumulation model and training adjustment strategy include: It receives real-time user action data, predicted micro-bias sequences, and personalized training adjustment strategies; The real-time motion data is compared with the micro-deviation accumulation model to calculate new micro-deviation accumulation parameters; The parameter values ​​of the micro-bias accumulation model are updated by combining the user's historical training data and action deviation. The training strategy parameters, including movement amplitude, sequence, and auxiliary cue information, are adjusted based on the updated micro-bias accumulation model.

[0014] Preferably, the real-time adjustment of training movements and training process includes: Receive the updated micro-bias accumulation model and training adjustment strategy; calculate the specific action amplitude, sequence, and auxiliary cues for each action phase based on the training adjustment strategy; Generate new training instructions, including adjustments to the range of motion, updates to the sequence of motions, and virtual assistive prompts; New training instructions are transmitted to the motion adjustment module to update the training motion; Monitor the execution of training actions in real time and adjust training process parameters according to new training instructions; The training instructions and training process adjustment information are output for the execution and optimization of actions in the next training cycle.

[0015] The present invention has the following beneficial effects: 1. In this invention, the micro-deviation analysis module extracts joint position, angle, velocity and acceleration features from standardized motion data, and combines nonlinear system modeling with user physiological characteristics and historical training data to realize a personalized micro-deviation accumulation model, generating real-time micro-deviation prediction and training adjustment strategies, which solves the problem that user micro-movement deviations are difficult to detect and that long-term accumulation leads to potential load or chronic injury.

[0016] 2. In this invention, the phased motion data of each joint is analyzed by the motion adjustment module, and the local stress state and possible abnormal locations are predicted by the load prediction model. This realizes a closed-loop optimization training strategy that dynamically adjusts the motion amplitude, sequence and virtual auxiliary prompts, which solves the problem that the user's next action in a complex motion combination may cause local load abnormalities and existing software only provides passive prompts.

[0017] 3. In this invention, the collaborative optimization module performs correlation analysis on the motion data of each joint, calculates the coupling relationship between joint angles, velocities and accelerations, and compares it with historical motion patterns to identify abnormal coupling situations. This enables the generation of closed-loop optimization strategies and real-time updates of training movements, solving the problem that abnormal coupling of multiple parts of the body cannot be detected when users perform compound movements and that long-term training may lead to poor motion patterns. Attached Figure Description

[0018] Figure 1 This is an architectural diagram of a cervical and lumbar spine rehabilitation training system that integrates posture monitoring and motion guidance, as proposed in this invention. Detailed Implementation

[0019] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example 1: In a first embodiment of the present invention, the present invention provides a cervical and lumbar spine rehabilitation training system integrating posture monitoring and motion guidance, such as... Figure 1 As shown, it includes the following modules: The data acquisition and preprocessing module is used to collect the user's cervical spine, thoracic spine, lumbar spine and related joints in real time, and generate motion data after standardization processing; Furthermore, standardization processes include: The three-dimensional pose data of the user's cervical spine, thoracic spine, lumbar spine and related joints are converted into a unified reference coordinate system; The motion data acquired from multiple joints is time-aligned, and the motion data is processed using adaptive filtering or low-pass filtering methods. The motion amplitude of each joint is normalized. For lost or abnormally collected data, interpolation algorithms or prediction methods based on historical data are used to complete the data, and the processed data is integrated to generate standardized action data.

[0021] Specifically, the data acquisition and preprocessing module is responsible for collecting real-time three-dimensional posture data of the user's cervical spine, thoracic spine, lumbar spine, and related joints, and generating final motion data after standardizing this data. This module includes multiple processing steps to ensure that the acquired data is accurate, reliable, and meets the needs of subsequent analysis and training.

[0022] In the standardization process, the first step is to transform the 3D pose data of each user joint into a unified reference coordinate system. This process is achieved through rotation matrices and displacement vectors, aiming to eliminate coordinate system differences caused by different devices or acquisition environments. Assume the original 3D pose data of the user joints... Data is acquired in a local coordinate system, with the goal of transforming the data to a standardized global reference coordinate system. The conversion can be performed using the following formula:

[0023] in, This is data in the target reference coordinate system. It is a rotation matrix used to eliminate rotational differences between coordinate systems. It is a displacement vector used to adjust the origin of the coordinate system.

[0024] Time alignment is performed on the motion data acquired from multiple joints. Since there may be time discrepancies during data acquisition, it is necessary to synchronize the motion data of each joint. In this embodiment, adaptive filtering or low-pass filtering methods are used to smooth and align the time-series data. For each joint... Its time Motion data at time After time alignment, the following is obtained:

[0025] in, It is the time point after the target is aligned. It is aligned data processed by filtering methods.

[0026] During data alignment, it is also necessary to normalize the range of motion of each joint to ensure that the data dimensions are consistent across different joints. By normalizing the range of motion of each joint, the motion data of all joints are ensured to be within a uniform quantitative range, making subsequent analysis more accurate and stable.

[0027] During data acquisition, data loss or anomalies may sometimes occur due to equipment limitations or environmental factors, resulting in data incompleteness. Therefore, this embodiment employs an interpolation algorithm to complete the data. Assume that at a certain moment... Motion data of a certain joint Missing values ​​can be filled in using the following linear interpolation method:

[0028] in, and These represent the known data points before and after the missing data points, respectively. The interpolation formula is calculated based on the data at the time points before and after the missing data points to ensure the smoothness and continuity of the data.

[0029] Through the above processing steps, the generated standardized motion data provides a reliable data foundation for subsequent analysis, modeling, and personalized training strategy generation. The data acquisition and preprocessing module effectively removes noise and fills in missing data during real-time acquisition and processing, ensuring high data quality and continuity, and providing accurate input for subsequent motion recognition and prediction.

[0030] The micro-deviation analysis module generates a user-specific micro-deviation accumulation model based on motion data, predicts micro-deviations in real time, and generates personalized training adjustment strategies based on the user's physiological characteristics, historical training data, and current actions. Furthermore, generating the micro-bias accumulation model includes: Extract joint position, angle, motion speed, and acceleration features from standardized motion data; Time series modeling is performed on the extracted feature data to form a feature sequence for each action stage; Nonlinear system modeling methods are used to train the feature sequences to establish a micro-bias accumulation model; By combining model parameters with users' physiological characteristics and historical training data, a user-personalized micro-bias accumulation model is formed.

[0031] Furthermore, real-time prediction of micro-biases includes: Receive standardized user action data and a micro-deviation accumulation model; Real-time motion data is matched with a micro-bias accumulation model, and feature sequences are compared for each motion stage. A recurrent neural network is used to calculate the real-time feature sequence and predict the micro-deviation value at the next time step. The prediction results are used to generate a sequence of action deviations, which are then used to generate personalized training and adjustment strategies.

[0032] Furthermore, generating personalized training adjustment strategies includes: Receive real-time predicted micro-deviation sequences and user physiological characteristic data; Analyze the micro-deviation sequence to identify deviations in the amplitude, angle, and sequence of movements; The training adjustment parameters are determined by matching the motion deviation with the user's historical training data; Generate training adjustment strategies, including adjusting the range of motion, the sequence of motions, and auxiliary prompts.

[0033] Specifically, the micro-deviation analysis module generates a personalized micro-deviation accumulation model based on the user's motion data and predicts micro-deviations in real time. Simultaneously, it generates personalized training adjustment strategies based on the user's physiological characteristics, historical training data, and current movements. Through a series of calculations and modeling processes, this module dynamically adjusts the user's motion training plan to optimize their movement posture, prevent sports injuries, and improve exercise efficiency.

[0034] The process of generating a micro-bias cumulative model begins with the extraction of standardized motion data. Specifically, it first requires extracting key features such as the user's joint positions, angles, movement speeds, and accelerations from the standardized motion data. To accurately reflect the dynamic changes of each joint during the user's movement, feature extraction algorithms are used to extract the spatial positions of the joints from the continuous motion data. ,angle ,speed and acceleration These feature data are crucial for constructing micro-bias models. For each joint... The feature data can be represented as:

[0035] After extracting this data, the next step is to perform time series modeling on the feature data to form a feature sequence for each action stage. The goal of time series modeling is to process action data in chronological order, enabling the model to capture the dynamic changes at each stage of the action. By using appropriate modeling methods, such as recurrent neural networks (RNNs) or long short-term memory networks (LSTMs), feature sequences can be generated for each action stage, accurately representing the user's motion state at different time steps.

[0036] A nonlinear system modeling approach is employed to train the feature sequences and establish a micro-bias accumulation model. Nonlinear modeling methods are suitable for handling complex dynamic systems, such as joint movements in human motion. By training the model, the system can learn the nonlinear relationships of joint movements and predict the micro-biases generated at different stages of the movement. This process is accomplished by optimizing the parameters of the micro-bias model, with the optimization objective being to minimize the model's prediction error and ensure the accuracy of the micro-bias model.

[0037] To create a personalized micro-bias accumulation model, the model parameters need to be combined with the user's physiological characteristics and historical training data. The user's physiological characteristics (such as joint flexibility and muscle strength) and their historical training data (such as past athletic performance and movement patterns) significantly influence the micro-bias of movement and therefore need to be considered in the model. By adjusting these parameters, a micro-bias accumulation model tailored to each user's physiological conditions and exercise habits can be generated.

[0038] For micro-bias prediction, the system first receives standardized motion data from the user and a pre-trained micro-bias accumulation model. Then, by matching real-time motion data with the micro-bias accumulation model, the system compares the feature sequences of each motion stage and calculates the real-time feature sequences using a recurrent neural network. Specifically, it uses an RNN or LSTM model to perform temporal prediction on the real-time motion data and calculates the predicted micro-bias values. The prediction results are then applied to the generation of subsequent training and adjustment strategies.

[0039] Based on real-time predicted micro-deviation values, the system can generate movement deviation sequences. These sequences reflect the deviation of the user's current movements, thus influencing the generation of personalized training adjustment strategies. Specifically, the system determines adjustment strategies by analyzing deviations in the amplitude, angle, and sequence of movements. For example, if the angle deviation of a certain joint is too large, the system may suggest that the user adjust their posture to avoid overloading; if the range of motion of a certain joint is too small, it may suggest increasing the range of motion to improve training effectiveness.

[0040] Based on the above analysis, the system will generate personalized training adjustment strategies. These strategies include adjusting movement range and sequence, and providing auxiliary prompts. Adjusting movement range and sequence helps users correct bad habits during exercise, ensuring proper form; while auxiliary prompts provide real-time feedback to guide users in optimizing their exercise methods and avoiding potential injuries.

[0041] By analyzing the subtle deviations in a user's movements in real time and combining them with the user's physiological characteristics and historical training data, it is possible to effectively tailor personalized training adjustment strategies for each user. This helps users maintain optimal posture during exercise, optimize exercise results, and reduce the risk of injury caused by improper posture or overexertion.

[0042] The action adjustment module is used to predict the local load anomalies caused by the user's next action based on the training adjustment strategy, and generate optimization instructions to update the training action by adjusting the action amplitude, sequence and virtual auxiliary prompts. Furthermore, predicting localized load anomalies caused by the user's next action includes: Perform joint-by-joint and stage-by-stage analysis on motion data to calculate the load state of each joint in the next motion time step; The load prediction model is used to calculate the local stress of each joint to determine the possible location and magnitude of abnormal loads. Generate a sequence of local load anomalies to guide adjustments to the magnitude, sequence, and auxiliary prompts of actions.

[0043] Specifically, the motion adjustment module predicts potential localized load anomalies in the user's next movement based on their training adjustment strategy, generates optimization instructions, and then adjusts the amplitude and sequence of training movements, as well as providing virtual assistance prompts. The goal of this module is to prevent sports injuries and improve exercise performance by dynamically adjusting movement posture and training load.

[0044] Predicting localized load anomalies caused by a user's next action requires detailed, segmented analysis of the motion data. Based on the collected user motion data, the module first divides each motion cycle into several stages and then analyzes the motion of each joint step by step. For example, for each joint... Its load state in the next time step can be expressed as:

[0045] in, Indicates joint The force state at the next time step and Represent the spatial position, velocity, and acceleration of the joint, respectively; functions This is a mapping function for calculating the load state. By analyzing the load on each joint at different stages of motion, the system can assess its potential stress in the next motion.

[0046] A load prediction model is used to accurately calculate the local forces on each joint, thereby determining the location and magnitude of potential load anomalies. In this embodiment, a load prediction model trained based on historical data can be used. This model utilizes features such as joint position, angle, velocity, and acceleration, combined with historical motion data, to predict the potential load anomalies that each joint may face in the next time step. For example, if the model calculates that the force on a certain joint exceeds the normal range, it may mean that the joint is overloaded or unbalanced during motion.

[0047] Local load anomalies can be calculated using the following formula:

[0048] in, Indicates joint The load anomaly occurred in the next time step. To predict the actual load, This represents the expected load under normal conditions. Using this method, the system can determine which joints might experience abnormal loads in the next movement.

[0049] By comprehensively analyzing the stress on each joint, the system can generate a sequence of localized load anomalies, serving as a basis for subsequent adjustments to training movements. This data sequence includes potential load anomalies at different time steps for each joint, and the training problems these anomalies may cause. Based on this data, the system can guide athletes to adjust the range and sequence of movements and provide virtual assistance prompts, thereby optimizing training effectiveness.

[0050] Based on the generated abnormal load data sequence, the motion adjustment module can generate corresponding optimization instructions to guide users on how to adjust their movement patterns. These optimization instructions include adjusting the range of motion, adjusting the sequence of movements, and providing real-time assistance prompts. Specifically, when excessive load is detected on a joint, the system can suggest that the user reduce the range of motion of that joint or change its movement sequence to avoid continuing to bear excessive pressure. When unbalanced movements are detected, the system can provide real-time assistance prompts to help users adjust their posture or use appropriate training equipment to reduce the risk of injury.

[0051] The motion adjustment module can dynamically adjust the user's exercise training plan through accurate local load prediction and personalized training adjustment strategies. This avoids the problem of excessive local load, helps users improve training results, and ensures safety during exercise.

[0052] The collaborative optimization module identifies abnormal coupling of multiple body parts based on user action data, generates collaborative optimization strategies, and transmits optimization instructions to the action adjustment module. Furthermore, identifying multi-site motion coupling anomalies includes: Receive real-time user action data and micro-deviation accumulation model; Correlation analysis is performed on the motion data of each joint to calculate the coupling relationship between joint angles, velocities, and accelerations; By comparing the inter-joint coupling relationship with historical movement patterns, abnormal coupling situations can be identified.

[0053] Specifically, the collaborative optimization module identifies abnormal coupling between multiple body parts based on the user's real-time motion data, generates corresponding collaborative optimization strategies, and then transmits these optimization instructions to the motion adjustment module to further optimize training results. By precisely analyzing the relationships between various joints in the user's motion data, the collaborative optimization module detects and corrects potential abnormal couplings, ensuring coordination among all body parts during movement, thereby reducing injuries or inefficiencies caused by uncoordinated movements.

[0054] The collaborative optimization module receives real-time user motion data and a micro-bias accumulation model. Real-time motion data includes information such as the user's joint positions, angles, speeds, and accelerations during training. The micro-bias accumulation model is a predictive model built based on historical data and personalized physiological characteristics, used to analyze potential deviations in the user's movements. By combining these two types of data, the collaborative optimization module can comprehensively understand the user's actual state during exercise and provide the necessary foundation for subsequent analysis.

[0055] The module performs correlation analysis on the motion data of each joint to calculate the coupling relationships between joint angles, velocities, and accelerations. Joint coupling refers to how multiple joints coordinate their movements in time and space when performing the same action. For example, the angle changes, velocity synchronization, and accelerations between the shoulder and elbow joints, and the hip and knee joints, should exhibit certain regularities under a given motion pattern. When the coordination between these joints becomes abnormal, it may lead to uneven stress on certain joints, resulting in sports injuries or reduced efficiency.

[0056] The collaborative optimization module can use the following mathematical model to model and analyze the coupling relationship between joints:

[0057] in, Indicates joint and joints At time step Coupling relationship at time, and Joints and joints Location, and For its speed, and Its acceleration. Function This represents the coupling feature mapping function between joints, which can be used to quantify the cooperative relationship between joints.

[0058] Once the coupling relationships between joints are calculated, the collaborative optimization module compares these relationships with historical motion patterns to identify abnormal coupling. In this step, the system evaluates whether the current movement exhibits uncoordinated or abnormal coupling based on known normal motion patterns from historical training data. If the current movement deviates significantly from historical patterns, it may indicate that the movements of certain joints are no longer coordinated, leading to abnormal coupling.

[0059] For example, the angle and velocity of one joint may be inconsistent with those of other joints, or unnatural acceleration changes may occur in the movement of certain joints; these could indicate coupling anomalies. By analyzing these discrepancies, the system can effectively identify potential motion coordination problems.

[0060] Based on the identified abnormal coupling, the collaborative optimization module generates collaborative optimization strategies and passes these optimization instructions to the motion adjustment module. These optimization instructions include how to adjust the range of motion and sequence of joint movements, and how to improve the coordination between joints. Through real-time adjustment feedback, the system can effectively improve the user's training movements and avoid injuries caused by improper movement coordination.

[0061] The collaborative optimization module, through precise analysis of inter-joint coupling relationships, combined with a micro-deviation accumulation model and historical data, can detect and correct abnormal couplings in user movements in real time, ensuring coordination and consistency of multiple parts during training, optimizing exercise effects, and effectively reducing the risk of sports injuries.

[0062] The model update module is used to dynamically update the micro-bias accumulation model and training adjustment strategy based on the real-time action data of each training session; Furthermore, updating the micro-bias accumulation model and training adjustment strategies includes: It receives real-time user action data, predicted micro-bias sequences, and personalized training adjustment strategies; The real-time motion data is compared with the micro-deviation accumulation model to calculate new micro-deviation accumulation parameters; The parameter values ​​of the micro-bias accumulation model are updated by combining the user's historical training data and action deviation. The training strategy parameters, including movement amplitude, sequence, and auxiliary cue information, are adjusted based on the updated micro-bias accumulation model.

[0063] Specifically, the model update module dynamically updates the micro-bias accumulation model and training adjustment strategy based on the user's real-time action data, predicted micro-bias sequences, and personalized training adjustment strategies during each training session. In this way, the system can continuously optimize the personalized model and strategy according to the training progress to adapt to the user's training needs and physiological changes, ensuring that all data and strategies remain in optimal condition during training.

[0064] In its implementation, the model update module first receives real-time motion data from the user. This data includes multi-dimensional information such as the position, angle, velocity, and acceleration of each joint during training. Additionally, the module receives micro-bias prediction results (i.e., micro-bias sequences) from the previous step, as well as personalized training adjustment strategies generated based on historical data and the current model. The micro-bias sequence represents the deviation between the actual movement and the ideal movement during training; these deviations affect the training effect and the user's motion state.

[0065] The model update module compares the received real-time motion data with the existing micro-bias cumulative model to calculate new micro-bias cumulative parameters. This process identifies anomalies or biases in the motion data by evaluating the difference between the performance of each movement during training and the predictions of the micro-bias model. Specifically, the micro-bias cumulative model includes parameters built based on individual user characteristics (such as joint range of motion, muscle strength, etc.) and historical training data. By comparing these parameters with real-time data, the system can dynamically adjust them to accurately reflect the current training status.

[0066] By combining the user's historical training data with the deviation of their current actions, the model update module updates the parameter values ​​of the micro-bias cumulative model. The key to this step is combining the user's past training data with the current deviation to identify potential patterns and adjust the model parameters using algorithms such as weighted averaging and least squares to more accurately reflect the user's physiological state and training progress. The specific formula can be expressed as:

[0067] in, Indicates at time step The update amount of parameters in the upper infinitesimal cumulative model. It is the training data at the current moment. It is the output predicted by the micro-bias model. This is the learning rate or weighting coefficient. Using this formula, the model can continuously adjust its predictive power to adapt to the user's changing training status.

[0068] Based on the updated micro-bias accumulation model, the model update module will further adjust the parameters of the training strategy. These adjustments include optimizations to movement amplitude, sequence, and auxiliary prompts. For example, if the micro-bias prediction indicates that the movement of a certain joint has excessive deviation, the system may prompt the user to adjust the movement amplitude of that joint or modify the movement sequence to avoid further deviation accumulation. In addition, auxiliary prompts may also be updated in real time according to the current movement status, providing more customized training guidance.

[0069] During the generation of training adjustment strategies, the system not only relies on the updating of the micro-bias model but also incorporates the user's personalized needs and physiological characteristics. For example, for users with limited joint range of motion, the system helps them better complete training movements by adjusting the amplitude and sequence of movements or adding auxiliary prompts. This adaptive adjustment capability ensures the personalization and efficiency of the training process.

[0070] The model update module continuously updates the micro-bias accumulation model and training adjustment strategies by receiving and processing user motion data in real time, thereby achieving dynamic optimization tailored to individual user needs. Through comprehensive analysis of real-time and historical data, the system can precisely adjust model parameters and training strategies, ensuring that each training session is adapted to the user's current physical condition to the greatest extent possible, promoting healthier and more scientific training results. Simultaneously, this dynamic optimization process effectively avoids training inefficiencies or sports injuries caused by training plans that are not adapted to changes in the user's physiological characteristics or athletic abilities.

[0071] The training strategy update module generates new training instructions based on the updated micro-bias accumulation model and training adjustment strategy, and adjusts the training actions and training process in real time.

[0072] Furthermore, real-time adjustments to training movements and processes include: Receive the updated micro-bias accumulation model and training adjustment strategy; calculate the specific action amplitude, sequence, and auxiliary cues for each action phase based on the training adjustment strategy; Generate new training instructions, including adjustments to the range of motion, updates to the sequence of motions, and virtual assistive prompts; New training instructions are transmitted to the motion adjustment module to update the training motion; Monitor the execution of training actions in real time and adjust training process parameters according to new training instructions; The training instructions and training process adjustment information are output for the execution and optimization of actions in the next training cycle.

[0073] Specifically, the training strategy update module generates new training instructions based on the updated micro-bias accumulation model and training adjustment strategy, and adjusts the training actions and process in real time. It dynamically updates the training instructions according to the user's constantly changing action states and training needs during training to ensure the effectiveness and personalization of the training.

[0074] In the specific implementation process, the training strategy update module first receives the updated micro-bias accumulation model and training adjustment strategy from the front-end module. After the dynamic update in the previous stage, the micro-bias accumulation model can accurately reflect the user's current training status and bias accumulation. Based on this information, the training adjustment strategy forms a corresponding adjustment scheme for the action amplitude, sequence, and auxiliary prompts. This information provides the necessary basic data for generating new training instructions.

[0075] Based on the training adjustment strategy, the training strategy update module calculates the specific movement amplitude, sequence, and auxiliary cues for each action phase. In this phase, the system performs a detailed analysis of the execution of each action, combining prediction information provided by the micro-bias accumulation model to calculate the optimal execution method for each action phase. Specifically, for each action phase, the system determines the most suitable movement amplitude, sequence, and whether virtual auxiliary cues are needed for the current training phase. Adjustments to movement amplitude can be based on real-time micro-bias prediction results, such as joint angle deviations and movement speeds; adjustments to the movement sequence are determined based on the user's training progress and the priority of the target task; auxiliary cues are provided through a virtual training assistant to help the user adjust posture, movements, etc.

[0076] Based on these calculations, the training strategy update module generates new training instructions, which consist of three main parts: amplitude adjustment, sequence update, and virtual assistance prompts. Amplitude adjustment is typically performed when the user's deviation is significant, to help the user complete the movement more accurately; sequence update is based on the user's training goals and performance to ensure optimal movement flow; and virtual assistance prompts provide feedback based on real-time movement feedback, guiding the user to correct posture or adjust movements when the user deviates from the ideal training mode.

[0077] New training instructions are transmitted to the motion adjustment module for real-time updates to the user's training movements. Based on the received instructions, the motion adjustment module adjusts the execution of the training movements in real time, ensuring that the movements in each training cycle align with the user's personalized training plan and goals. This timely transmission and updating of training instructions allows users to receive optimized training guidance based on the latest training data in each training cycle.

[0078] The training strategy update module is also responsible for monitoring the execution of training movements in real time and adjusting the parameters of the training process based on new training instructions and feedback. If the user exhibits unexpected behavior or movement deviations during execution, the system can correct them immediately and feed the new adjustment information back to the training process to further optimize the effectiveness of subsequent training. The system monitors key parameters such as joint angles, movement speed, and acceleration to determine whether the user is training according to the adjusted movement instructions and adjusts the training strategy based on the monitored real-time data feedback.

[0079] The training strategy update module outputs training instructions and training process adjustment information for the execution and optimization of movements in the next training cycle. This information includes not only update instructions such as movement range and sequence, but may also include potential risks that the user may encounter during training (such as movements that are too large or too small, uneven stress on joints, etc.). The system will further optimize the training instructions based on this information, so that each training session can achieve further improvement based on the previous round.

[0080] Through the above methods, the training strategy update module can realize a training process that is dynamically adjusted based on real-time data, enabling each user to receive personalized and refined training instructions, thereby improving training effectiveness and reducing the risk of sports injuries.

[0081] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A cervical and lumbar spine rehabilitation training system integrating posture monitoring and motion guidance, characterized in that, Includes the following modules: The data acquisition and preprocessing module is used to collect the user's cervical spine, thoracic spine, lumbar spine and related joints in real time, and generate motion data after standardization processing; The micro-deviation analysis module generates a user-specific micro-deviation accumulation model based on motion data, predicts micro-deviations in real time, and generates personalized training adjustment strategies based on the user's physiological characteristics, historical training data, and current actions. The action adjustment module is used to predict the local load anomalies caused by the user's next action based on the training adjustment strategy, and generate optimization instructions to update the training action by adjusting the action amplitude, sequence and virtual auxiliary prompts. The collaborative optimization module identifies abnormal coupling of multiple body parts based on user action data, generates collaborative optimization strategies, and transmits optimization instructions to the action adjustment module. The model update module is used to dynamically update the micro-bias accumulation model and training adjustment strategy based on the real-time action data of each training session; The training strategy update module generates new training instructions based on the updated micro-bias accumulation model and training adjustment strategy, and adjusts the training actions and training process in real time.

2. The cervical and lumbar spine rehabilitation training system integrating posture monitoring and motion guidance according to claim 1, characterized in that, The standardization process includes: The three-dimensional pose data of the user's cervical spine, thoracic spine, lumbar spine and related joints are converted into a unified reference coordinate system; The motion data acquired from multiple joints is time-aligned, and the motion data is processed using adaptive filtering or low-pass filtering methods. The motion amplitude of each joint is normalized. For lost or abnormally collected data, interpolation algorithms or prediction methods based on historical data are used to complete the data, and the processed data is integrated to generate standardized action data.

3. The cervical and lumbar spine rehabilitation training system integrating posture monitoring and motion guidance according to claim 1, characterized in that, The generation of the micro-bias accumulation model includes: Extract joint position, angle, motion speed, and acceleration features from standardized motion data; Time series modeling is performed on the extracted feature data to form a feature sequence for each action stage; Nonlinear system modeling methods are used to train the feature sequences to establish a micro-bias accumulation model; By combining model parameters with users' physiological characteristics and historical training data, a user-personalized micro-bias accumulation model is formed.

4. The cervical and lumbar spine rehabilitation training system integrating posture monitoring and motion guidance according to claim 1, characterized in that, The real-time prediction micro-bias includes: Receive standardized user action data and a micro-deviation accumulation model; Real-time motion data is matched with a micro-bias accumulation model, and feature sequences are compared for each motion stage. A recurrent neural network is used to calculate the real-time feature sequence and predict the micro-deviation value at the next time step. The prediction results are used to generate a sequence of action deviations, which are then used to generate personalized training and adjustment strategies.

5. The cervical and lumbar spine rehabilitation training system integrating posture monitoring and motion guidance according to claim 1, characterized in that, The personalized training adjustment strategy includes: Receive real-time predicted micro-deviation sequences and user physiological characteristic data; Analyze the micro-deviation sequence to identify deviations in the amplitude, angle, and sequence of movements; The training adjustment parameters are determined by matching the motion deviation with the user's historical training data; Generate training adjustment strategies, including adjusting the range of motion, the sequence of motions, and auxiliary prompts.

6. The cervical and lumbar spine rehabilitation training system integrating posture monitoring and motion guidance according to claim 1, characterized in that, The predicted local load anomalies caused by the user's next action include: Perform joint-by-joint and stage-by-stage analysis on motion data to calculate the load state of each joint in the next motion time step; The load prediction model is used to calculate the local stress of each joint to determine the possible location and magnitude of abnormal loads. Generate a sequence of local load anomalies to guide adjustments to the magnitude, sequence, and auxiliary prompts of actions.

7. The cervical and lumbar spine rehabilitation training system integrating posture monitoring and motion guidance according to claim 1, characterized in that, The identification of multi-part motion coupling anomalies includes: Receive real-time user action data and micro-deviation accumulation model; Correlation analysis is performed on the motion data of each joint to calculate the coupling relationship between joint angles, velocities, and accelerations; By comparing the inter-joint coupling relationship with historical movement patterns, abnormal coupling situations can be identified.

8. The cervical and lumbar spine rehabilitation training system integrating posture monitoring and motion guidance according to claim 1, characterized in that, The updated micro-bias accumulation model and training adjustment strategy include: It receives real-time user action data, predicted micro-bias sequences, and personalized training adjustment strategies; The real-time motion data is compared with the micro-deviation accumulation model to calculate new micro-deviation accumulation parameters; The parameter values ​​of the micro-bias accumulation model are updated by combining the user's historical training data and action deviation. The training strategy parameters, including movement amplitude, sequence, and auxiliary cue information, are adjusted based on the updated micro-bias accumulation model.

9. The cervical and lumbar spine rehabilitation training system integrating posture monitoring and motion guidance according to claim 1, characterized in that, The real-time adjustment of training movements and training process includes: Receive the updated micro-bias accumulation model and training adjustment strategy; calculate the specific action amplitude, sequence, and auxiliary prompts for each action phase based on the training adjustment strategy; Generate new training instructions, including adjustments to the range of motion, updates to the sequence of motions, and virtual assistive prompts; New training instructions are transmitted to the motion adjustment module to update the training motions; Monitor the execution of training actions in real time and adjust training process parameters according to new training instructions; The training instructions and training process adjustment information are output for the execution and optimization of actions in the next training cycle.

Images