Personalized rehabilitation training method based on state monitoring

Abstract

The invention discloses a personalized rehabilitation training method based on state monitoring, and particularly relates to the technical field of intelligent medical treatment. According to the method, the technical problems of insufficient individuation, poor safety, limited effect and the like caused by single monitoring dimension, lagging adjustment strategy and lack of self-optimization capability in the existing rehabilitation training are solved. According to the scheme, physiological, motion and subjective multi-dimensional data of a patient are synchronously collected through a multi-source sensor and input into a machine learning model for real-time state evaluation after fusion processing, a personalized training scheme is generated and executed accordingly, meanwhile, parameters are dynamically adjusted according to real-time monitoring data in training, and the training efficiency is improved. And the model and the strategy are optimized by analyzing historical data after the period is ended. According to the invention, safe, accurate and self-adaptive evolution personalized rehabilitation training is realized, and the safety, adaptability and rehabilitation effect of training are improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent medical technology, and more specifically, to a personalized rehabilitation training method based on condition monitoring. Background Technology

[0002] In the field of modern clinical rehabilitation medicine, personalized rehabilitation training is a core means to improve patient prognosis and accelerate functional recovery. Its core objective is to dynamically adjust the training plan based on the patient's real-time physiological and functional state to ensure the safety and effectiveness of treatment. With the development of sensor technology, the Internet of Things, and artificial intelligence, the traditional manual adjustment mode, which relies on the therapist's subjective experience, is gradually evolving into a data-driven, automated, and intelligent rehabilitation mode. Limitations of existing technologies:

[0003] 1. The singularity of monitoring dimensions and the isolation of analysis

[0004] Most systems rely on a single type of data, such as focusing only on joint movement angles or a specific physiological parameter, lacking the simultaneous collection and fusion analysis of multi-dimensional information including physiological, movement, and subjective feelings. This one-sided assessment cannot comprehensively and accurately reflect the patient's complex real condition, fails to provide sufficient data support for treatment plan development, and is prone to distorted or misjudged assessment results.

[0005] 2. The lag and rigidity of adjustment strategies

[0006] Even when technical solutions can collect data, their analysis and decision-making often occur before or after training, rather than during the process. Training parameters are usually preset and fixed, or can only be adjusted to a limited extent based on simple thresholds (such as stopping when the threshold is exceeded). There is a lack of a dynamic response mechanism that can optimize in real time, smoothly, and intelligently according to changes in the patient's condition within a safe range. This makes the training process inflexible and difficult to adapt to the fluctuations in the patient's condition.

[0007] 3. Lack of self-optimization capability in the system

[0008] Most existing systems remain at the level of executing preset instructions or providing simple feedback, and their evaluation models and solution libraries are often static after deployment. They cannot learn and evolve from the massive amounts of individual and group data accumulated over a long period of time, nor can they automatically discover the deep correlation between training parameters and rehabilitation effects. Therefore, they cannot make training strategies evolve synchronously with the progress of patient rehabilitation, which limits their long-term application value.

[0009] Therefore, a personalized rehabilitation training method based on state monitoring is proposed to address the above problems. Summary of the Invention

[0010] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide a personalized rehabilitation training method based on status monitoring to solve the problems mentioned in the background art.

[0011] To achieve the above objectives, the present invention provides the following technical solution: a personalized rehabilitation training method based on state monitoring, the method comprising:

[0012] Collect multi-dimensional status monitoring data of patients during rehabilitation training in real time or near real time. The data includes at least two types of physiological parameters, kinematic parameters and subjective feedback parameters.

[0013] The collected multi-dimensional status monitoring data is fused with the patient's basic information to form a comprehensive status assessment vector; this comprehensive status assessment vector is then input into a pre-trained rehabilitation assessment model, and a machine learning algorithm is used to calculate a quantitative indicator representing the patient's current rehabilitation status.

[0014] Based on this quantitative indicator, a personalized rehabilitation training plan containing multiple parameters such as training intensity, training frequency, training mode, training duration, movement range and rest interval is matched from the preset plan library or generated in real time.

[0015] Control the rehabilitation training equipment to execute the personalized rehabilitation training program, and continuously collect and update multi-dimensional status monitoring data during the execution process;

[0016] The training protocol parameters are dynamically adjusted in real time based on updated monitoring data, including early warning adjustments when data exceeds a safety threshold and adaptive adjustments when data indicates a change in the patient's condition.

[0017] After the training cycle ends, the rehabilitation assessment model and training program generation strategy are adaptively optimized based on the correlation analysis between historical monitoring data and training effect indicators.

[0018] Preferably, the physiological parameters include one or more combinations of heart rate, heart rate variability, electromyography (EMG) signal, electroencephalography (EEG) signal, blood oxygen saturation, and body temperature; the kinematic parameters include one or more combinations of joint angle, angular velocity, angular acceleration, displacement, velocity, acceleration, muscle strength, and endurance; and the subjective feedback parameters include one or more of fatigue level, pain level, and comfort level input through a human-computer interaction interface, and these parameters are collected collaboratively by multiple devices in wearable devices, motion capture devices, and interactive terminals.

[0019] Preferably, the rehabilitation assessment model is constructed in the following way: collecting historical rehabilitation training data samples, which contain the correspondence between multi-dimensional state monitoring data and optimal training program parameters; using machine learning algorithms to train the data samples to establish a non-linear mapping relationship from multi-dimensional inputs to program outputs; and periodically updating and optimizing the model parameters with new training data to improve its predictive accuracy and personalization.

[0020] Preferably, the dynamic adjustment includes: setting a safe threshold range based on the patient's individual physiological baseline and rehabilitation stage; immediately triggering an early warning adjustment when the monitoring data exceeds the safe threshold, including reducing training intensity, pausing training, or issuing an alarm; and gradually adjusting the training intensity, frequency, duration, amplitude, or interval according to predetermined rules when the monitoring data is within the safe range but shows a change in status.

[0021] Preferably, the adaptive optimization includes: analyzing the correlation patterns between historical state monitoring data collected during long-term training and training effect indicators; adjusting the internal parameters and weight allocation of the rehabilitation assessment model based on the correlation analysis results; and updating the training program templates and parameter adjustment rules in the program library so that the system can adapt to the dynamic changes in the patient's rehabilitation progress.

[0022] Preferably, it also includes generating a comprehensive rehabilitation training report, which includes trends in status data changes during the training process, records of program adjustments, phased assessment results, future training recommendations, and risk warning information, and is presented to medical professionals and patients through a visual interface.

[0023] Preferably, the calculation and decision-making process of the personalized rehabilitation training program is implemented by executing on a local computing device, processing on a cloud server, or using an end-to-cloud collaborative approach to ensure the real-time performance and reliability of data processing.

[0024] Preferably, the rehabilitation training equipment includes one or more combinations of active training devices, passive training devices, active-passive combined rehabilitation robots, electrical stimulation devices, and virtual reality interactive systems, and the training scheme adaptively adjusts the control strategy according to the equipment type.

[0025] Preferably, the entire training process forms a closed-loop feedback system of monitoring, evaluation, program generation, execution, and re-monitoring. Through multiple iterations, it continuously approaches the patient's optimal rehabilitation training trajectory, achieving truly personalized adaptive rehabilitation training.

[0026] The technical effects and advantages of this invention are as follows:

[0027] Compared with existing technologies, this invention achieves a revolutionary improvement in the rehabilitation training process by constructing a complete technical solution integrating multi-dimensional state monitoring, real-time artificial intelligence assessment and dynamic decision-making, and closed-loop adaptive optimization. Its core advantage lies in employing multi-source heterogeneous sensor fusion technology to simultaneously collect physiological, kinematic, and subjective feedback data. Through feature fusion, it constructs a comprehensive assessment vector that fully reflects the patient's state, solving the problem of inaccurate assessment based on a single data dimension. Technically, the system utilizes a pre-trained and continuously optimized machine learning model to perform deep analysis of the fused data, generating precise personalized training plans, and controlling the execution of rehabilitation equipment through IoT technology. Its significant advantage lies in establishing a real-time, two-layer dynamic adjustment mechanism that ensures training safety (early warning adjustment) while smoothly optimizing the training experience (adaptive adjustment), enabling the training process to flexibly respond to real-time changes in the patient's state. More importantly, the system possesses strong self-evolution capabilities. By analyzing the correlation between historical training data and effects, it continuously optimizes the assessment model and plan library, allowing the training strategy to evolve synchronously with the patient's rehabilitation process, achieving true long-term personalization. Attached Figure Description

[0028] Figure 1 This is a system framework diagram of the present invention.

[0029] Figure 2 This is a flowchart of the personalized rehabilitation training process of the present invention.

[0030] Figure 3 This is a flowchart of the multidimensional data fusion and intelligent evaluation process of the present invention.

[0031] Figure 4 This is a flowchart of the adaptive optimization and system evolution of the present invention. Detailed Implementation

[0032] The following will refer to the appendices in the embodiments of the present invention. Figure 1-4 The technical solutions in the embodiments of the present invention are clearly and completely described herein. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0033] Example 1:

[0034] A personalized rehabilitation training method based on state monitoring, the method comprising:

[0035] Collect multi-dimensional status monitoring data of patients during rehabilitation training in real time or near real time. The data includes at least two types of physiological parameters, kinematic parameters and subjective feedback parameters.

[0036] The collected multi-dimensional status monitoring data is fused with the patient's basic information to form a comprehensive status assessment vector; this comprehensive status assessment vector is then input into a pre-trained rehabilitation assessment model, and a machine learning algorithm is used to calculate a quantitative indicator representing the patient's current rehabilitation status.

[0037] Based on this quantitative indicator, a personalized rehabilitation training plan containing multiple parameters such as training intensity, training frequency, training mode, training duration, movement range and rest interval is matched from the preset plan library or generated in real time.

[0038] Control the rehabilitation training equipment to execute the personalized rehabilitation training program, and continuously collect and update multi-dimensional status monitoring data during the execution process;

[0039] The training protocol parameters are dynamically adjusted in real time based on updated monitoring data, including early warning adjustments when data exceeds a safety threshold and adaptive adjustments when data indicates a change in the patient's condition.

[0040] After the training cycle ends, the rehabilitation assessment model and training program generation strategy are adaptively optimized based on the correlation analysis between historical monitoring data and training effect indicators.

[0041] The system first collects various state data of patients during the training process through a distributed sensor network. These sensor networks can include wearable devices, environmental sensors, and dedicated monitoring devices, which work together to obtain comprehensive patient state information.

[0042] The collected multi-source data undergoes data fusion processing, integrating it with the patient's basic information and historical training records. Feature extraction and dimensionality reduction techniques are used to form a feature vector that comprehensively reflects the patient's condition. This feature vector is input into a pre-trained machine learning model, which analyzes complex nonlinear relationships and outputs a comprehensive quantitative evaluation index. Based on this evaluation index, the system employs an intelligent matching algorithm to select the most suitable training scheme from a dynamic scheme library for the current condition, or generates a completely new personalized scheme in real time.

[0043] Once the plan is finalized, the rehabilitation training equipment is controlled via an IoT protocol to execute the training plan, and changes in the patient's condition are continuously monitored during execution. Monitoring data is fed back to the system in real time, and the system dynamically optimizes the training parameters based on preset adjustment strategies.

[0044] In this embodiment, the physiological parameters include one or more combinations of heart rate, heart rate variability, electromyography signal, electroencephalography signal, blood oxygen saturation, and body temperature; the kinematic parameters include one or more combinations of joint angle, angular velocity, angular acceleration, displacement, velocity, acceleration, muscle strength, and endurance; and the subjective feedback parameters include one or more of fatigue level, pain level, and comfort level input through the human-computer interaction interface. These parameters are collected collaboratively by multiple devices in wearable devices, motion capture devices, and interactive terminals.

[0045] Physiological parameter monitoring is achieved through medical-grade wearable devices. These devices employ non-invasive or minimally invasive sensing technologies, enabling long-term continuous monitoring of key physiological indicators. Kinematic parameter acquisition utilizes motion capture systems based on inertial measurement units, optical motion capture devices, and mechanical sensors. These devices can accurately measure various dynamic and kinematic characteristics during patient movement. Subjective feedback parameters are collected through a specially designed human-computer interaction interface. The interface employs intuitive visual analog scales and natural language interaction, allowing patients to easily express their feelings.

[0046] All acquisition devices ensure data time consistency through a unified time synchronization protocol, use advanced signal processing algorithms to filter and denoise the raw data, and employ anomaly detection algorithms to identify and remove abnormal data points, thereby ensuring the accuracy and reliability of the acquired data.

[0047] In this embodiment, the rehabilitation assessment model is constructed in the following way: collecting historical rehabilitation training data samples, which contain the correspondence between multi-dimensional state monitoring data and optimal training scheme parameters; using machine learning algorithms to train the data samples to establish a non-linear mapping relationship from multi-dimensional inputs to scheme outputs; periodically updating and optimizing the model parameters with new training data to improve its prediction accuracy and personalization. The model training uses a combination of various machine learning algorithms, including deep neural networks, support vector machines, and ensemble learning, and selects the best model structure through cross-validation and hyperparameter optimization techniques.

[0048] Special attention is paid to the unique characteristics of medical data during training, employing specialized data preprocessing methods to address missing values ​​and imbalanced data. After model deployment, an online learning mechanism is established to periodically update the model incrementally with new training data, enabling continuous learning and improvement. To adapt to the characteristics of different patient groups, transfer learning techniques are used to tailor the general model to specific patient groups or individuals.

[0049] By combining the prediction results of multiple models using model ensemble methods, the accuracy and robustness of the evaluation are improved. Simultaneously, a model performance monitoring system is established to continuously track the model's prediction performance, automatically triggering a retraining process when a decline in model performance is detected.

[0050] In this embodiment, the dynamic adjustment includes: setting a safe threshold range based on the patient's individual physiological baseline and rehabilitation stage; immediately triggering an early warning adjustment when monitoring data exceeds the safe threshold, including reducing training intensity, pausing training, or issuing an alarm; and gradually adjusting the training intensity, frequency, duration, amplitude, or interval according to predetermined rules when monitoring data is within the safe range but shows a change in status. The dynamic adjustment mechanism employs an intelligent hierarchical decision-making architecture to achieve both safety and adaptability in the training process. The system first dynamically sets the safe threshold range based on the patient's individual physiological characteristics and current rehabilitation stage using an adaptive algorithm. This threshold range automatically adjusts as the patient's rehabilitation progresses.

[0051] A real-time data monitoring pipeline is established to continuously analyze the collected monitoring data. When data exceeds a safety threshold, an early warning process is immediately initiated. The early warning process employs a tiered response strategy, taking different adjustment measures based on the degree and duration of exceeding the threshold. These measures include gradually reducing training intensity, temporarily suspending training, switching training modes, or issuing multi-level alarms.

[0052] For situations where the status changes but remain within a safe range, the system employs a fuzzy logic-based decision-making mechanism to smoothly and gradually adjust the training parameters based on the trend and magnitude of the status changes. The adjustment decision considers the synergistic effects of multiple factors and uses a multi-objective optimization algorithm to ensure that the adjustment scheme meets both training effectiveness requirements and patient safety.

[0053] All adjustment decisions were meticulously recorded and added to the decision log, which was used for subsequent effect evaluation and decision optimization. Throughout the adjustment process, interactive communication with the patient was maintained, and adjustment information was promptly fed back through the human-machine interface to ensure the patient understood the changes in their training status.

[0054] In this embodiment, the adaptive optimization includes: analyzing the correlation patterns between historical state monitoring data collected during long-term training and training effect indicators; adjusting the internal parameters and weight allocation of the rehabilitation assessment model based on the correlation analysis results; and updating the training program templates and parameter adjustment rules in the program library to enable the system to adapt to the dynamic changes in the patient's rehabilitation progress. The system establishes a dedicated data warehouse to store multi-dimensional data accumulated during long-term training, and employs a distributed storage architecture to ensure data security and scalability. The data analysis phase utilizes various data mining techniques, including association rule analysis, cluster analysis, and time series analysis, to deeply explore the complex relationship between training data and rehabilitation effects. Causal relationship inference algorithms are used to identify key factors affecting training effects and determine the interaction mechanisms between these factors.

[0055] Based on the analysis results, a parameter optimization algorithm was used to fine-tune the rehabilitation assessment model, adjusting the feature weights and decision parameters within the model to more accurately reflect the patient's actual condition. Simultaneously, an evolutionary mechanism for the training protocols was established, optimizing templates and rules in the protocol library by analyzing the characteristics of successful training cases.

[0056] In this embodiment, a comprehensive rehabilitation training report is also generated. This report includes trends in status data during the training process, records of program adjustments, phased assessment results, future training recommendations, and risk warning information. It is presented to medical professionals and patients through a visual interface. The rehabilitation training report generation system employs advanced visualization analysis technology to present comprehensive training information. The report content covers multi-dimensional data throughout the training process, displaying trends in various monitoring indicators through time-series charts, and visually showcasing multiple dimensions of training effectiveness using heatmaps and radar charts.

[0057] The report meticulously records all decisions made to adjust the training program, including the reasons for the adjustments, the content of the adjustments, and their effects. The periodic assessment section employs a comprehensive scoring system to quantitatively evaluate the training effectiveness from multiple perspectives and demonstrates rehabilitation progress through trend analysis.

[0058] Future training recommendations are based on historical data analysis and predictive models, providing personalized optimization suggestions for training programs. The risk warning section uses color coding and warning levels to identify potential risk factors and provides corresponding preventative measures. Report generation combines template-based and personalized approaches, ensuring both standardized report content and meeting individualized needs.

[0059] The report output supports multiple formats and display methods, allowing for dynamic exploration through an interactive interface, and also supports the generation of standardized document formats. The reporting system also provides data comparison functionality, supporting horizontal and vertical comparisons of training effects at different time points, helping medical professionals and patients comprehensively understand rehabilitation progress.

[0060] In this embodiment, the calculation and decision-making process of the personalized rehabilitation training program is implemented through execution on a local computing device, processing on a cloud server, or using an edge-cloud collaborative approach, ensuring the real-time performance and reliability of data processing. The calculation and decision-making system employs an intelligent distributed architecture to balance computational efficiency and real-time requirements. The local computing device is responsible for handling tasks with high real-time requirements, including data acquisition preprocessing, real-time monitoring, and emergency response, ensuring low-latency execution of critical functions.

[0061] Cloud servers handle large-scale data processing and complex computational tasks, including historical data analysis, machine learning model training, and optimization algorithm execution. Data synchronization and task coordination between the local machine and the cloud are achieved through secure communication protocols, and data compression and differential transmission technologies are used to optimize network transmission efficiency.

[0062] Computational tasks are intelligently allocated based on real-time requirements, computational complexity, and data sensitivity, establishing a load balancing mechanism to ensure stable system operation. Security mechanisms include multiple protection measures such as data transmission encryption, access control, and operation log auditing to ensure patient privacy and data security.

[0063] The system establishes a fault detection and automatic recovery mechanism, which can automatically switch tasks when a computing node fails, ensuring service continuity. Computing resources are elastically scaled according to actual needs, automatically expanding resources during peak training periods and releasing excess resources during idle periods, thereby optimizing resource utilization.

[0064] In this embodiment, the rehabilitation training equipment includes one or more combinations of active training devices, passive training devices, combined active and passive rehabilitation robots, electrical stimulation devices, and virtual reality interactive systems. The training scheme adaptively adjusts the control strategy according to the equipment type. The computing and decision-making system employs an intelligent distributed architecture to balance computational efficiency and real-time requirements. Local computing devices handle tasks with high real-time requirements, including data acquisition preprocessing, real-time monitoring, and emergency response, ensuring low-latency execution of critical functions. Cloud servers undertake large-scale data processing and complex computational tasks, including historical data analysis, machine learning model training, and optimization algorithm execution.

[0065] Data synchronization and task coordination are achieved between the local machine and the cloud via a secure communication protocol, and data compression and differential transmission technologies are used to optimize network transmission efficiency. Computational tasks are intelligently allocated based on real-time requirements, computational complexity, and data sensitivity, and a load balancing mechanism is established to ensure stable system operation.

[0066] The security mechanism includes multiple protection measures such as data transmission encryption, access control, and operation log auditing to ensure patient privacy and data security.

[0067] The system establishes a fault detection and automatic recovery mechanism, which can automatically switch tasks when a computing node fails, ensuring service continuity.

[0068] Computing resources are elastically scaled according to actual needs, automatically expanding resources during peak training periods and releasing excess resources during idle periods, thereby optimizing resource utilization.

[0069] In this embodiment, the entire training process forms a closed-loop feedback system of monitoring, evaluation, program generation, execution, and re-monitoring. Through multiple iterative cycles, it continuously approaches the patient's optimal rehabilitation training trajectory, achieving truly personalized adaptive rehabilitation training. The intelligent closed-loop feedback system continuously optimizes the training process through the collaborative work of multiple functional modules. The system establishes a complete data flow pipeline, forming a seamless closed loop from data acquisition, processing, and analysis to decision-making.

[0070] Each training session generates new data, which is used to optimize decisions for the next training session, forming a progressively improving iterative process. Reinforcement learning algorithms explore the effects of different training strategies to gradually discover the optimal training trajectory best suited for each individual patient.

[0071] Digital twin technology is used to create virtual models of patients, predicting the potential effects of different training programs through simulation to aid in the decision-making process. The system employs a multi-objective optimization approach to balance the requirements of multiple objectives, including training effectiveness, safety, and comfort.

[0072] The human-machine collaborative decision-making mechanism combines the experience and knowledge of professional medical personnel with data-driven insights to ensure that decisions are both scientific and in line with clinical practice. The system builds a knowledge base to accumulate successful training cases and best practices, and provides reference plans for new patients through case-based reasoning technology.

[0073] The workflow of this invention:

[0074] Distributed wearable biosensors, high-precision motion capture units, and intelligent interactive terminals synchronously and continuously collect multi-dimensional physiological signals, kinematic parameters, and subjective feedback data from patients during rehabilitation training. The acquired multi-source heterogeneous data undergoes preprocessing, including time alignment, filtering, denoising, and feature extraction, before being fused into a feature vector representing the patient's real-time state. This vector is then input into a pre-trained machine learning evaluation model, which outputs a comprehensive quantitative assessment of the patient's current functional level, fatigue level, and rehabilitation potential through deep learning. Based on this dynamic assessment, the system's intelligent decision engine matches or generates a personalized training plan from an adaptive plan library, including parameters such as training mode, intensity, duration, and intervals. This plan is then distributed to the rehabilitation robot and electroacupuncture device via a standardized protocol. The system uses devices such as stimulators or virtual reality systems as execution terminals. During execution, the system makes real-time judgments based on continuously monitored data streams. If the data exceeds a safety threshold, protective interventions such as load reduction, pause, or alarms are immediately triggered. If the data is within a safe range but shows fluctuations, the parameters are progressively optimized and adjusted according to preset rules to balance training load and patient tolerance. After a single training cycle, the system automatically starts an offline optimization process. It uses big data correlation analysis technology to uncover causal patterns between training parameters and effects, and uses incremental learning algorithms to update the evaluation model and solution library. This enables the system to make more accurate decisions in subsequent cycles, thereby building a highly personalized rehabilitation training ecosystem that integrates continuous monitoring, intelligent evaluation, dynamic execution, closed-loop feedback, and autonomous evolution.

[0075] Finally, the following points should be noted: First, in the description of this application, it should be noted that, unless otherwise specified and limited, the terms "installation", "connection", and "linkage" should be interpreted broadly, and can be mechanical or electrical connections, or internal connections between two components, or direct connections. "Up", "down", "left", "right", etc. are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may change.

[0076] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other.

[0077] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. 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 personalized rehabilitation training method based on state monitoring, characterized in that... The method includes: Collect multi-dimensional status monitoring data of patients during rehabilitation training in real time or near real time. The data includes at least two types of physiological parameters, kinematic parameters and subjective feedback parameters. The collected multi-dimensional status monitoring data is fused with the patient's basic information to form a comprehensive status assessment vector; this comprehensive status assessment vector is then input into a pre-trained rehabilitation assessment model, and a machine learning algorithm is used to calculate a quantitative indicator representing the patient's current rehabilitation status. Based on this quantitative indicator, a personalized rehabilitation training plan containing multiple parameters such as training intensity, training frequency, training mode, training duration, movement range and rest interval is matched from the preset plan library or generated in real time. Control the rehabilitation training equipment to execute the personalized rehabilitation training program, and continuously collect and update multi-dimensional status monitoring data during the execution process; The training protocol parameters are dynamically adjusted in real time based on updated monitoring data, including early warning adjustments when data exceeds a safety threshold and adaptive adjustments when data indicates a change in the patient's condition. After the training cycle ends, the rehabilitation assessment model and training program generation strategy are adaptively optimized based on the correlation analysis between historical monitoring data and training effect indicators.

2. The personalized rehabilitation training method based on state monitoring according to claim 1, characterized in that... The physiological parameters include one or more combinations of heart rate, heart rate variability, electromyography (EMG) signals, electroencephalography (EEG) signals, blood oxygen saturation, and body temperature. The kinematic parameters include one or more combinations of joint angles, angular velocity, angular acceleration, displacement, velocity, acceleration, muscle strength, and endurance. The subjective feedback parameters include one or more of fatigue level, pain level, and comfort level input through a human-computer interaction interface. These parameters are collected collaboratively by multiple devices in wearable devices, motion capture devices, and interactive terminals.

3. The personalized rehabilitation training method based on state monitoring according to claim 1, characterized in that... The rehabilitation assessment model is constructed in the following way: collecting historical rehabilitation training data samples, which contain the correspondence between multi-dimensional state monitoring data and optimal training program parameters; using machine learning algorithms to train the data samples to establish a non-linear mapping relationship from multi-dimensional inputs to program outputs; and periodically updating and optimizing the model parameters with new training data to improve its predictive accuracy and personalization.

4. The personalized rehabilitation training method based on state monitoring according to claim 1, characterized in that... The dynamic adjustment includes: setting a safe threshold range based on the patient's individual physiological baseline and rehabilitation stage; immediately triggering early warning adjustments when the monitoring data exceeds the safe threshold, including reducing training intensity, pausing training, or issuing an alarm; and gradually adjusting the training intensity, frequency, duration, amplitude, or interval according to predetermined rules when the monitoring data is within the safe range but shows a change in status.

5. The personalized rehabilitation training method based on state monitoring according to claim 1, characterized in that... The adaptive optimization includes: analyzing the correlation patterns between historical state monitoring data collected during long-term training and training effect indicators; adjusting the internal parameters and weight allocation of the rehabilitation assessment model based on the correlation analysis results; and updating the training program templates and parameter adjustment rules in the program library so that the system can adapt to the dynamic changes in the patient's rehabilitation progress.

6. The personalized rehabilitation training method based on state monitoring according to claim 1, characterized in that... It also includes generating a comprehensive rehabilitation training report, which contains trends in status data during the training process, records of program adjustments, phased assessment results, future training recommendations, and risk warning information, and is presented to medical professionals and patients through a visual interface.

7. The personalized rehabilitation training method based on state monitoring according to claim 1, characterized in that... The calculation and decision-making process of the personalized rehabilitation training program is implemented by executing on local computing devices, processing on cloud servers, or using an end-to-cloud collaborative approach, ensuring the real-time performance and reliability of data processing.

8. The personalized rehabilitation training method based on state monitoring according to claim 1, characterized in that... The rehabilitation training equipment includes one or more combinations of active training devices, passive training devices, active-passive combined rehabilitation robots, electrical stimulation devices, and virtual reality interactive systems, and the training program adaptively adjusts the control strategy according to the type of equipment.

9. The personalized rehabilitation training method based on state monitoring according to claim 1, characterized in that... The entire training process forms a closed-loop feedback system of monitoring, evaluation, program generation, execution, and re-monitoring. Through multiple iterations, it continuously approaches the patient's optimal rehabilitation training trajectory, achieving truly personalized adaptive rehabilitation training.

Images