A rehabilitation training control method based on digital OT evaluation and training

Abstract

The application provides a rehabilitation training control method based on digital OT evaluation and training, and relates to the technical field of digital evaluation, and comprises the following steps: pre-perception of a target user before rehabilitation training is performed, a user pre-perception matrix is obtained, and an immersive experience rehabilitation module is constructed synchronously; rehabilitation training big data mining is performed, a multi-dimensional rehabilitation training analysis domain is established, rehabilitation training control decisions are made, and a rehabilitation training control space is obtained; dynamic decomposition disturbance aggregation learning is performed according to a rehabilitation training evaluation event set, a digital OT evaluator is established; multi-channel rehabilitation evaluation optimization is performed, and multiple rehabilitation training optimization groups are established; reproduction evolution iterative optimization is performed, and a rehabilitation training optimization result is determined. The application solves the technical problem that the rehabilitation training scheme in the prior art is often designed based on a general training mode, lacks targeted optimization for individual needs of different users, and thus leads to uncertainty and inefficiency of the rehabilitation process.

Description

Technical Field

[0001] This invention relates to the field of digital assessment technology, and more specifically to a rehabilitation training control method based on digital OT assessment and training. Background Technology

[0002] Rehabilitation training is a training process that helps individuals restore or improve physical function. It typically targets physical dysfunction caused by injury, illness, or other reasons. It primarily uses a series of planned activities to help people regain motor skills, increase muscle strength, improve coordination, and enhance daily living abilities. Traditional rehabilitation training often relies on standardized training procedures and fixed training plans, which frequently fail to make dynamic adjustments based on the user's real-time condition, rehabilitation progress, and physical feedback. This uniform training approach leads to delayed and unsuitable training effects, making it difficult to accurately support and continuously optimize the user's rehabilitation process. Furthermore, it easily results in overtraining or undertraining, leading to inefficiency or potential harm. Summary of the Invention

[0003] This application provides a rehabilitation training control method based on digital OT assessment and training, which aims to solve the technical problem that existing rehabilitation training programs are often designed based on general training models and lack targeted optimization for the needs of different individual users, resulting in uncertainty and inefficiency in the rehabilitation process.

[0004] This application discloses a rehabilitation training control method based on digital OT assessment and training. The method includes: performing pre-rehabilitation training perception on the target user to obtain a user pre-perception matrix, and simultaneously constructing an immersive experience rehabilitation module; performing rehabilitation training big data mining based on the user pre-perception matrix to establish a multi-dimensional rehabilitation training analytical domain, and making rehabilitation training control decisions based on the multi-dimensional rehabilitation training analytical domain to obtain a rehabilitation training control space; performing dynamic decomposition, perturbation, and aggregation learning based on a rehabilitation training assessment event set to establish a digital OT evaluator, which includes a functional recovery gain assessment channel, a rehabilitation training risk assessment channel, and a rehabilitation training participation assessment channel; based on the immersive experience rehabilitation module, performing multi-channel rehabilitation assessment optimization on the rehabilitation training control space according to the digital OT evaluator to establish multiple rehabilitation training optimization groups; and performing iterative optimization through the immersive experience rehabilitation module, the digital OT evaluator, and the multi-dimensional rehabilitation training analytical domain to determine the rehabilitation training optimization result.

[0005] One or more technical solutions provided in this application have at least the following beneficial effects: By conducting pre-rehabilitation training perception analysis on target users, a user pre-perception matrix is ​​generated, providing foundational data for subsequent personalized rehabilitation training. Combined with an immersive rehabilitation module, users can participate in training within a virtual environment. This interactive experience not only enhances the fun of training but also allows for dynamic adjustments to training content based on real-time feedback. This highly interactive training method strengthens user engagement and participation. Through big data mining of the pre-perception matrix, a multi-dimensional rehabilitation training analytical domain is generated, providing decision support for training control. By dynamically decomposing perturbation aggregation learning, a digital OT evaluator is established, enabling multi-channel evaluation. These channels assess functional recovery gain, rehabilitation training risk, and participation, ensuring comprehensive monitoring of rehabilitation training effectiveness from multiple dimensions. This not only improves the accuracy of the evaluation but also provides... Training optimization provides rich data support; by combining the immersive rehabilitation module with the digital OT evaluator, the rehabilitation training control space is evaluated through multiple channels to determine the most suitable training plan for the user. This adaptive adjustment allows the training process of each user to be dynamically optimized according to their current condition, thereby achieving the best rehabilitation effect. By breeding and evolving multiple rehabilitation training optimization groups and continuously conducting multiple rounds of iterative optimization, the training plan is constantly adjusted and optimized. This iterative optimization mechanism can identify and correct the deficiencies in the training plan during the training process, so as to maximize the training effect. Through multi-dimensional evaluation and intelligent optimization mechanisms, the goals, intensity and methods of rehabilitation training can be automatically adjusted to avoid the uncertainty of manual intervention, making rehabilitation training more scientific, systematic and more in line with the user's needs.

[0006] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0007] Figure 1 This application provides a schematic flowchart of a rehabilitation training control method based on digital OT assessment and training.

[0008] Figure 2 This application provides a schematic diagram of the rehabilitation training big data mining process in a rehabilitation training control method based on digital OT assessment and training. Detailed Implementation

[0009] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0010] like Figure 1 As shown in the embodiment of this application, a rehabilitation training control method based on digital OT assessment and training is provided, the method comprising: Conduct pre-rehabilitation training perception for target users, obtain the user pre-perception matrix, and simultaneously construct an immersive experience rehabilitation module.

[0011] Before rehabilitation training begins, pre-treatment perception is conducted on the target user. The resulting pre-treatment perception dataset includes basic user information, user monitoring information, current training target information, and training environment information. This perception data is collected through legal and compliant channels, meeting data protection and privacy requirements to ensure data security and reliability. The pre-treatment perception dataset undergoes cleaning, including removing noise or invalid data, and is organized into a structured matrix to obtain the user's pre-treatment perception matrix. Based on this matrix, digital twin modeling is performed, creating a virtual model of the user. This virtual model reflects the user's state in the real world, providing an immersive rehabilitation training environment.

[0012] Based on the user's prior perception matrix, big data mining of rehabilitation training is performed to establish a multidimensional rehabilitation training analytical domain, and rehabilitation training control decisions are made based on the multidimensional rehabilitation training analytical domain to obtain the rehabilitation training control space.

[0013] By leveraging a user's prior perception matrix for big data analysis, key factors influencing rehabilitation outcomes are identified. This data is then analyzed and modeled to establish a multidimensional rehabilitation training analytical domain—a multidimensional data space—to analyze various characteristics of rehabilitation training. This domain encompasses multiple dimensions, such as functional recovery, rehabilitation training risks, and participation in rehabilitation training. Decisions are made based on this established multidimensional analytical domain. Specifically, reasoning and optimization are performed based on these multidimensional data inputs to determine different rehabilitation training control schemes. These schemes involve varying training methods, intensity, duration, and frequency. Personalized rehabilitation training plans are provided based on the user's specific needs and condition, forming a rehabilitation training control space containing multiple alternative schemes.

[0014] A digital OT evaluator is established by dynamically decomposing perturbation aggregation learning based on the set of rehabilitation training assessment events. The digital OT evaluator includes a functional recovery gain assessment channel, a rehabilitation training risk assessment channel, and a rehabilitation training participation assessment channel.

[0015] The rehabilitation training assessment event set consists of a series of key assessment indicators related to rehabilitation effectiveness, such as progress in functional recovery, training risks, and patient participation. These indicators help assess the achievements and challenges faced by users during the rehabilitation process. Dynamic decomposition-perturbation aggregation learning involves dynamically analyzing the rehabilitation training assessment event set. It uses perturbation aggregation learning to decompose and aggregate the data. Here, perturbation refers to modeling and learning about potential interfering factors during training, such as environmental changes and changes in user behavior. Through multiple training iterations and data processing, these perturbation factors are analyzed to ultimately determine their impact on different assessment channels.

[0016] This process allows for the dynamic capture of changes during rehabilitation and the creation of a digital OT assessor. The digital OT assessor is an intelligent system capable of multi-dimensional evaluation based on input data. Specifically, the functional recovery gain assessment channel evaluates the user's progress and effectiveness in functional recovery; the rehabilitation training risk assessment channel assesses potential risks during rehabilitation, such as overtraining and injury risk; and the rehabilitation training participation assessment channel evaluates the user's initiative and consistency in participating in rehabilitation training, with a focus on user engagement.

[0017] Based on the immersive experience rehabilitation module, the rehabilitation training control space is optimized through multi-channel rehabilitation assessment using the digital OT evaluator, and multiple rehabilitation training optimization groups are established.

[0018] The rehabilitation training program is simulated and virtually tested using an immersive rehabilitation experience module. This process simulates a real training environment and user interaction, resulting in multiple sets of virtual rehabilitation training data. This virtual data is then input into a digital OT evaluator, which assesses the training process across different dimensions, including functional recovery gain, training risk, and engagement. This results in a multi-channel rehabilitation assessment graph, with each channel representing a different assessment indicator. After obtaining the multi-channel assessment graph, multi-dimensional optimization is performed based on the weights and thresholds of each assessment channel. For example, the control space is optimized based on the functional recovery gain assessment threshold to select the optimal rehabilitation training program. Based on this multi-channel assessment, multiple optimized programs are grouped into rehabilitation training optimization clusters. Each cluster represents a specific training optimization result, reflecting the optimal rehabilitation training program across multiple dimensions, including functional recovery, training risk, and engagement.

[0019] The immersive rehabilitation module, the digital OT evaluator, and the multidimensional rehabilitation training analytical domain are used to perform iterative optimization of the multiple rehabilitation training optimization groups to determine the optimization results of rehabilitation training.

[0020] Propagation and evolution involve iterative optimization through repeated training sessions simulating user experiences in a virtual environment. The immersive rehabilitation module simulates interactions between different rehabilitation training programs and the user, generating new virtual data. This new virtual data is then input into a digital OT evaluator for re-evaluation, resulting in new assessment results. Utilizing a multidimensional rehabilitation training analytical domain allows for more precise analysis and optimization of each rehabilitation training optimization group. This domain provides greater control and analysis, ensuring the optimization process covers all relevant training factors. Iterative optimization involves continuously adjusting different parameters through multiple rounds of training and evaluation, and providing feedback on the training results to drive continuous improvement of the training program. After multiple rounds of propagation and evolution, a comprehensive weight allocation is performed based on the rehabilitation adaptability (i.e., rehabilitation effect) of each program, ultimately yielding a comprehensive score for each training program. Based on this comprehensive score, the optimal rehabilitation training result—the most suitable rehabilitation training program for the target user—is determined.

[0021] Furthermore, such as Figure 2 As shown, based on the user's prior perception matrix, big data mining of rehabilitation training is performed to establish a multi-dimensional rehabilitation training analytical domain, including: Rehabilitation training big data retrieval is performed based on the user's prior perception matrix to obtain a matching rehabilitation training sample set; confidence evaluation and cleaning are performed based on the matching rehabilitation training sample set to obtain a confidence rehabilitation training sample set; and multidimensional rehabilitation training control trigger feature parsing is performed based on the confidence rehabilitation training sample set to generate the multidimensional rehabilitation training parsing domain.

[0022] The core of the big data retrieval process, based on the user's prior perception matrix, is to match this matrix with an existing rehabilitation training sample library to find samples most similar to the target user's training needs, health status, and recovery goals. Sample matching uses similarity metrics such as Euclidean distance and cosine similarity to determine the similarity between the user and historical rehabilitation samples. For example, if the target user's prior perception matrix is ​​very close to data in certain rehabilitation samples, such as user age, physical condition, injury type, or training goals, then these samples will be selected. Through big data retrieval, the set of training samples that best matches the target user is found from a large amount of historical data, forming the matched rehabilitation training sample set.

[0023] Confidence assessment involves analyzing and evaluating the quality of the matched sample set to eliminate samples that do not meet the standards, including outliers, incomplete data, or inapplicable parts of historical data. Samples with low confidence are eliminated or marked as outliers. The cleaned sample set is called the confidence rehabilitation training sample set, which contains high-quality and reliable training data.

[0024] Based on a confidence-based rehabilitation training sample set, multidimensional rehabilitation training control trigger feature analysis is performed. The goal is to extract multidimensional features from the sample set, representing key factors in rehabilitation training. Each feature corresponds to different rehabilitation training parameters, such as training intensity, training duration, range of motion, and training method. Based on these features, trigger feature analysis is conducted to identify factors that decisively influence the effectiveness of rehabilitation training. For example, some users exhibit better functional recovery under specific rehabilitation methods, or certain environmental settings significantly impact training outcomes. Based on this feature analysis, data analysis techniques, such as principal component analysis and cluster analysis, are used to extract a key multidimensional feature space, ultimately generating a multidimensional rehabilitation training analytic domain. This multidimensional rehabilitation training analytic domain is a multidimensional data space where each dimension represents different rehabilitation training characteristics, providing a multidimensional data-based decision support framework for subsequent rehabilitation training control.

[0025] Furthermore, a digital OT evaluator is established by dynamically decomposing perturbation aggregation learning based on the rehabilitation training assessment event set, including: The rehabilitation training assessment event set is dynamically decomposed to obtain a historical set of functional recovery gain assessment, a historical set of rehabilitation training risk assessment, and a historical set of rehabilitation training participation assessment. Multi-scale perturbation aggregation learning is performed on the historical set of functional recovery gain assessment to generate the functional recovery gain assessment channel. Multi-scale perturbation aggregation learning is performed on the historical set of rehabilitation training risk assessment to generate the rehabilitation training risk assessment channel. Multi-scale perturbation aggregation learning is performed on the historical set of rehabilitation training participation assessment to generate the rehabilitation training participation assessment channel. The functional recovery gain assessment channel, the rehabilitation training risk assessment channel, and the rehabilitation training participation assessment channel are connected as parallel nodes to generate the digital OT evaluator.

[0026] Through dynamic decomposition, the rehabilitation training assessment event set is separated and organized according to different assessment dimensions. Among them, the functional recovery gain assessment history set is based on the assessment data of the user's recovery progress and effect during the rehabilitation process. It mainly reflects the user's rehabilitation progress and functional recovery under a specific training program. The rehabilitation training risk assessment history set contains data related to risks in the training process, such as the frequency of sports injuries, the possibility of overtraining, and the inappropriateness of training intensity. The rehabilitation training participation assessment history set is data related to user participation, reflecting the user's activity level, participation time, and training willingness in rehabilitation training.

[0027] Multi-scale perturbation aggregation learning refers to perturbing data at multiple scales, such as time scale and training intensity scale, to enhance the model's ability to capture potential patterns and changes in the data. For historical sets of functional recovery gain assessments, multi-scale perturbation aggregation learning can deeply explore the changing patterns of functional recovery gain, analyzing not only the impact of a single dimension but also the synergistic effects of multiple factors. Specifically, perturbation analysis of the historical set of functional recovery gain assessments involves adjusting some training parameters or contextual factors to simulate the possible gains in user functional recovery under different training scenarios. Perturbation aggregation refers to merging these different perturbations, such as different training methods, intensities, and frequencies, to identify which training methods and parameter combinations are most effective for functional recovery. Based on the results of multi-scale perturbation analysis, a functional recovery gain assessment channel is trained, which can evaluate the level of functional recovery gain of users in real time during different rehabilitation training sessions and dynamically adjust it according to the user's actual situation.

[0028] Similar to the analysis of functional recovery gains, for the historical set of rehabilitation training risk assessment, multi-scale perturbation aggregation learning is used to identify and assess the risks of different training programs. Training risks come from multiple aspects, such as excessive training intensity, non-standard training movements, and lack of supervision. Through perturbation analysis, these potential risk factors are identified, and a rehabilitation training risk assessment channel is constructed based on training history and feedback data.

[0029] For the historical set of rehabilitation training participation assessment, multi-scale perturbation aggregation learning is also used to analyze user participation. Participation is not only related to the user's initiative and level of engagement, but also closely related to factors such as the suitability of training and feedback mechanisms. Through perturbation analysis, we can identify which factors have a greater impact on user participation and construct a rehabilitation training participation assessment channel.

[0030] By connecting the functional recovery gain assessment channel, rehabilitation training risk assessment channel, and rehabilitation training participation assessment channel in parallel, it is possible to simultaneously evaluate and optimize the user's rehabilitation process from multiple dimensions—recovery, risk, and participation—rather than relying solely on the assessment results of a single dimension. These parallel assessment channels combine to form a digital OT assessor, which integrates multiple factors to provide a more comprehensive and accurate evaluation of rehabilitation training programs.

[0031] Furthermore, based on the historical set of functional recovery gain assessments, multi-scale perturbation aggregation learning is performed to generate the functional recovery gain assessment channel, including: The data distribution characteristics are analyzed based on the historical set of functional recovery gain assessment to obtain sample distribution characteristics; multi-scale perturbation feature decision-making is performed based on the sample distribution characteristics to obtain multiple perturbation feature strategies; the historical set of functional recovery gain assessment is perturbed in multiple rounds based on the multiple perturbation feature strategies to obtain multiple functional recovery gain assessment perturbation sets; adversarial training is performed on the multiple functional recovery gain assessment perturbation sets respectively to obtain multiple functional recovery gain assessment models; and aggregation learning is performed on the multiple functional recovery gain assessment models based on the multiple adversarial training loss characteristics to establish the functional recovery gain assessment channel.

[0032] The distribution characteristics of the historical data set for functional recovery gain assessment are analyzed. That is, statistical analysis is performed on the functional recovery status of different users to uncover patterns and trends in the data. The main goal is to identify the data characteristics of functional recovery gain, including: functional recovery speed, such as whether the recovery is fast or stable; whether there are significant fluctuations or instabilities during the recovery process; and how individual differences among users, such as age and underlying health conditions, affect the recovery process.

[0033] Multi-scale perturbation feature decision-making is based on the characteristics of sample distribution. This means perturbing and optimizing the functional recovery gain at multiple scales. Multi-scale perturbation feature decision-making integrates multiple factors and generates multiple perturbation feature strategies. For example, how to adjust the content of rehabilitation training under different training intensities, how to adjust the training intensity at different time periods, and whether it is necessary to adjust the strategy according to the different individual characteristics of users.

[0034] Multiple rounds of perturbation are applied to the historical set of functional recovery gain assessments based on the obtained perturbation feature strategies. Each round of perturbation represents an adjustment to training parameters or methods, such as adjusting exercise intensity, training duration, and frequency. Through multiple rounds of perturbation, the different performances of functional recovery gain under different training conditions can be simulated. Multiple perturbation sets of functional recovery gain assessments are generated through these rounds of perturbation. These sets contain functional recovery data under different perturbation conditions, and each perturbation set represents the possible changes and effects of the rehabilitation process under a specific perturbation strategy.

[0035] Adversarial training is a deep learning technique that aims to improve the robustness and accuracy of a model by generating adversarial examples—perturbed data samples. In this process, adversarial training is performed on multiple perturbation sets for functional recovery gain (FGG) assessment. This allows the model to distinguish and adapt to FGG data under various perturbation conditions, ultimately generating multiple FGG assessment models, each capable of evaluating a user's rehabilitation process based on a specific perturbation set. Through adversarial training, the model can make accurate predictions of recovery gain under different training conditions.

[0036] During adversarial training, each model generates a loss value, or prediction error, which reflects the model's performance. By comparing the adversarial training loss characteristics of multiple models, we can identify which models perform well in specific situations and which models may have overfitting or large errors. Based on the loss characteristics of multiple adversarial training models, we employ ensemble learning techniques to aggregate the outputs of multiple functional recovery gain assessment models, forming a more powerful assessment model. The aggregation process includes methods such as weighted averaging and ensemble learning, thereby improving the overall assessment effect. Through ensemble learning, a comprehensive functional recovery gain assessment channel is ultimately formed. This channel can accurately assess the user's functional recovery under various perturbation conditions and provide real-time feedback for rehabilitation training.

[0037] Furthermore, based on the immersive experience rehabilitation module, and according to the digital OT evaluator, the rehabilitation training control space is optimized through multi-channel rehabilitation assessment, establishing multiple rehabilitation training optimization groups, including: The immersive rehabilitation experience module virtually evolves each rehabilitation training control scheme within the rehabilitation training control space to obtain multiple sets of virtual rehabilitation training data. These virtual data are then input into the digital OT evaluator to obtain a multi-channel rehabilitation assessment map. Based on the multi-channel assessment map, the rehabilitation training control space is optimized for functional recovery gain according to a functional recovery gain assessment threshold to obtain a first rehabilitation training optimization group. Based on the multi-channel assessment map, the rehabilitation training control space is optimized for rehabilitation training risk according to a rehabilitation training risk assessment threshold to obtain a second rehabilitation training optimization group. Based on the multi-channel assessment map, the rehabilitation training control space is optimized for rehabilitation training participation according to a rehabilitation training participation assessment threshold to obtain a third rehabilitation training optimization group. The first, second, and third rehabilitation training optimization groups are then added to the multiple rehabilitation training optimization groups.

[0038] Virtual evolution means that each rehabilitation training program is evolved and optimized virtually through simulation. The immersive experience rehabilitation module is the core of virtual evolution. It generates virtual rehabilitation training data for each training program through simulation technology. This virtual data reflects the changes and performance of the training program at different rehabilitation stages, similar to training in reality, but simulated through a virtual environment.

[0039] Multiple sets of virtual rehabilitation training data are input into the digital OT evaluator. The digital OT evaluator analyzes these virtual data to generate a multi-channel rehabilitation assessment map. The multi-channel rehabilitation assessment map includes the functional recovery gain assessment coefficient, rehabilitation training risk assessment coefficient, and rehabilitation training participation assessment coefficient corresponding to each rehabilitation training control program, indicating the advantages and disadvantages of each rehabilitation training control program.

[0040] A functional recovery gain assessment threshold is set, representing the minimum required functional recovery gain. Any training program with a functional recovery gain below this threshold is deemed ineffective and fails to meet rehabilitation needs. Based on the functional recovery gain assessment coefficients in the multi-channel rehabilitation assessment atlas, the rehabilitation training control space is optimized for functional recovery gain. This involves selecting all training programs from the atlas that have a gain greater than the functional recovery gain assessment threshold, retaining only those programs with good recovery effects. After this selection, a new set of training programs, the first rehabilitation training optimization group, is obtained. These programs possess high functional recovery gains and meet the requirements.

[0041] A rehabilitation training risk assessment threshold is established, representing the minimum required risk level for rehabilitation training. Any training program with a risk level exceeding this threshold is deemed ineffective, potentially too dangerous, or unsuitable for the user. Based on the rehabilitation training risk assessment coefficients in a multi-channel rehabilitation assessment graph, the rehabilitation training control space is optimized for risk assessment. All training programs with risk assessment coefficients below the threshold are selected, indicating lower risk and suitability for user rehabilitation. This selection forms a second optimization group, encompassing all training programs that meet the risk assessment threshold requirements.

[0042] The rehabilitation training participation assessment threshold is another set standard, indicating that the user's participation in the rehabilitation training program should be higher than this threshold. If the participation is too low, it means that the training program does not match the user's interest or physical condition, and cannot be maintained for a long time. Based on the rehabilitation training participation assessment coefficient in the multi-channel rehabilitation assessment map, the rehabilitation training control space is optimized to select all training programs with participation assessment coefficients higher than the participation assessment threshold. These programs have high user participation, can effectively mobilize user enthusiasm, and help to maintain consistent rehabilitation training. After screening, a third rehabilitation training optimization group is formed, which includes all training programs with high participation. These programs can enhance user engagement and ensure the continuity of training.

[0043] The first, second, and third rehabilitation training optimization groups represent the best-performing training programs in terms of functional recovery, risk control, and participation optimization, respectively. By merging these three groups, a more comprehensive set of multiple rehabilitation training optimization groups that meet various criteria is generated. The merged optimization groups include all optimized programs, ensuring that the training programs can meet user needs in multiple dimensions.

[0044] Furthermore, the immersive rehabilitation experience module, the digital OT evaluator, and the multidimensional rehabilitation training analytical domain are used to perform iterative optimization of the multiple rehabilitation training optimization groups to determine the rehabilitation training optimization results, including: Based on the immersive experience rehabilitation module and the digital OT evaluator, the first rehabilitation training optimization group is subjected to reproductive evolution optimization according to the multidimensional rehabilitation training analytical domain to obtain a first rehabilitation training evolution optimization set; based on the immersive experience rehabilitation module and the digital OT evaluator, the second rehabilitation training optimization group is subjected to reproductive evolution optimization according to the multidimensional rehabilitation training analytical domain to obtain a second rehabilitation training evolution optimization set; based on the immersive experience rehabilitation module and the digital OT evaluator, the third rehabilitation training optimization group is subjected to reproductive evolution optimization according to the multidimensional rehabilitation training analytical domain to obtain a third rehabilitation training evolution optimization set; weights are allocated based on the multidimensional rehabilitation evaluation indicators of the digital OT evaluator to establish a rehabilitation fitness analytical model; the rehabilitation fitness is iteratively optimized according to the rehabilitation fitness analytical model for the first rehabilitation training evolution optimization set, the second rehabilitation training evolution optimization set, and the third rehabilitation training evolution optimization set to generate the rehabilitation training optimization result.

[0045] The first rehabilitation training optimization group is any one of multiple rehabilitation training optimization groups, serving as the current analysis object. Based on the immersive experience rehabilitation module and digital OT evaluator, the training programs in the first rehabilitation training optimization group are propagated and evolved. This propagation and evolution refers to simulating the process of biological evolution, involving the combination, variation, and selection of programs to generate more optimized training programs. The multidimensional rehabilitation training analytical domain provides more dimensions of control and analysis, ensuring that the optimization process covers all relevant training factors. After propagation and evolution, the final first rehabilitation training evolution optimization set is obtained, containing the programs optimized through evolution. These programs have been improved in multiple dimensions, possessing stronger adaptability and effectiveness, and therefore have greater potential than the programs in the original optimization group.

[0046] Similar to the first rehabilitation training optimization group, the second and third rehabilitation training optimization groups were propagated and evolved through immersive experience rehabilitation modules and digital OT assessors. These schemes were further optimized based on the multidimensional rehabilitation training analytical domain. Finally, through the evolution process, the second and third rehabilitation training evolution optimization sets were generated, which respectively included training schemes optimized in terms of safety and participation.

[0047] Multidimensional rehabilitation assessment indicators include functional recovery gain indicators, rehabilitation training risk indicators, and rehabilitation training participation indicators. These indicators provide evaluation criteria for each rehabilitation training program, indicating the program's performance in different aspects. Weights are assigned to each indicator. For example, by analyzing historical training data, the relative importance of each indicator to the rehabilitation effect is estimated; for instance, functional recovery gain is given a higher weight, while rehabilitation risk and participation are given relatively lower weights. Based on the weight assignment results, a rehabilitation fitness analytical model is established. This model includes a rehabilitation fitness analytical function, which is: Rehabilitation Fitness = Functional Recovery Gain Indicator * Functional Recovery Gain Weight - Rehabilitation Training Risk Indicator * Rehabilitation Training Risk Weight + Rehabilitation Training Participation Indicator * Rehabilitation Training Participation Weight.

[0048] Iterative optimization, based on the results of the rehabilitation fitness analytical model, continuously updates the training scheme to optimize the final rehabilitation effect. This means finding the training scheme with the highest rehabilitation fitness. Specifically, from the current first, second, and third rehabilitation training evolution optimization sets, training schemes with high fitness are selected as the starting point for the next iteration. Various optimization algorithms, such as genetic algorithms and simulated annealing, are used to cross-pollinate and mutate the selected training schemes to generate new ones. The newly generated training schemes are evaluated according to the rehabilitation fitness analytical model, their rehabilitation fitness is calculated, and the training scheme with the highest rehabilitation fitness is selected for the next round of optimization. This process continues for multiple rounds until predetermined stopping conditions are met, such as reaching maximum fitness or no significant improvement. Finally, a set of training schemes with the highest rehabilitation fitness value is selected, representing the optimal rehabilitation path for the target user under multi-dimensional evaluation criteria.

[0049] Furthermore, based on the immersive experience rehabilitation module and the digital OT evaluator, the first rehabilitation training optimization group is subjected to reproductive evolution optimization according to the multidimensional rehabilitation training analytical domain to obtain the first rehabilitation training evolution optimization set, including: The functional recovery gain distribution characteristics are analyzed based on the first rehabilitation training optimization group to obtain functional recovery gain distribution features; reproductive attention allocation is performed on the first rehabilitation training optimization group based on the functional recovery gain distribution features to obtain a first reproductive attention allocation result; based on the multidimensional rehabilitation training analytical domain, the first rehabilitation training optimization group is expanded based on the first reproductive attention allocation result to obtain a first rehabilitation training expansion group; the first rehabilitation training expansion group is virtually evolved based on the immersive experience rehabilitation module to obtain a first training virtual evolution distribution; the first training virtual evolution distribution is input into the digital OT evaluator to obtain multiple training evaluation sequences; training evaluation constraints are constructed based on the functional recovery gain evaluation threshold, rehabilitation training risk evaluation threshold, and rehabilitation training participation evaluation threshold; and the first rehabilitation training expansion group is optimized and identified in combination with the multiple training evaluation sequences to generate the first rehabilitation training evolution optimization set.

[0050] Functional recovery gain refers to the degree of improvement in actual functional recovery brought about by each rehabilitation training program during the user's rehabilitation process. This gain can be the change in a certain physiological or behavioral function before and after the user performs a certain training task, such as joint range of motion, muscle strength, motor coordination, etc.

[0051] By analyzing the functional recovery gains of various training programs in the first rehabilitation training optimization group, the distribution patterns of the gains were identified. Functional recovery gains are represented by the gain value of each training program. Statistical analysis was performed on these gain values, including: distribution pattern analysis to identify whether the gain values ​​follow a normal, skewed, or multimodal distribution; mean and variance analysis to calculate the mean and variance of each training program to assess the overall effect and differences between programs; and cluster analysis to group the gain data according to certain similarity criteria to identify which training programs are effective for specific groups and which are more universally applicable. The output functional recovery gain distribution characteristics are: concentrated within a certain range or fluctuating significantly across the entire distribution interval, which can help further optimize and adjust the selection of training programs.

[0052] In evolutionary algorithms, reproductive attention allocation means determining which training methods should receive more attention, be iterated upon, and optimized, and which can be relatively ignored by weighting the attention given to different training methods. Training methods with high functional recovery gains and significant effects are given higher attention weights, based on the gain distribution characteristics—that is, prioritizing the best-performing training method. The first reproductive attention allocation result obtained in this way includes the attention level of each training method, indicating the extent to which each training method will be amplified or ignored in subsequent optimization iterations.

[0053] In evolutionary algorithms, reproduction and expansion refer to generating new individuals from existing high-performing individuals through operations such as crossover and mutation. In the context of rehabilitation training optimization, expansion refers to generating new variants or extending the scope of existing high-quality training programs. During expansion, optimization is performed based on the multidimensional rehabilitation training analytic domain, for example, by changing the weights or parameters of a certain dimension to generate new training programs.

[0054] Propagation and expansion include: cross-operation, which involves cross-fertilizing currently performing training programs to generate new ones, such as merging different training modules from two programs to create a new combination; and mutation operation, which involves modifying selected training programs by making minor adjustments, such as adding new training movements, adjusting training frequency or intensity, to generate diverse training programs. Based on the previous propagation attention allocation results, programs that receive more attention are further manipulated and mutated during the expansion process. Through propagation and expansion, a new first rehabilitation training expansion group is generated. These newly generated training programs contain more possibilities and optimization dimensions, further promoting the improvement of rehabilitation training.

[0055] For each training program in the first rehabilitation training expansion group, the immersive experience rehabilitation module is used to simulate the rehabilitation training process of different users, evaluate its effectiveness, and generate virtual data corresponding to each training program. This includes users' physiological feedback data, such as muscle activity, movement trajectory, and joint angles, as well as users' behavioral data, such as training duration, fatigue level, and movement completion. The virtual evolution distribution of the first training program is obtained, which represents the virtual data corresponding to the training programs within the first rehabilitation training expansion group. This virtual data describes the effects of different training programs in the virtual environment.

[0056] The digital OT evaluator analyzes the virtual evolution distribution of the first training session and generates evaluation results based on functional recovery gain, risk control, and participation, namely multiple training evaluation sequences. Each training evaluation sequence includes a functional recovery gain evaluation coefficient, a rehabilitation training risk evaluation coefficient, and a rehabilitation training participation evaluation coefficient.

[0057] The functional recovery gain assessment threshold is a minimum required level of functional recovery gain to ensure that the training program can bring about sufficient functional recovery. The rehabilitation training risk assessment threshold is a risk ceiling to avoid excessive risks such as injury or overtraining. The rehabilitation training participation assessment threshold is a participation floor to ensure that the training program can maintain sufficient user engagement. These thresholds serve as constraints to ensure that only training programs that meet all conditions can proceed to the next stage of optimization and evolution. For multiple training assessment sequences, based on the above assessment constraints, training programs that meet all assessment thresholds are optimized and selected to enter the subsequent optimization process. Through multiple rounds of screening and identification, the best training programs that meet the conditions are selected to form the first rehabilitation training evolution optimization set.

[0058] Furthermore, pre-rehabilitation training perception is conducted on target users to obtain a user pre-perception matrix, including: Based on the target user's pre-rehabilitation training perception, a pre-perception dataset is obtained, which includes user basic information, user monitoring information, current training target information, and training environment information; the pre-perception dataset is cleaned to generate the user pre-perception matrix.

[0059] The pre-rehabilitation dataset is designed to comprehensively understand the personalized needs, current status, and training environment of target users. It is a collection of key user data through a series of information collection tools and sensor data. This dataset provides data support for subsequent rehabilitation training and helps to customize personalized rehabilitation plans. The collection of this pre-rehabilitation dataset is carried out through legal and compliant channels and meets data protection and privacy security requirements to ensure the security and reliability of the data.

[0060] The user's basic information includes their age, gender, physical condition, and lifestyle habits. This information helps to understand the user's basic needs and physical condition, allowing for targeted design of rehabilitation training. User monitoring information includes real-time physiological data collected through sensors, wearable devices, or health monitoring systems, such as heart rate, blood pressure, muscle electrical activity, and gait analysis. This helps to track the user's physiological changes in real time and identify problems during rehabilitation training. Current training goal information refers to the user's current rehabilitation goals, such as restoring specific functions, improving motor skills, reducing pain, or improving quality of life. Clear goals help to personalize training plans. Training environment information includes the environmental conditions in which rehabilitation training takes place, such as temperature, humidity, spatial layout, and equipment conditions. The training environment also has a certain impact on the user's training effectiveness.

[0061] Pre-perception datasets often contain missing values, noisy data, or inconsistent records. To ensure the validity and accuracy of the data, the dataset needs to be cleaned, including removing duplicate data, filling in missing values, data correction, and data standardization. After data cleaning, various user information is constructed into a matrix format to generate a user pre-perception matrix, which is used for subsequent data analysis and modeling.

[0062] Furthermore, an immersive rehabilitation experience module is being built concurrently, including: The immersive experience rehabilitation module is obtained by performing digital twin modeling based on the user's prior perception matrix.

[0063] By inputting data from the user's pre-perception matrix into the modeling system and using computer simulation technology, a digital twin model is constructed based on the user's personalized data. This model can not only reflect the user's physical condition but also simulate their reactions in different training environments. The user's digital twin model is embedded into a virtual or immersive environment to simulate different rehabilitation training scenarios, such as gait training, strength training, and coordination training. The resulting immersive rehabilitation module allows users to undergo rehabilitation training in a digital and interactive way, as if they were actually there.

[0064] Furthermore, the multidimensional rehabilitation assessment indicators include functional recovery gain indicators, rehabilitation training risk indicators, and rehabilitation training participation indicators.

[0065] Functional recovery gain indicators are used to assess the actual contribution of rehabilitation training to the functional recovery of the target user; rehabilitation training risk indicators are used to assess the potential risk factors and their impact during rehabilitation training; and rehabilitation training participation indicators are used to assess the target user's participation and enthusiasm during rehabilitation training. The comprehensive application of these indicators helps to provide comprehensive and accurate guidance for the optimization and decision-making of rehabilitation training.

[0066] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A rehabilitation training control method based on digital operational assessment and training, characterized in that, include: Conduct pre-rehabilitation training perception for target users, obtain the user pre-perception matrix, and simultaneously construct an immersive experience rehabilitation module; Based on the user's prior perception matrix, big data mining of rehabilitation training is performed to establish a multidimensional rehabilitation training analytical domain. Based on the multidimensional rehabilitation training analytical domain, rehabilitation training control decisions are made to obtain a rehabilitation training control space. Based on multidimensional data input, reasoning and optimization are performed to determine different rehabilitation training control schemes, involving different training methods, training intensity, duration, and frequency. According to the user's specific needs and status, rehabilitation training schemes are provided, and a rehabilitation training control space is formed, which includes multiple alternative schemes. A digital OT evaluator is established by performing dynamic decomposition, perturbation, and aggregation learning based on the rehabilitation training assessment event set. The digital OT evaluator includes a functional recovery gain assessment channel, a rehabilitation training risk assessment channel, and a rehabilitation training participation assessment channel. The rehabilitation training assessment event set consists of a series of key assessment indicators related to rehabilitation effectiveness, including: progress of functional recovery, training risk, and patient participation. Based on the immersive experience rehabilitation module, the rehabilitation training control space is optimized through multi-channel rehabilitation assessment using the digital OT evaluator, and multiple rehabilitation training optimization groups are established. The immersive experience rehabilitation module, the digital OT evaluator, and the multidimensional rehabilitation training analytical domain are used to perform iterative optimization of the multiple rehabilitation training optimization groups to determine the rehabilitation training optimization results. Specifically, the immersive rehabilitation experience module, the digital OT evaluator, and the multidimensional rehabilitation training analytical domain are used to iteratively optimize the multiple rehabilitation training optimization groups to determine the rehabilitation training optimization results, including: Based on the immersive experience rehabilitation module and the digital OT evaluator, the first rehabilitation training optimization group is subjected to breeding evolution optimization according to the multidimensional rehabilitation training analytical domain to obtain the first rehabilitation training evolution optimization set. Based on the immersive experience rehabilitation module and the digital OT evaluator, the second rehabilitation training optimization group is subjected to breeding evolution optimization according to the multidimensional rehabilitation training analytical domain to obtain the second rehabilitation training evolution optimization set. Based on the immersive experience rehabilitation module and the digital OT evaluator, the third rehabilitation training optimization group is subjected to breeding evolution optimization according to the multidimensional rehabilitation training analytical domain to obtain the third rehabilitation training evolution optimization set. Based on the multidimensional rehabilitation assessment indicators of the digital OT assessor, a weight allocation is performed to establish a rehabilitation fitness analytical model. According to the rehabilitation fitness analytical model, the rehabilitation fitness of the first rehabilitation training evolution optimization set, the second rehabilitation training evolution optimization set, and the third rehabilitation training evolution optimization set is iteratively optimized to generate the rehabilitation training optimization result. This includes conducting pre-rehabilitation training perception assessments on target users to obtain a user pre-perception matrix, including: Based on the target user, a pre-rehabilitation training pre-perception is performed to obtain a pre-perception dataset, which includes user basic information, user monitoring information, current training target information, and training environment information; Clean the pre-perception dataset to generate the user pre-perception matrix; The immersive rehabilitation experience module, which is built concurrently, includes: The immersive experience rehabilitation module is obtained by performing digital twin modeling based on the user's prior perception matrix.

2. The rehabilitation training control method based on digital OT assessment and training as described in claim 1, characterized in that, Based on the aforementioned user prior perception matrix, big data mining of rehabilitation training is performed to establish a multidimensional rehabilitation training analytical domain, including: Based on the user's prior perception matrix, a rehabilitation training big data retrieval is performed to obtain a matching rehabilitation training sample set; Based on the matched rehabilitation training sample set, confidence evaluation and cleaning are performed to obtain a confidence rehabilitation training sample set. Based on the confidence-based rehabilitation training sample set, multidimensional rehabilitation training control trigger feature parsing is performed to generate the multidimensional rehabilitation training parsing domain.

3. The rehabilitation training control method based on digital OT assessment and training as described in claim 1, characterized in that, A digital OT assessor is established by dynamically decomposing and aggregating perturbations based on a set of rehabilitation training assessment events, including: The rehabilitation training assessment event set is dynamically decomposed to obtain a historical set of functional recovery gain assessment, a historical set of rehabilitation training risk assessment, and a historical set of rehabilitation training participation assessment. Based on the historical set of functional recovery gain assessment, multi-scale perturbation aggregation learning is performed to generate the functional recovery gain assessment channel; Based on the historical set of rehabilitation training risk assessment, multi-scale perturbation aggregation learning is performed to generate the rehabilitation training risk assessment channel. Based on the historical set of rehabilitation training participation assessment, multi-scale perturbation aggregation learning is performed to generate the rehabilitation training participation assessment channel; The functional recovery gain assessment channel, the rehabilitation training risk assessment channel, and the rehabilitation training participation assessment channel are connected as parallel nodes to generate the digital OT evaluator.

4. The rehabilitation training control method based on digital OT assessment and training as described in claim 3, characterized in that, Based on the historical set of functional recovery gain assessments, multi-scale perturbation aggregation learning is performed to generate the functional recovery gain assessment channel, including: Based on the historical set of functional recovery gain assessment, the data distribution characteristics are analyzed to obtain the sample distribution characteristics; Based on the sample distribution characteristics, multi-scale perturbation feature decision-making is performed to obtain multiple perturbation feature strategies; The functional recovery gain evaluation history set is perturbed in multiple rounds according to the multiple perturbation feature strategies to obtain multiple functional recovery gain evaluation perturbation sets; Adversarial training is performed on the multiple perturbation sets for functional recovery gain evaluation to obtain multiple functional recovery gain evaluation models. The multiple function recovery gain evaluation models are aggregated and learned based on the characteristics of multiple adversarial training loss to establish the function recovery gain evaluation channel.

5. The rehabilitation training control method based on digital OT assessment and training as described in claim 1, characterized in that, Based on the immersive experience rehabilitation module, the rehabilitation training control space is optimized through multi-channel rehabilitation assessment using the digital OT evaluator, establishing multiple rehabilitation training optimization groups, including: The immersive experience rehabilitation module virtually evolves each rehabilitation training control scheme within the rehabilitation training control space to obtain multiple virtual rehabilitation training data. The multiple virtual rehabilitation training data are input into the digital OT evaluator to obtain a multi-channel rehabilitation assessment map. Based on the multi-channel rehabilitation assessment map, the rehabilitation training control space is optimized for functional recovery gain according to the functional recovery gain assessment threshold to obtain the first rehabilitation training optimization group. Based on the multi-channel rehabilitation assessment map, the rehabilitation training control space is optimized for rehabilitation training risk according to the rehabilitation training risk assessment threshold to obtain a second rehabilitation training optimization group. Based on the multi-channel rehabilitation assessment map, the rehabilitation training control space is optimized according to the rehabilitation training participation assessment threshold to obtain the third rehabilitation training optimization group. Add the first rehabilitation training optimization group, the second rehabilitation training optimization group, and the third rehabilitation training optimization group to the plurality of rehabilitation training optimization groups.

6. The rehabilitation training control method based on digital OT assessment and training as described in claim 1, characterized in that, Based on the immersive experience rehabilitation module and the digital OT evaluator, the first rehabilitation training optimization group is subjected to reproductive evolution optimization according to the multidimensional rehabilitation training analytical domain to obtain the first rehabilitation training evolution optimization set, including: Based on the first rehabilitation training optimization group, the functional recovery gain distribution characteristics are analyzed to obtain the functional recovery gain distribution features; Based on the functional recovery gain distribution characteristics, the first rehabilitation training optimization group is subjected to reproductive attention allocation to obtain the first reproductive attention allocation result. Based on the multidimensional rehabilitation training parsing domain, the first rehabilitation training optimization group is expanded and multiplied according to the first reproductive attention allocation result to obtain the first rehabilitation training expansion group. The first rehabilitation training expansion group is virtually evolved based on the immersive experience rehabilitation module to obtain the first training virtual evolution distribution; The first training virtual evolution distribution is input into the digital OT evaluator to obtain multiple training evaluation sequences; Training evaluation constraints are constructed based on the functional recovery gain assessment threshold, rehabilitation training risk assessment threshold, and rehabilitation training participation assessment threshold. The first rehabilitation training expansion group is then optimized and identified by combining the multiple training evaluation sequences to generate the first rehabilitation training evolution optimization set.

7. The rehabilitation training control method based on digital OT assessment and training as described in claim 1, characterized in that, The multidimensional rehabilitation assessment indicators include functional recovery gain indicators, rehabilitation training risk indicators, and rehabilitation training participation indicators.

Images