CN117457218A

CN117457218A - Interactive rehabilitation training assisting method and system

Info

Publication number: CN117457218A
Application number: CN202311778569.6A
Authority: CN
Inventors: 崔丽娜; 邓南楠
Original assignee: Shenzhen Jianyikang Medical Instrument Technology Co ltd
Current assignee: Shenzhen Jianyikang Medical Instrument Technology Co ltd
Priority date: 2023-12-22
Filing date: 2023-12-22
Publication date: 2024-01-26
Anticipated expiration: 2043-12-22
Also published as: CN117457218B

Abstract

The invention relates to the technical field of computer-aided medical treatment, in particular to an interactive rehabilitation training assisting method and system. According to the invention, the rehabilitation plan is more flexible and accurate through the use of the self-adaptive neural network and the fuzzy logic controller, the training adjustment algorithm based on reinforcement learning can be adjusted according to the specific condition of a patient, the real-time adjustment and optimization of the training plan are ensured, the application of the cross-modal fusion technology introduces multi-sensory stimulation in the rehabilitation training, the effectiveness of the training is enhanced, the use of the dynamic system theory and the nonlinear prediction technology provides deep holes in the aspect of data analysis, the establishment of a more targeted and effective rehabilitation strategy is facilitated, the micro-service architecture and the technical integration of API integration are ensured, and the flexibility, the expandability and the continuous progress of the rehabilitation process are ensured.

Description

Interactive rehabilitation training assisting method and system

Technical Field

The invention relates to the technical field of computer-aided medical treatment, in particular to an interactive rehabilitation training assisting method and system.

Background

The field of computer-assisted medical technology is a high-tech field covering a wide range of applications, with computer technology being used to support, optimize and improve medical and health care procedures. Information technology, biomedical engineering, data science, medicine, and other multidisciplinary knowledge are integrated to develop and apply software, hardware, and other computing tools to assist doctors and medical professionals in diagnosis, treatment, care, and research.

The interactive rehabilitation training assistance method is a part of the technical field of computer-assisted medical treatment and is focused on providing a customized rehabilitation training scheme. The purpose is through technical means, such as virtual reality, motion tracking, biofeedback etc. provides the rehabilitation training that interdynamic and participation are stronger to increase patient's participation and training's effectiveness. Aims to enable patients to participate in the rehabilitation process more actively through the high-tech means, thereby accelerating the rehabilitation progress and improving the rehabilitation effect. The method generally integrates various sensors and tracking devices and combines professional rehabilitation software to realize real-time monitoring and feedback of patient movement. By means of the method, the rehabilitation state of the patient can be accurately estimated, the training program can be adjusted in real time, and an interesting interaction mode is provided to increase the interest and participation of training.

The traditional rehabilitation training method lacks in-depth consideration of individual difference, uses a more universal and fixed training program, and causes poor training effect or does not adapt to the specific requirements of individual patients. The traditional method is relatively backward in real-time adjustment and optimization of training plans, and is difficult to quickly adjust according to the instant performance and feedback of patients. The lack of application of multi-sensory stimulation and cross-modality fusion also places limitations on traditional rehabilitation training in terms of improving patient engagement and training outcome. The traditional method is not fine enough in terms of data analysis and personalized strategy formulation, lacks support of deep learning and complex algorithms, and limits the effectiveness of the traditional method in adapting to rapidly-developed medical technology environments.

Disclosure of Invention

The invention aims to solve the defects in the prior art and provides an interactive rehabilitation training assisting method and system.

In order to achieve the above purpose, the present invention adopts the following technical scheme: the interactive rehabilitation training assisting method comprises the following steps of S1: based on the preliminary medical records and rehabilitation requirements of the patient, performing potential assessment by adopting a deep learning algorithm and time sequence analysis, and generating a rehabilitation potential analysis result;

S2: based on the rehabilitation potential analysis result, a rehabilitation training plan is designed by using a self-adaptive neural network and a fuzzy logic controller, and a preliminary rehabilitation training plan is generated;

s3: based on the preliminary rehabilitation training plan, a training adjustment algorithm based on reinforcement learning is utilized to adjust training difficulty and type according to real-time performance of a patient, and a dynamically adjusted rehabilitation training scheme is generated;

s4: based on the dynamically adjusted rehabilitation training scheme, performing multi-sense stimulation integration by adopting a cross-mode fusion technology to generate a multi-sense fusion training scheme;

s5: based on the multi-sense fusion training scheme, performing data analysis by using a dynamic system theory and a nonlinear prediction technology to generate an optimized rehabilitation scheme;

s6: based on the optimized rehabilitation scheme, applying a micro-service architecture and API integration to perform technology integration, and generating a rehabilitation training architecture integrating a new technology;

s7: based on the rehabilitation training framework integrating the new technology, the natural language processing technology is adopted to analyze the feedback of the patient to perform continuous optimization, and a continuous optimization rehabilitation training framework is generated.

As a further aspect of the present invention, the rehabilitation potential analysis results include comprehensive assessment of patient rehabilitation potential, limiting factor identification, and preliminary rehabilitation advice, the preliminary rehabilitation training plan includes customized training activities, scheduling and expected targets, the dynamically adjusted rehabilitation training scheme is specifically training difficulty and type automatically adjusted based on real-time feedback and scheduling of the patient, the multi-sense fusion training scheme is specifically visual and auditory multi-sense element fusion training activities, the optimized rehabilitation scheme is specifically targeted rehabilitation training strategy, and the continuously optimized rehabilitation training framework is specifically training content and method adjusted according to patient feedback and progress.

As a further scheme of the invention, based on the preliminary medical records and rehabilitation requirements of patients, the potential evaluation is carried out by adopting a deep learning algorithm and time series analysis, and the specific steps for generating the rehabilitation potential analysis result are as follows: preparing data by adopting a data preprocessing technology based on the preliminary medical records of the patient, and generating a data preprocessing result;

s102: based on the data preprocessing result, identifying key rehabilitation indexes by adopting a characteristic extraction technology, and generating a characteristic extraction report;

s103: based on the feature extraction report, analyzing the rehabilitation potential of the patient by using a deep learning algorithm, and generating a deep learning potential analysis result;

s104: based on the deep learning potential analysis result, evaluating the change trend of the rehabilitation potential of the patient by combining a time sequence analysis method, and generating a rehabilitation potential analysis result;

the data preprocessing technology comprises a normalization method, missing value estimation and outlier processing, the feature extraction technology comprises a principal component analysis method and a linear discriminant analysis method, the deep learning algorithm is specifically a convolutional neural network, and the time sequence analysis method comprises an autoregressive model and a moving average model.

As a further scheme of the present invention, based on the rehabilitation potential analysis result, a rehabilitation training program is designed by using an adaptive neural network and a fuzzy logic controller, and the specific steps of generating a preliminary rehabilitation training program are as follows: based on the rehabilitation potential analysis result, classifying the rehabilitation types of the patient by using a data clustering algorithm to generate a patient rehabilitation type classification;

S202: based on the patient rehabilitation type classification, a rule engine is adopted to make a preliminary training strategy according to the rehabilitation type, and a preliminary training strategy draft is generated;

s203: based on the preliminary training strategy draft, generating a training plan optimized by the self-adaptive neural network by using the self-adaptive neural network;

s204: based on the training program optimized by the self-adaptive neural network, performing final fine adjustment and optimization on the training program by using a fuzzy logic controller to generate a preliminary rehabilitation training program;

the data clustering algorithm is specifically a K-means clustering method, the rule engine is specifically a decision tree algorithm, the self-adaptive neural network is specifically a multi-layer perceptron and a back propagation algorithm, and the fuzzy logic controller comprises a fuzzy rule set and an inference mechanism.

As a further aspect of the present invention, based on the preliminary rehabilitation training plan, a training adjustment algorithm based on reinforcement learning is utilized to adjust training difficulty and type according to real-time performance of a patient, and the specific steps of generating a dynamically adjusted rehabilitation training scheme are as follows: simulating a patient rehabilitation training scene by adopting an environment modeling technology based on the preliminary rehabilitation training plan to generate a training scene simulation;

S302: based on the training scene simulation, applying a reinforcement learning algorithm to learn a patient behavior mode and generating a proxy learning model;

s303: based on the agent learning model, implementing a reward mechanism, and generating an optimization training adjustment strategy;

s304: based on the optimized training adjustment strategy, a dynamic programming algorithm is applied to adjust the training difficulty and type, and the method is suitable for real-time performance of patients and generates a dynamically adjusted rehabilitation training scheme;

the environment modeling technology comprises a state space model and a dynamic system simulation, the reinforcement learning algorithm is specifically a Q learning algorithm, the reward mechanism comprises accumulated return optimization and behavior evaluation, and the dynamic planning algorithm is specifically a Bellman equation.

As a further scheme of the invention, based on the dynamically adjusted rehabilitation training scheme, the cross-modal fusion technology is adopted to integrate multi-sensory stimulation, and the specific steps for generating the multi-sensory fusion training scheme are as follows: based on the dynamically adjusted rehabilitation training scheme, analyzing the interaction effect of the multi-sensory stimulation by adopting a data analysis technology, and generating a sensory stimulation interaction analysis result;

s402: based on the sensory stimulus interaction analysis result, a multi-sensory fusion technology is applied to integrate visual and auditory stimuli, and a multi-sensory synchronous scheme is generated;

S403: designing a multi-mode rehabilitation training activity based on the multi-sense synchronous scheme, and generating a multi-mode training activity design;

s404: based on the multi-mode training activity design, integrating multi-sense stimulation content by adopting a sensing weighted fusion technology to generate a multi-sense fusion training scheme;

the data analysis technology is specifically correlation analysis and interactive mode identification, the multi-sense fusion technology is specifically a sense synchronization algorithm and a mode fusion method, the multi-mode rehabilitation training activities comprise audiovisual coordination training and a sense integration strategy, and the sense weighting fusion technology is specifically a weighted average method and a sense stimulation optimization fusion algorithm.

As a further scheme of the invention, based on the multi-sense fusion training scheme, the data analysis is performed by applying a dynamic system theory and a nonlinear prediction technology, and the specific steps for generating the optimized rehabilitation scheme are as follows: based on the multi-sense fusion training scheme, adopting dynamic system modeling to analyze the time sequence characteristics of training data, and generating a dynamic system training data model;

s502: based on the dynamic system training data model, a nonlinear time sequence analysis technology is applied to reveal a complex mode in training data, and a nonlinear time sequence analysis result is generated;

S503: based on the nonlinear time sequence analysis result, predicting rehabilitation progress by using a machine learning prediction model, and generating a rehabilitation progress prediction model;

s504: based on the rehabilitation progress prediction model, comprehensively analyzing data by adopting a decision tree algorithm, optimizing a training plan, and generating an optimized rehabilitation scheme;

the dynamic system modeling is specifically a state space model, the nonlinear time sequence analysis technology is specifically a chaos theory, the machine learning prediction model is specifically a support vector machine, and the decision tree algorithm is specifically a random forest.

As a further scheme of the present invention, based on the optimized rehabilitation scheme, the technical integration is performed by applying the micro-service architecture and API integration, and the specific steps of generating the rehabilitation training architecture of the integrated new technology are as follows: planning a service module of a rehabilitation training system by adopting a micro-service architecture design principle based on the optimized rehabilitation scheme to generate a micro-service architecture design;

s602: based on the micro-service architecture design, integrating each service module by using an API management tool to generate an API integration scheme;

s603: based on the API integration scheme, combining micro-service interface development to generate a micro-service interface development result;

S604: based on the micro-service interface development result, applying a containerization technology to implement system integration, and generating a rehabilitation training framework integrating a new technology;

the micro-service architecture design principle is specifically a field driving design, the API management tool is specifically an API gateway, the micro-service interface development is specifically a RESTful API design, and the containerization technology is specifically a Docker.

As a further scheme of the present invention, based on the rehabilitation training architecture integrated with the new technology, the specific steps of generating a continuously optimized rehabilitation training frame by analyzing patient feedback to perform continuous optimization by adopting a natural language processing technology are as follows: based on the rehabilitation training architecture integrating the new technology, analyzing the written feedback of the patient by adopting a text mining technology, and generating a text analysis result of the feedback of the patient;

s702: based on the feedback text analysis result of the patient, applying emotion analysis technology to evaluate the emotional response of the patient and generating a patient emotion analysis report;

s703: based on the patient emotion analysis report, performing deep analysis by using a natural language processing algorithm, extracting deep patient requirements and suggestions, and generating a deep patient requirement analysis result;

S704: based on the deep patient demand analysis result, integrating the analysis result by adopting a decision support algorithm, continuously optimizing a rehabilitation training scheme, and generating a continuously optimized rehabilitation training frame;

the text mining technology is particularly word frequency-inverse document frequency analysis, the emotion analysis technology is particularly emotion polarity analysis, the natural language processing algorithm is particularly a long-time and short-time memory network, and the decision support algorithm is particularly fuzzy logic and a data-driven decision tree.

The interactive rehabilitation training auxiliary system comprises a rehabilitation potential analysis module, a strategy making and optimizing module, a scene simulation learning module, a dynamic training plan adjusting module, a multi-sense fusion design module, a system modeling progress prediction module, a micro-service architecture integration module and a feedback analysis decision module;

the rehabilitation potential analysis module is used for carrying out analysis and evaluation by combining a deep neural network algorithm and a time sequence analysis method based on the preliminary medical records of the patient and adopting a data preprocessing and feature extraction technology to generate a rehabilitation potential analysis result;

the strategy making and optimizing module makes a preliminary training strategy by utilizing a rule engine based on a rehabilitation potential analysis result, optimizes and fine-tunes a training plan by utilizing a self-adaptive neural network and a fuzzy logic controller, and generates a preliminary rehabilitation training plan;

The scene simulation learning module simulates a training scene by using an environment modeling technology based on a preliminary rehabilitation training plan, learns a patient behavior mode by adopting a reinforcement learning algorithm, optimizes a training plan adjustment strategy and generates a proxy learning model;

the dynamic training plan adjusting module is used for implementing a rewarding mechanism based on the agent learning model, adjusting training difficulty and type through a dynamic planning algorithm, adapting to real-time performance of a patient and generating a dynamically adjusted rehabilitation training scheme;

the multisensory fusion design module is based on a dynamically adjusted rehabilitation training scheme, analyzes the interaction effect of multisensory stimulation by using a multivariate data analysis technology, integrates visual and auditory stimulation by combining the multisensory fusion technology, designs a multisensory training activity, and generates a multisensory fusion training scheme;

the system modeling progress prediction module analyzes the time sequence characteristics of training data by using a dynamic system modeling technology based on a multi-sense fusion training scheme, reveals a complex mode by using a nonlinear time sequence analysis technology, predicts rehabilitation progress by using a machine learning prediction model, and generates a rehabilitation progress prediction model;

the micro-service architecture integration module is used for planning a rehabilitation training system service module based on a rehabilitation progress prediction model by adopting a micro-service architecture design principle, integrating the service module by using an API management tool, and implementing system integration by using a containerization technology to generate a rehabilitation training architecture integrating a new technology;

The feedback analysis decision module is based on a rehabilitation training framework integrating a new technology, adopts a text mining technology to analyze the written feedback of a patient, adopts an emotion analysis technology to evaluate the emotional response of the patient, utilizes a natural language processing technology and a decision support algorithm to analyze an integration result, continuously optimizes a rehabilitation training scheme, and generates a continuously optimized rehabilitation training framework.

Compared with the prior art, the invention has the advantages and positive effects that:

according to the invention, the rehabilitation plan is more flexible and accurate through the use of the self-adaptive neural network and the fuzzy logic controller, and can be adjusted according to the specific condition of a patient. The training adjustment algorithm based on reinforcement learning further ensures real-time adjustment and optimization of the training program so that the training program is more suitable for the current state and the requirements of the patient. The application of the cross-modal fusion technology introduces multi-sensory stimulation in rehabilitation training, so that the training effectiveness is enhanced, and the participation and experience of patients are improved. The use of dynamic system theory and nonlinear prediction techniques provides a deep hole in data analysis, helping to formulate a more targeted and efficient rehabilitation strategy. The integration of the micro-service architecture and the API integration techniques ensures flexibility, scalability and continued progress of the overall rehabilitation process.

Drawings

FIG. 1 is a schematic workflow diagram of the present invention;

FIG. 2 is a S1 refinement flowchart of the present invention;

FIG. 3 is a S2 refinement flowchart of the present invention;

FIG. 4 is a S3 refinement flowchart of the present invention;

FIG. 5 is a S4 refinement flowchart of the present invention;

FIG. 6 is a S5 refinement flowchart of the present invention;

FIG. 7 is a S6 refinement flowchart of the present invention;

FIG. 8 is a S7 refinement flowchart of the present invention;

fig. 9 is a system flow diagram of the present invention.

Detailed Description

The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.

In the description of the present invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," and the like indicate orientations or positional relationships based on the orientation or positional relationships shown in the drawings, merely to facilitate describing the present invention and simplify the description, and do not indicate or imply that the devices or elements referred to must have a specific orientation, be configured and operated in a specific orientation, and therefore should not be construed as limiting the present invention. Furthermore, in the description of the present invention, the meaning of "a plurality" is two or more, unless explicitly defined otherwise.

Embodiment one: referring to fig. 1, the present invention provides a technical solution: the interactive rehabilitation training assisting method comprises the following steps of S1: based on the preliminary medical records and rehabilitation requirements of the patient, performing potential assessment by adopting a deep learning algorithm and time sequence analysis, and generating a rehabilitation potential analysis result;

s2: based on the rehabilitation potential analysis result, a rehabilitation training plan is designed by using the self-adaptive neural network and the fuzzy logic controller, and a preliminary rehabilitation training plan is generated;

s3: based on a preliminary rehabilitation training plan, a training adjustment algorithm based on reinforcement learning is utilized to adjust training difficulty and type according to real-time performance of a patient, and a dynamically adjusted rehabilitation training scheme is generated;

s4: based on a dynamically adjusted rehabilitation training scheme, performing multisensory stimulation integration by adopting a cross-modal fusion technology to generate a multisensory fusion training scheme;

s5: based on a multisensory fusion training scheme, performing data analysis by using a dynamic system theory and a nonlinear prediction technology to generate an optimized rehabilitation scheme;

s6: based on an optimized rehabilitation scheme, applying a micro-service architecture and API integration to perform technology integration, and generating a rehabilitation training architecture integrating a new technology;

s7: based on a rehabilitation training framework integrating new technology, the natural language processing technology is adopted to analyze patient feedback for continuous optimization, and a continuous optimization rehabilitation training framework is generated.

The rehabilitation potential analysis results comprise comprehensive assessment of the rehabilitation potential of the patient, limiting factor identification and preliminary rehabilitation advice, the preliminary rehabilitation training plan comprises customized training activities, scheduling and expected targets, the dynamically adjusted rehabilitation training scheme is specifically training difficulty and type based on real-time feedback and progress automatic adjustment of the patient, the multi-sense fusion training scheme is specifically fusion training activities of visual and auditory multi-sense elements, the optimized rehabilitation scheme is specifically targeted rehabilitation training strategies, and the continuously optimized rehabilitation training framework is specifically training content and method adjusted according to the feedback and progress of the patient.

The interactive rehabilitation training auxiliary method provides a personalized rehabilitation path for patients through the rehabilitation potential assessment by the deep learning algorithm and the time sequence analysis, can accurately assess the rehabilitation potential of the patients based on specific conditions of the patients, such as initial medical records and rehabilitation demands, and identifies limiting factors, so that the most suitable rehabilitation plan is customized for each patient, and the personalized method is beneficial to improving the rehabilitation efficiency and reducing unnecessary time waste.

The application of the self-adaptive neural network and the fuzzy logic controller further enhances the flexibility and accuracy of the rehabilitation training program, can design the training program which accords with the specific situation of the patient according to the rehabilitation potential analysis result, comprises the customized training activities, scheduling and expected targets, and is more suitable for the actual situation of the patient, thereby being beneficial to improving the rehabilitation effect.

The training adjustment algorithm based on reinforcement learning can dynamically adjust the training difficulty and type according to the real-time performance of the patient, the training program can respond to the progress and feedback of the patient in real time, the rehabilitation activity is ensured to be always carried out on the most suitable level, and the dynamic adjustment mechanism not only improves the flexibility of the rehabilitation training, but also can remarkably improve the participation degree and the power of the patient.

The application of the multi-sensory fusion training regimen provides a richer and interactive rehabilitation experience for the patient. The multi-sense elements such as vision and hearing are integrated into the training activities, so that the sense organs of a patient can be stimulated, the interestingness and effectiveness of rehabilitation training are enhanced, and the multi-sense interaction mode plays an important role in improving the participation degree and rehabilitation power of the patient.

The continuous optimization and individuation of the rehabilitation scheme are ensured by adopting the data analysis of the dynamic system theory and the nonlinear prediction technology, the training strategy can be finely adjusted according to the continuous feedback and the rehabilitation progress of the patient, and the rehabilitation activity is ensured to always keep the best effect.

The application of micro-service architecture and API integration, and the method for analyzing patient feedback by natural language processing technology provide strong technical support and continuous optimization possibility for the rehabilitation training framework. Not only improves the flexibility and expandability of the rehabilitation training method, but also ensures the continuous improvement of the experience of the patient, and ensures that the rehabilitation process is more efficient, personalized and user-friendly.

Referring to fig. 2, based on the preliminary medical records and rehabilitation requirements of the patient, the potential evaluation is performed by adopting a deep learning algorithm and time series analysis, and the specific steps of generating the rehabilitation potential analysis result are as follows: preparing data by adopting a data preprocessing technology based on the preliminary medical records of the patient, and generating a data preprocessing result;

s102: based on the data preprocessing result, identifying key rehabilitation indexes by adopting a feature extraction technology, and generating a feature extraction report;

In step S101, data is prepared based on the patient preliminary medical records by employing a data preprocessing technique, including normalization methods, missing value estimation, and outlier processing, to ensure accuracy and consistency of the data. The normalization method is used for normalizing data with different scales, the missing value estimation processes blank or missing information in the data set, and the outlier processing identifies and corrects irregular points in the data. Through the steps, a data preprocessing result is generated, and a solid foundation is laid for subsequent analysis.

In step S102, based on the data preprocessing result, a feature extraction technique is used to identify a key rehabilitation index. Feature extraction techniques, such as principal component analysis and linear discriminant analysis, are used to extract the most informative features from complex data sets, reflecting the patient's rehabilitation needs and potential, and generating feature extraction reports.

In step S103, based on the feature extraction report, the patient' S rehabilitation potential is analyzed using a deep learning algorithm, in particular a convolutional neural network. The deep learning algorithm can identify complex modes and relations in the data, generate a deep learning potential analysis result and provide strong support for accurate assessment of rehabilitation potential.

In step S104, based on the deep learning potential analysis result, the trend of the patient' S rehabilitation potential change is estimated in combination with a time series analysis method, such as an autoregressive model and a moving average model. The time series analysis is helpful for understanding the long-term trend and periodic change of the rehabilitation potential of the patient, and generating a rehabilitation potential analysis result.

Referring to fig. 3, based on the rehabilitation potential analysis result, a rehabilitation training program is designed by using an adaptive neural network and a fuzzy logic controller, and the specific steps of generating a preliminary rehabilitation training program are as follows: based on the rehabilitation potential analysis result, classifying the rehabilitation types of the patient by using a data clustering algorithm to generate a patient rehabilitation type classification;

S202: based on the classification of the rehabilitation types of the patients, a rule engine is adopted to make a preliminary training strategy according to the rehabilitation types, and a preliminary training strategy draft is generated;

s203: based on the preliminary training strategy draft, generating a training plan optimized by the self-adaptive neural network by utilizing the self-adaptive neural network;

s204: based on the training program optimized by the self-adaptive neural network, performing final fine adjustment and optimization on the training program by using the fuzzy logic controller to generate a preliminary rehabilitation training program;

the data clustering algorithm is specifically a K-means clustering method, the rule engine is specifically a decision tree algorithm, the self-adaptive neural network is specifically a multi-layer perceptron and a back propagation algorithm, and the fuzzy logic controller comprises a fuzzy rule set and an reasoning mechanism.

In step S201, based on the rehabilitation potential analysis result, the patients are classified according to the rehabilitation types by using a data clustering algorithm, such as K-means clustering method, and they are divided into different groups according to the rehabilitation potential and the requirement of the patients, so as to ensure that each patient can receive the rehabilitation training most suitable for the situation of the patient, thereby generating the patient rehabilitation type classification.

Step S202 is entered, based on the classification of the rehabilitation types of the patients, a rule engine, in particular a decision tree algorithm is adopted, a preliminary training strategy is formulated according to the rehabilitation types, the process involves analyzing specific requirements of different rehabilitation types, and a corresponding training plan is formulated, so that a preliminary training strategy draft is generated.

In step S203, the training program is optimized using an adaptive neural network, including multi-layer perceptron and back propagation algorithm, based on the preliminary training strategy draft. The adaptive neural network can adjust the training program according to the specific feedback and progress of the patient, and ensure the effectiveness and individuality of the training program, so that the training program optimized by the adaptive neural network is generated.

In step S204, a fuzzy logic controller is applied to the training program based on the adaptive neural network optimization, including fuzzy rule sets and inference mechanisms, to perform final fine tuning and optimization on the training program. The fuzzy logic controller can process uncertainty and ambiguity, so that the training program is more flexible and has strong adaptability, and a preliminary rehabilitation training program is generated.

Referring to fig. 4, based on the preliminary rehabilitation training plan, the training difficulty and type are adjusted according to the real-time performance of the patient by using the training adjustment algorithm based on reinforcement learning, and the specific steps of generating the dynamically adjusted rehabilitation training scheme are as follows: simulating a patient rehabilitation training scene by adopting an environment modeling technology based on the preliminary rehabilitation training plan to generate a training scene simulation;

s302: based on training scene simulation, applying a reinforcement learning algorithm to learn a patient behavior mode and generating a proxy learning model;

s304: based on the optimized training adjustment strategy, the training difficulty and type are adjusted by applying a dynamic programming algorithm, the real-time performance of a patient is adapted, and a dynamically adjusted rehabilitation training scheme is generated;

the environment modeling technology comprises a state space model and a dynamic system simulation, the reinforcement learning algorithm is specifically a Q learning algorithm, the rewarding mechanism comprises accumulated return optimization and behavior evaluation, and the dynamic planning algorithm is specifically a Belman equation.

In step S301, based on the preliminary rehabilitation training plan, the patient rehabilitation training scene is simulated by adopting the environment modeling technology, including the state space model and the dynamic system simulation, so as to create a virtual environment, simulate various rehabilitation training scenes possibly encountered by the patient, help better understand and predict the performance of the patient under different training conditions, provide a basis for subsequent training adjustment, and generate a training scene simulation.

In step S302, a reinforcement learning algorithm, in particular a Q learning algorithm, is applied to learn the behavior pattern of the patient based on the training scenario simulation. The reinforcement learning algorithm can be continuously learned and adjusted according to the reaction and the behaviors of the patient through the agent learning model, so that a more personalized and effective training strategy is generated.

In step S303, a reward mechanism is implemented to optimize a training program adjustment strategy based on the agent learning model. A rewarding mechanism including cumulative rewards optimization and behavioral assessment for encouraging the patient to conduct rehabilitation-beneficial behaviors. Through the mechanism, the training plan can be dynamically adjusted, so that the training plan is more in line with the current rehabilitation state and the current requirements of the patient, and an optimized training adjustment strategy is generated.

In step S304, based on the optimized training adjustment strategy, a dynamic programming algorithm, in particular, a bellman equation is applied to adjust the training difficulty and type to adapt to the real-time performance of the patient, allowing the training program to flexibly adjust according to the instant feedback and progress of the patient, ensuring that the training is always performed at an optimum level, thereby generating a dynamically adjusted rehabilitation training scheme.

Referring to fig. 5, a dynamic adjustment-based rehabilitation training scheme is adopted to integrate multisensory stimulation by using a cross-modal fusion technology, and the specific steps of generating the multisensory fusion training scheme are as follows: based on a dynamically adjusted rehabilitation training scheme, analyzing the interaction effect of the multi-sensory stimulation by adopting a data analysis technology, and generating a sensory stimulation interaction analysis result;

s402: based on the sensory stimulation interaction analysis result, a multi-sensory fusion technology is applied to integrate visual and auditory stimuli, and a multi-sensory synchronous scheme is generated;

S403: based on a multisensory synchronization scheme, designing a multi-mode rehabilitation training activity, and generating a multi-mode training activity design;

s404: based on the multi-modal training activity design, integrating multi-sense stimulation content by adopting a sense weighted fusion technology to generate a multi-sense fusion training scheme;

the data analysis technology is specifically correlation analysis and interactive mode recognition, the multi-sense fusion technology is specifically a sense synchronization algorithm and a mode fusion method, the multi-mode rehabilitation training activities comprise audiovisual coordination training and a sense integration strategy, and the sense weighting fusion technology is specifically a weighted average method and a sense stimulation optimization fusion algorithm.

In step S401, sensory stimulation interaction analysis is performed by using a data analysis technology, and correlation analysis and interaction pattern recognition are performed;

example code:

python；

Copy code；

import pandas as pd；

from scipy.stats import pearsonr；

let # assume df is the DataFrame containing multisensory stimulation data;

for example, # df=pd.dataframe ({ 'visual data': [..], 'auditory data': [.], and 'tactile data': [.]);

# calculate the correlation between visual and auditory data;

corralation, _pearsonr (df [ 'visual data' ], df [ 'auditory data' ];

print (f 'visual and auditory data correlation { correlation }');

In step S402, multisensory synchronization is performed using multisensory fusion techniques;

example code:

python；

Copy code；

def synchronize_sensory_data(visual_data, auditory_data):

# the code example here is simplified, and the actual implementation may require complex time synchronization logic;

# assume that visual_data and audio_data are the same in length;

synchronized_data = (visual_data + auditory_data) / 2；

return synchronized_data；

# example data;

visual_data=df [ 'visual data' ];

audioy_data=df [ 'auditory data'

# obtaining synchronized sensory data;

multisensory_data = synchronize_sensory_data(visual_data, auditory_data)；

in step S403, a multi-modal rehabilitation training activity design is performed;

example code:

python；

Copy code；

def design_training_activities(multisensory_data):

design training activities based on multisensory data;

the example herein is conceptual;

activities = []；

for data in multisensory_data:

activity=f "audiovisual coordination task { data }";

activities.append(activity)；

return activities；

training_activities = design_training_activities(multisensory_data)；

in step S404, a multi-sense fusion training scheme is generated using a perceptual weighted fusion technique;

example code:

python；

Copy code；

def weighted_sensory_integration(activities, visual_weight, auditory_weight):

carrying out sensory data fusion by using a weighted average method, # and (4);

integrated_activities = []；

for activity in activities:

integrated_activity = activity * visual_weight + activity * auditory_weight；

integrated_activities.append(integrated_activity)；

return integrated_activities；

example weights #;

visual_weight = 0.6；

auditory_weight = 0.4；

integrated_training_plan = weighted_sensory_integration(training_activities, visual_weight, auditory_weight)。

referring to fig. 6, based on the multi-sensory fusion training scheme, the data analysis is performed by using the dynamic system theory and the nonlinear prediction technology, and the specific steps of generating the optimized rehabilitation scheme are as follows: based on a multisensory fusion training scheme, adopting dynamic system modeling to analyze the time sequence characteristics of training data, and generating a dynamic system training data model;

S502: based on a dynamic system training data model, a nonlinear time sequence analysis technology is applied to reveal a complex mode in training data, and a nonlinear time sequence analysis result is generated;

s503: based on nonlinear time sequence analysis results, predicting rehabilitation progress by using a machine learning prediction model, and generating a rehabilitation progress prediction model;

In step S501, based on the multisensory fusion training scheme, the application of the state space model is related to by adopting dynamic system modeling to analyze the time sequence characteristics of the training data, so as to construct a dynamic system representation of the training data, which is helpful for capturing the time dynamic change in the training process, and generating a dynamic system training data model, thereby providing a foundation for deep time sequence analysis.

In step S502, based on the dynamic system training data model, a nonlinear time sequence analysis technology, in particular, a chaos theory is applied to reveal a complex mode in training data, which is helpful for understanding nonlinear characteristics and internal rules in the rehabilitation training process, generating a nonlinear time sequence analysis result, and revealing a deep structure and dynamics of the training data.

In step S503, based on the nonlinear time series analysis result, a machine learning prediction model, in particular, a support vector machine is used to predict rehabilitation progress, the analyzed time series data is used to train the prediction model, and the future rehabilitation trend and potential result of the patient are estimated, so as to generate a rehabilitation progress prediction model.

In step S504, based on the rehabilitation progress prediction model, a decision tree algorithm, particularly a random forest, is adopted to comprehensively analyze data and optimize the training plan, and by comprehensively considering the data and the prediction result in multiple aspects, a more accurate adjustment suggestion is provided for the rehabilitation training plan to generate an optimized rehabilitation scheme.

Referring to fig. 7, based on the optimized rehabilitation scheme, the technology integration is performed by applying the micro-service architecture and the API integration, and the specific steps of generating the rehabilitation training architecture integrating the new technology are as follows: based on an optimized rehabilitation scheme, planning a service module of a rehabilitation training system by adopting a micro-service architecture design principle, and generating a micro-service architecture design;

s603: based on an API integration scheme, combining micro-service interface development to generate a micro-service interface development result;

S604: based on the micro-service interface development result, applying a containerization technology, implementing system integration, and generating a rehabilitation training framework integrating a new technology;

the design principle of the micro-service architecture is specifically a field driving design, the API management tool is specifically an API gateway, the development of the micro-service interface is specifically a RESTful API design, and the containerization technology is specifically a Docker.

In step S601, a service module of the rehabilitation training system is planned by adopting a micro-service architecture design principle based on an optimized rehabilitation scheme. The design principle, especially the field driving design, is beneficial to identifying and defining key services and functional modules in the rehabilitation training system, ensures the flexibility and the expandability of the system structure, generates a micro-service architecture design and lays a foundation for subsequent service development and integration.

In step S602, based on the micro-service architecture design, the service modules are integrated using an API management tool, in particular an API gateway. The API management tool is helpful for uniformly managing and scheduling interaction among different micro services, and ensures efficient communication and data exchange among services.

In step S603, based on the API integration scheme, a micro-service interface is developed, in particular a RESTful API design. The RESTful API design provides a compact and efficient way for interactions between services, helping to achieve seamless integration between parts of the system.

In step S604, based on the micro-service interface development results, a containerization technique, such as Docker, is applied to implement system integration. The containerization technology is beneficial to uniformly deploying and running the micro-services under different environments, and ensures the reliability and stability of the system. By the technology, a rehabilitation training framework integrating a new technology is generated, and the overall coordination and high efficiency of the system are ensured.

Referring to fig. 8, based on the rehabilitation training architecture of the integrated new technology, the specific steps of generating the rehabilitation training framework of continuous optimization by analyzing patient feedback through natural language processing technology are as follows: based on a rehabilitation training architecture integrating a new technology, analyzing the written feedback of a patient by adopting a text mining technology, and generating a text analysis result of the feedback of the patient;

s702: based on the feedback text analysis result of the patient, the emotional response of the patient is estimated by applying an emotional analysis technology, and a patient emotional analysis report is generated;

s703: based on the patient emotion analysis report, carrying out deep analysis by using a natural language processing algorithm, extracting deep patient demands and suggestions, and generating a deep patient demand analysis result;

s704: based on the deep patient demand analysis result, integrating the analysis result by adopting a decision support algorithm, continuously optimizing the rehabilitation training scheme, and generating a continuously optimized rehabilitation training frame;

The text mining technology is particularly word frequency-inverse document frequency analysis, the emotion analysis technology is particularly emotion polarity analysis, the natural language processing algorithm is particularly a long-time and short-time memory network, and the decision support algorithm is particularly a fuzzy logic and data-driven decision tree.

In step S701, based on the rehabilitation training architecture integrated with the new technology, the text mining technology is adopted to analyze the written feedback of the patient, and relates to word frequency-inverse document frequency analysis, so as to extract key information and trend from the written feedback of the patient, thereby being helpful for understanding the general viewpoint and focus of the patient, generating the text analysis result of the feedback of the patient, and providing a foundation for deep analysis.

In step S702, based on the results of the patient feedback text analysis, emotional response of the patient is evaluated by applying emotion analysis techniques, in particular emotion polarity analysis. The emotion analysis technology can identify and evaluate emotion tendencies in feedback, such as positive, negative or neutral, so that a patient emotion analysis report is generated, the emotion feedback of a patient on a rehabilitation training scheme is facilitated to be understood, and an important index is provided for optimization of the training scheme.

In step S703, based on the patient emotion analysis report, a natural language processing algorithm, especially a long-short time memory network, is used to perform deep analysis, so as to extract deep patient needs and suggestions, and understand the concrete comments and needs of the patient on the rehabilitation training scheme. Through the deep analysis, a deep patient demand analysis result is generated, and a basis is provided for more accurate adjustment of a training scheme.

In step S704, based on the deep patient demand analysis result, a decision support algorithm, in particular a fuzzy logic and data driven decision tree, is adopted to integrate the analysis result, and the rehabilitation training scheme is continuously optimized by combining the patient demand, feedback and emotion analysis result so as to meet the actual demand of the patient, so that a continuously optimized rehabilitation training frame is generated.

Referring to fig. 9, the interactive rehabilitation training assistance system comprises a rehabilitation potential analysis module, a strategy formulation optimization module, a scene simulation learning module, a dynamic training plan adjustment module, a multi-sense fusion design module, a system modeling progress prediction module, a micro-service architecture integration module and a feedback analysis decision module;

the strategy making and optimizing module makes a preliminary training strategy by utilizing a rule engine based on the rehabilitation potential analysis result, optimizes and fine-tunes a training plan by utilizing the self-adaptive neural network and the fuzzy logic controller, and generates a preliminary rehabilitation training plan;

the dynamic training plan adjusting module is used for implementing a rewarding mechanism based on the agent learning model, adjusting the training difficulty and type through a dynamic planning algorithm, adapting to real-time performance of a patient and generating a dynamically adjusted rehabilitation training scheme;

The feedback analysis decision module analyzes the written feedback of the patient by adopting a text mining technology based on a rehabilitation training framework integrating a new technology, evaluates the emotional response of the patient by adopting an emotion analysis technology, analyzes the integration result by utilizing a natural language processing technology and a decision support algorithm, continuously optimizes the rehabilitation training scheme and generates a continuously optimized rehabilitation training framework.

The system ensures high individuality of the rehabilitation scheme through application of the rehabilitation potential analysis module and the strategy formulation optimization module. By means of deep learning and time sequence analysis, rehabilitation potential of each patient can be accurately estimated, and a rehabilitation plan meeting specific requirements of each patient is formulated by combining a rule engine and self-adaptive network optimization, so that rehabilitation efficiency is improved, and rehabilitation experience and satisfaction of the patient can be improved.

The scene simulation learning module and the dynamic training plan adjusting module further enhance the real-time responsiveness and adaptability of rehabilitation training. Through simulation training scene and reinforcement study, can be according to patient's real-time performance adjustment training degree of difficulty and type, ensure that training is gone on in optimum level all the time to improve rehabilitation training's effect.

The multi-sense organ fusion design module enhances interactivity and interestingness of rehabilitation training by integrating multi-mode stimulation, such as vision and hearing, and the multi-sense organ fusion training can stimulate multiple senses of a patient, so that participation of the patient is improved, and the rehabilitation process is more vivid and interesting.

The system modeling progress prediction module and the feedback analysis decision module provide powerful tools for continuous optimization and decision support of the system. By analyzing and predicting rehabilitation progress, the system is able to continuously optimize rehabilitation regimens, with feedback analysis ensuring that the actual experience and needs of the patient are fully considered and taken into account.

The application of the micro-service architecture integrated module ensures the flexibility, expandability and maintainability of the system. The micro-service architecture enables the system to flexibly integrate various services and technologies, adapt to rapidly changing technical environments and patient requirements, and improve the overall efficiency and reliability of the system.

In summary, the interactive rehabilitation training auxiliary system not only improves the efficiency and effect of rehabilitation training, but also provides more personalized and interactive rehabilitation experience for patients through advanced technical integration and modularized design, and simultaneously ensures the flexibility and sustainability of the system.

The present invention is not limited to the above embodiments, and any equivalent embodiments which can be changed or modified by the technical disclosure described above can be applied to other fields, but any simple modification, equivalent changes and modification made to the above embodiments according to the technical matter of the present invention will still fall within the scope of the technical disclosure.

Claims

1. The interactive rehabilitation training auxiliary method is characterized by comprising the following steps of carrying out potential assessment by adopting a deep learning algorithm and time sequence analysis based on preliminary medical records and rehabilitation demands of patients to generate a rehabilitation potential analysis result;

based on the rehabilitation potential analysis result, a rehabilitation training plan is designed by using a self-adaptive neural network and a fuzzy logic controller, and a preliminary rehabilitation training plan is generated;

based on the preliminary rehabilitation training plan, a training adjustment algorithm based on reinforcement learning is utilized to adjust training difficulty and type according to real-time performance of a patient, and a dynamically adjusted rehabilitation training scheme is generated;

based on the dynamically adjusted rehabilitation training scheme, performing multi-sense stimulation integration by adopting a cross-mode fusion technology to generate a multi-sense fusion training scheme;

Based on the multi-sense fusion training scheme, performing data analysis by using a dynamic system theory and a nonlinear prediction technology to generate an optimized rehabilitation scheme;

based on the optimized rehabilitation scheme, applying a micro-service architecture and API integration to perform technology integration, and generating a rehabilitation training architecture integrating a new technology;

based on the rehabilitation training framework integrating the new technology, the natural language processing technology is adopted to analyze the feedback of the patient to perform continuous optimization, and a continuous optimization rehabilitation training framework is generated.

2. The interactive rehabilitation training assistance method according to claim 1, characterized in that the rehabilitation potential analysis results comprise comprehensive assessment of patient rehabilitation potential, limiting factor identification and preliminary rehabilitation advice, the preliminary rehabilitation training plan comprises customized training activities, scheduling and expected targets, the dynamically adjusted rehabilitation training scheme is specifically training difficulty and type automatically adjusted based on real-time feedback and scheduling of patients, the multi-sense fusion training scheme is specifically fusion training activities of visual and auditory multi-sense elements, the optimized rehabilitation scheme is specifically targeted rehabilitation training strategy, and the continuously optimized rehabilitation training framework is specifically training content and method adjusted according to patient feedback and progress.

3. The interactive rehabilitation training assistance method according to claim 1, wherein the specific steps of performing potential assessment by using a deep learning algorithm and time series analysis based on the preliminary medical records of the patient and the rehabilitation requirements, and generating a rehabilitation potential analysis result are that preparing data by using a data preprocessing technology based on the preliminary medical records of the patient, and generating a data preprocessing result;

based on the data preprocessing result, identifying key rehabilitation indexes by adopting a characteristic extraction technology, and generating a characteristic extraction report;

based on the feature extraction report, analyzing the rehabilitation potential of the patient by using a deep learning algorithm, and generating a deep learning potential analysis result;

based on the deep learning potential analysis result, evaluating the change trend of the rehabilitation potential of the patient by combining a time sequence analysis method, and generating a rehabilitation potential analysis result;

4. The interactive rehabilitation training assistance method according to claim 1, wherein the specific step of designing a rehabilitation training plan by using an adaptive neural network and a fuzzy logic controller based on the rehabilitation potential analysis result, and generating a preliminary rehabilitation training plan is to classify patient rehabilitation types by using a data clustering algorithm based on the rehabilitation potential analysis result, and generate patient rehabilitation type classification;

based on the patient rehabilitation type classification, a rule engine is adopted to make a preliminary training strategy according to the rehabilitation type, and a preliminary training strategy draft is generated;

based on the preliminary training strategy draft, generating a training plan optimized by the self-adaptive neural network by using the self-adaptive neural network;

based on the training program optimized by the self-adaptive neural network, performing final fine adjustment and optimization on the training program by using a fuzzy logic controller to generate a preliminary rehabilitation training program;

5. The interactive rehabilitation training assistance method according to claim 1, wherein based on the preliminary rehabilitation training plan, a training adjustment algorithm based on reinforcement learning is utilized to adjust training difficulty and type according to real-time performance of a patient, and the specific step of generating a dynamically adjusted rehabilitation training scheme is that based on the preliminary rehabilitation training plan, an environment modeling technology is adopted to simulate a patient rehabilitation training scene, and training scene simulation is generated;

based on the training scene simulation, applying a reinforcement learning algorithm to learn a patient behavior mode and generating a proxy learning model;

based on the agent learning model, implementing a reward mechanism, and generating an optimization training adjustment strategy;

based on the optimized training adjustment strategy, a dynamic programming algorithm is applied to adjust the training difficulty and type, and the method is suitable for real-time performance of patients and generates a dynamically adjusted rehabilitation training scheme;

6. The interactive rehabilitation training assistance method according to claim 1, wherein based on the dynamically adjusted rehabilitation training scheme, a cross-modal fusion technology is adopted to integrate multiple sensory stimuli, and the specific step of generating the multiple sensory fusion training scheme is that based on the dynamically adjusted rehabilitation training scheme, a data analysis technology is adopted to analyze the interaction effect of the multiple sensory stimuli, so as to generate a sensory stimulus interaction analysis result;

Based on the sensory stimulus interaction analysis result, a multi-sensory fusion technology is applied to integrate visual and auditory stimuli, and a multi-sensory synchronous scheme is generated;

designing a multi-mode rehabilitation training activity based on the multi-sense synchronous scheme, and generating a multi-mode training activity design;

based on the multi-mode training activity design, integrating multi-sense stimulation content by adopting a sensing weighted fusion technology to generate a multi-sense fusion training scheme;

7. The interactive rehabilitation training assistance method according to claim 1, wherein based on the multi-sense fusion training scheme, data analysis is performed by using a dynamic system theory and a nonlinear prediction technology, and the specific step of generating an optimized rehabilitation scheme is that based on the multi-sense fusion training scheme, a dynamic system modeling analysis is adopted to analyze time series characteristics of training data, so as to generate a dynamic system training data model;

Based on the dynamic system training data model, a nonlinear time sequence analysis technology is applied to reveal a complex mode in training data, and a nonlinear time sequence analysis result is generated;

based on the nonlinear time sequence analysis result, predicting rehabilitation progress by using a machine learning prediction model, and generating a rehabilitation progress prediction model;

based on the rehabilitation progress prediction model, comprehensively analyzing data by adopting a decision tree algorithm, optimizing a training plan, and generating an optimized rehabilitation scheme;

8. The interactive rehabilitation training assistance method according to claim 1, wherein based on the optimized rehabilitation scheme, the technical integration is performed by applying a micro-service architecture and API integration, and the specific step of generating a rehabilitation training architecture integrating a new technology is to plan a service module of a rehabilitation training system by adopting a micro-service architecture design principle based on the optimized rehabilitation scheme, so as to generate a micro-service architecture design;

Based on the micro-service architecture design, integrating each service module by using an API management tool to generate an API integration scheme;

based on the API integration scheme, combining micro-service interface development to generate a micro-service interface development result;

based on the micro-service interface development result, applying a containerization technology to implement system integration, and generating a rehabilitation training framework integrating a new technology;

9. The interactive rehabilitation training assistance method according to claim 1, wherein based on the rehabilitation training architecture of the integrated new technology, the patient feedback is analyzed by adopting a natural language processing technology for continuous optimization, and the specific step of generating a continuous optimized rehabilitation training frame is that based on the rehabilitation training architecture of the integrated new technology, the patient written feedback is analyzed by adopting a text mining technology, and a patient feedback text analysis result is generated;

based on the feedback text analysis result of the patient, applying emotion analysis technology to evaluate the emotional response of the patient and generating a patient emotion analysis report;

Based on the patient emotion analysis report, performing deep analysis by using a natural language processing algorithm, extracting deep patient requirements and suggestions, and generating a deep patient requirement analysis result;

based on the deep patient demand analysis result, integrating the analysis result by adopting a decision support algorithm, continuously optimizing a rehabilitation training scheme, and generating a continuously optimized rehabilitation training frame;

10. An interactive rehabilitation training assistance system, characterized in that the interactive rehabilitation training assistance method according to any one of claims 1-9 comprises a rehabilitation potential analysis module, a strategy formulation optimization module, a scene simulation learning module, a dynamic training plan adjustment module, a multi-sense fusion design module, a system modeling progress prediction module, a micro-service architecture integration module, and a feedback analysis decision module;