An ai-driven traditional chinese and western medicine integrated health assessment scheme and data adaptation method

Abstract

This invention discloses an AI-driven integrated traditional Chinese and Western medicine health assessment scheme and data adaptation method, involving AI large-scale model technology, health sensing technology, multi-source heterogeneous data fusion technology, and the field of integrated traditional Chinese and Western medicine technology. Based on the TCM concepts of "prevention of disease" and "syndrome differentiation and treatment," this invention constructs a flexible deployment architecture for AI large-scale models and a collaborative system of edge-cloud intelligent agents. Through acquiring multi-dimensional physiological data, sample-driven machine learning, autonomous learning of TCM and Western medicine principles, adaptive optimization of assessment weights, personalized interaction at the edge, and intelligent early warning reminders, it achieves intelligent health status assessment and personalized conditioning output closed-loop management. This invention abandons manual mapping and reliance on knowledge graphs, using data to drive model autonomous learning, supplemented by TCM expert rules as boundary constraints and special scenario interventions. It solves problems such as data fragmentation, one-sided assessment, insufficient integration of TCM and Western medicine, superficial AI application, and weak edge response capabilities in existing technologies, improving the accuracy, timeliness, and applicability of health management, and possesses significant innovation and industrial value.

Description

Technical Field

[0001] This invention relates to the fields of AI large model technology, health sensing technology, multi-source heterogeneous data fusion technology, and integrated traditional Chinese and Western medicine technology, specifically to an AI-driven, edge-cloud collaborative, data-standardized, and deeply integrated traditional Chinese and Western medicine health assessment and data adaptation method. Background Technology

[0002] The current demand for health management has shifted from disease response to prevention, maintenance, and proactive risk control. Sub-health conditions, sleep disorders, metabolic fluctuations, cognitive changes, emotional fluctuations, and potential health risks are characterized by their insidious, gradual, and high incidence, placing high demands on continuous, multi-dimensional, standardized health data and systematic intelligent analysis.

[0003] The existing technology has five major flaws: The data acquisition dimensions are too limited and the standardization is insufficient, making it difficult to reflect the overall health status; Data is disconnected from in-depth analysis, lacking the ability to uncover and predict trends; The assessment logic is rigid, overly reliant on Western medical indicators, and disconnected from the health concepts of traditional Chinese medicine. There is insufficient integration of traditional Chinese and Western medicine, and a lack of a data-driven, systematic integration mechanism. AI technology is applied superficially, with weak edge response capabilities, rigid model deployment, weak self-learning ability, poor data adaptability, and no closed-loop iteration.

[0004] Health management needs to be based on traditional Chinese medicine principles, grounded in data, and driven by AI to achieve seamless integration of data acquisition, adaptation, learning, assessment, conditioning, and feedback throughout the entire process. Therefore, developing AI-driven, integrated traditional Chinese and Western medicine approaches, and edge-cloud collaborative health assessment and data adaptation methods is of great significance. Summary of the Invention

[0005] I. Purpose of the Invention This invention overcomes the shortcomings of existing technologies and provides an AI-driven method for integrated traditional Chinese and Western medicine health assessment and data adaptation. It achieves multi-source data standardization, AI model autonomous learning, deep integration of traditional Chinese and Western medicine, adaptive assessment weights, personalized interaction, intelligent early warning reminders, personalized suggestion output, and closed-loop iteration throughout the entire process, thereby improving the accuracy, scientificity, and practicality of health management. II. Technical Solution

[0006] The core technical logic of this invention is based on the traditional Chinese medicine concepts of "prevention of disease" and "syndrome differentiation and treatment," relying on leading AI technology to achieve a closed-loop process of modern data acquisition, deep integration of traditional Chinese and Western medicine, and personalized health management. The overall technical solution is designed around a progressive logic of "architecture deployment → data acquisition and adaptation → model calculation and alignment with traditional Chinese and Western medicine → health assessment and weight adjustment → personalized conditioning and closed-loop optimization → data security assurance." It leverages an edge-cloud intelligent agent collaborative architecture and the flexible deployment capabilities of large AI models to overcome the shortcomings of existing technologies. The specific core technical solutions are as follows: The core technical solution adopted in this invention is based on a flexible deployment architecture for large AI models and a collaborative architecture between edge and cloud intelligent agents, forming a layered, collaborative, and closed-loop technical system, specifically implemented as follows: Step 1: Architecture Deployment A flexible deployment architecture for the AI ​​large-scale model and a collaborative architecture between edge and cloud intelligent agents are constructed, clearly defining the layered collaborative logic and functional division of labor. The AI ​​large-scale model supports two independently selectable, non-binding deployment modes: First, an open-source large-scale model is trained and optimized, then deployed locally in different versions to form a proprietary model adapted to the health management scenario of this invention; second, a closed-source large-scale model is deployed with parameter adaptation, requiring only parameter debugging and adaptation without model training. Corresponding to these two deployment modes, the edge intelligent agent deploys a lightweight version of the AI ​​large-scale model, serving as a high-feedback, low-global-judgment terminal execution unit responsible for local data acquisition, preprocessing, personalized interaction, and real-time early warning and reminder services. The cloud intelligent agent deploys the full version of the AI ​​large-scale model, responsible for core computation, model iteration and optimization, intelligent benchmarking of traditional Chinese and Western medicine, and comprehensive health assessment. The two intelligent agents complete collaborative debugging to ensure smooth data transmission and model linkage, simultaneously adapting to both large-scale model deployment modes and accommodating different technology implementation needs.

[0007] The second step is data acquisition and local adaptation. Led by an edge agent, the system acquires, preprocesses, and standardizes multi-source data, providing high-quality data support for subsequent model computation and health assessment. Specifically, the edge agent deploys a lightweight interactive interface, allowing users to input simplified personalized health information such as age, gender, physical condition, and past health history. After information parsing and format standardization, the information is synchronized to the cloud agent. Simultaneously, it uniformly connects to various sensing devices centered around wearable smart devices, supporting multi-standard compatibility and dynamically scheduling the data collection frequency based on the user's health status, achieving low-power, high-precision data acquisition. The acquired multi-dimensional physiological characteristic data is clearly divided into basic and advanced indicators. Basic indicators include heart rate, blood oxygen, blood pressure, blood glucose, uric acid, menstrual cycle, sleep stage, sleep heart rate, sleep blood oxygen, sleep HRV, sleep respiratory rate, daily steps, steps taken, distance traveled, exercise time, and visual perception (graphics / texture / motion / ). (Stereoscopic vision), high-level indicators include ECG, EEG, and ductal electrodermatology; the edge agent performs local noise reduction, outlier filtering, and time-series verification on the acquired data, eliminating invalid data. Offline caching ensures data integrity and time-series consistency during network interruptions, synchronizing only standardized, qualified data to the cloud agent. A data feedback interface is also reserved to receive assessment and adjustment results from the cloud and present them to the user. Based on a lightweight AI model, the edge agent simultaneously enables personalized user interaction, intelligent early warnings, and status alerts, achieving a high-feedback front-end response.

[0008] The third step is to benchmark the model computation against the intelligent systems of traditional Chinese and Western medicine. Led by a cloud-based intelligent agent, this invention leverages a large AI model to perform machine learning, feature fusion, and autonomous association learning on standardized data, achieving a deep integration of modern physiological data and the theoretical system of Traditional Chinese Medicine (TCM). Specifically, the cloud-based intelligent agent integrates the computation and iteration modules of the large AI model. Depending on the deployment mode of the large model, it performs targeted model optimization: When deploying with an open-source large model, it continuously optimizes model training and iterates parameters based on TCM and Western medicine clinical data, user historical health data, and standardized samples synchronized by the edge intelligent agent, improving the accuracy of model analysis and inference; when deploying with a closed-source large model, only parameter adaptation and feedback optimization are performed, without model training, and the standardized data synchronized by the edge intelligent agent is used as calibration data for parameter debugging. This invention does not rely on manually defined mapping relationships or knowledge graphs. Through autonomous learning from massive samples using an AI large model, it constructs a data-driven association between physiological data and the TCM Zang-Xiang system. Simultaneously, it introduces TCM expert rules as boundary constraints and auxiliary interventions for special scenarios, ensuring rigorous and standardized diagnostic logic and providing scientific, stable, and iterative data support for TCM health preservation and diagnostic assessment.

[0009] Step 4: Health assessment and adaptive weight adjustment Based on the core computing power and adaptive evaluation mechanism of cloud-based intelligent agents, multi-dimensional accurate health assessments are achieved, solving the problem of insufficient adaptability of traditional models. Specifically, the adaptive model is deployed in a distributed manner across the edge-cloud intelligent agent. Lightweight model units on the edge side are responsible for preliminary data processing, anomaly identification, and real-time early warning. The core model units on the cloud side include weight adaptive optimization, cloud-based deep computing, and security collaboration modules. The weight adaptive optimization mechanism is deeply coupled with the large AI model, dynamically adjusting the weight ratio of each assessment indicator based on user physiological data, physical characteristics, and model learning samples to achieve personalized assessments for different health states. The cloud-based deep computing module, combined with the AI ​​model's autonomous learning of the correlation between traditional Chinese and Western medicine, completes accurate multi-dimensional health status assessments for users, outputting diagnostic results, health risk classifications, and early warning signals, accurately identifying various health risks and abnormal states.

[0010] Step 5: Personalized conditioning and closed-loop optimization Personalized health management plans are generated based on health assessment results, and a closed-loop iteration process is achieved through data feedback to improve the sustainability and effectiveness of health management. Specifically, the cloud-based intelligent agent outputs personalized health management plans (within the scope of health management, excluding disease diagnosis and prescription) based on the TCM (Traditional Chinese Medicine) syndrome differentiation assessment results and incorporating TCM and Western medicine theories. These plans are tailored to different syndrome types and health states, covering dietary adjustments, lifestyle guidance, exercise suggestions, and TCM health care. The edge intelligent agent presents the assessment results and health management plans to the user, while continuously acquiring post-intervention feedback data and physiological representation data. After standardization, this data is synchronized to the cloud-based intelligent agent as incremental learning samples for the AI ​​large-scale model. This data is used for model iteration optimization and dynamic adjustment of the assessment logic and health management plans, forming a complete collaborative closed loop of "acquisition-adaptation-assessment-health management-feedback-optimization".

[0011] Step 6: Data Security Assurance A data security mechanism is implemented throughout the entire technical process to ensure user privacy and data security, supporting the large-scale application of the technology. Specifically, the edge-cloud intelligent agent employs an encrypted communication mechanism, and the cloud intelligent agent integrates a data security encryption module to encrypt user-entered information, physiological data, model data, and evaluation results throughout the entire process, strictly complying with relevant laws and regulations. Simultaneously, it supports multi-regional data interconnection and interoperability, enabling cross-regional collaborative iteration of large AI models, providing security guarantees for the technology's adaptation to multiple scenarios and large-scale deployment.

[0012] Furthermore, the data adaptation method and personalized weight adaptive algorithm of this invention serve as core supporting technologies throughout the entire process: the data adaptation method is achieved through collaborative edge-cloud intelligent agents, encompassing user information adaptation, multi-dimensional data acquisition, preprocessing and screening, traditional Chinese and Western medicine benchmarking adaptation, and result output adaptation. It can simultaneously match two AI large-scale model deployment modes, providing high-precision data support for model iteration and health assessment. The personalized weight adaptive algorithm works deeply with the AI ​​large-scale model, relying on model iteration to continuously improve assessment accuracy, overcoming the technical bottleneck of insufficient adaptability of traditional fixed models, and ensuring the personalization and accuracy of assessment results. Simultaneously, physiological characterization data and user-entered information are used collaboratively as the basis for model learning and health assessment, but not as the sole standard for health assessment, ensuring the scientific rigor and comprehensiveness of the assessment.

[0013] In summary, the technical solution of this invention is logically coherent and progressively advanced, forming a complete technical chain from architecture deployment to closed-loop optimization. It not only highlights the core position of traditional Chinese medicine concepts but also deeply integrates leading AI technologies, realizing full-process intelligentization of data acquisition, model calculation, integration of traditional Chinese and Western medicine, health assessment, personalized interaction, intelligent early warning, and conditioning services. It effectively addresses the core shortcomings of existing technologies and possesses clear technical innovation and practicality. III. Beneficial Effects

[0014] To achieve continuous monitoring, accurate assessment, early warning, and rapid response of health status across all scenarios; Build a multi-dimensional, standardized, and continuous health data system to solve the problem of data fragmentation; AI-powered autonomous learning enables deep data integration between traditional Chinese medicine and Western medicine, resulting in more scientific assessments that better align with the principles of traditional Chinese medicine. Edge devices feature high feedback, personalized interaction, and intelligent early warning capabilities, enhancing user experience and response time. AI large-scale models are flexible in deployment and adaptable to different technical conditions and application scenarios; Weight adaptation and data standardization improve evaluation accuracy and model generalization ability; This creates a closed-loop process, enabling continuous iterative optimization of the system. It is secure and compliant with regulations, can be scaled up, and has significant industrial application value. Detailed Implementation

[0015] Example 1 (Local Deployment Mode of Open Source Large Model) Architecture Deployment: Build an edge-cloud collaborative architecture, adopt open-source AI large models for local deployment in different versions, with edge agents carrying lightweight models and cloud agents carrying complete models, and complete collaborative debugging.

[0016] Data acquisition and adaptation: The edge agent completes the acquisition, preprocessing, and standardization of user information and multi-dimensional physiological data, synchronizes them to the cloud, and realizes personalized interaction, real-time early warning and reminder services.

[0017] Model learning and alignment with traditional Chinese and Western medicine: The cloud-based intelligent agent drives the AI ​​large model to carry out machine learning based on clinical data, health records and standardized samples. It autonomously learns the intrinsic relationship between physiological data and TCM syndromes, and completes model iteration with the help of expert rule constraints.

[0018] Health assessment and weight optimization: The AI ​​big model dynamically adjusts the assessment weights and outputs diagnostic results, risk levels and early warning information.

[0019] Conditioning output and closed-loop iteration: Personalized conditioning plans are generated in the cloud, presented and feedback data is obtained at the edge, forming an incremental sample return optimization model.

[0020] Data security: End-to-end encrypted transmission and end-to-end encrypted data storage ensure compliance and security.

[0021] Example 2 (Closed-source large model parameter adaptation mode) The difference between this embodiment and Embodiment 1 is that: a closed-source AI large model parameter adaptation deployment is adopted, and full training is not performed. Only standardized data is used as calibration samples to complete parameter tuning; the edge end still retains data acquisition, personalized interaction, intelligent early warning and reminder functions. The remaining data processing, evaluation logic, conditioning output, closed-loop iteration and security mechanisms are the same, and the same evaluation and adaptation effect can be achieved in a lightweight deployment scenario.

[0022] The above description is only a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention; all modifications, equivalent substitutions, improvements, etc., within the spirit and principles of the present invention are included within the scope of protection of the present invention. Attached Figure Description Figure 1 is a block diagram of the overall system architecture of the present invention. Figure 2 is a schematic diagram of the system workflow of the present invention.

Claims

1. An AI-driven integrated traditional Chinese and Western medicine health assessment scheme and data adaptation method, characterized in that, The method is based on a flexible deployment of large AI models in an edge-cloud intelligent agent collaborative architecture. The large AI models support two independent modes: open-source training and optimization with version-based local deployment and closed-source parameter adaptation deployment. The edge intelligent agent carries a lightweight large AI model, while the cloud intelligent agent carries the complete large AI model. The method integrates the entire process of data acquisition, data adaptation, model operation, intelligent benchmarking between traditional Chinese and Western medicine, health assessment, adaptive weight adjustment, personalized interaction, intelligent early warning and reminders, and personalized suggestions to form a closed-loop system.

2. The method according to claim 1, characterized in that, Edge agents are used for personalized user interaction, acquisition of multi-source physiological data, data adaptation, offline caching, real-time anomaly identification, and emergency warning and reminder services. They synchronize standardized data to cloud agents and receive and present evaluation and suggestion results.

3. The method according to claim 1, characterized in that, Cloud-based intelligent agents are used for large-scale AI model computation, model iteration, intelligent benchmarking of traditional Chinese and Western medicine, comprehensive health assessment and data security management, to complete model training or parameter adaptation, and to build a sample-driven autonomous learning association mechanism between physiological indicators and the Zang-Xiang system of traditional Chinese medicine.

4. The method according to claim 1, characterized in that, The weight adaptive module is deployed simultaneously on the edge-cloud intelligent agent. On the edge side, it is used for preliminary data processing, anomaly identification, and real-time early warning, while on the cloud side, it is used for evaluating dynamic weight optimization, deep computing, and security collaboration.

5. The method according to claim 1, characterized in that, The weight adaptive module is deeply coupled with the AI ​​big model, dynamically adjusting the weights of evaluation indicators based on the user's physiological characteristics, physical condition and sample data, thereby improving the accuracy of the evaluation and the adaptability to different scenarios.

6. The method according to claim 1, characterized in that, The data adaptation process is executed collaboratively by edge-cloud intelligent agents, including user information structuring, data cleaning, and time series normalization, providing accurate data support for model iteration and health assessment.

7. The method according to claim 1, characterized in that, The health sensing devices used for data acquisition are based on wearable smart health devices, support multiple interface adaptations, and are uniformly accessed and scheduled by edge intelligent agents.

8. The method according to claim 1, characterized in that, Data acquisition includes basic physiological indicators, sleep-motor indicators, visual perception indicators, and advanced monitoring data such as electrocardiogram, electroencephalogram, and electrodermal conductance. These data, along with user-entered information, serve as the basis for model training and evaluation, but are not used as the sole indicator for health assessment.

9. The method according to claim 1, characterized in that, The edge-cloud intelligent agent uses encrypted communication, and the cloud intelligent agent achieves end-to-end data encryption and privacy protection, supporting cross-regional collaboration and model iteration.

Images