Method and device for intelligent analysis and evidence-based decision-making of personal health data

Abstract

The invention discloses a personal health data intelligent analysis and evidence-based decision-making method and device. The device comprises a data capturing module, a data cleaning module, a structured data processing module, a dynamic labeling module, a graph modeling module, an MOCK verification module and a graph database module. Wherein the data capturing module is used for collecting traditional Chinese and western medicine diagnosis and treatment and health data of a patient from various sources; and the data cleaning module is used for preprocessing the captured data. Through the dynamic labeling technology, the labeling strategy can be automatically adjusted according to data characteristics and application scenes, the problems of time consumption and error proneness of manual labeling are avoided, the labeling efficiency and accuracy are remarkably improved, through the map modeling technology, the complete medical knowledge map can be constructed, the deep relevance between data can be revealed, and the medical knowledge map modeling efficiency is improved. More comprehensive and deep reference is provided for clinical decision making, and the interpretability and application value of data are improved.

Description

Technical Field

[0001] The present invention belongs to the field of artificial intelligence in personal medical health technology, and in particular relates to a method and device for intelligent analysis of personal health data and evidence-based decision-making. Background Art

[0002] With the increasing prevalence of medical informatization and wearable devices, the amount of personal health data is exploding. This data encompasses a wide range of information, including both traditional Chinese and Western medical treatment records (such as physical examinations, allergen information, and daily diagnoses), physical examination indicators, and continuous monitoring data (such as fitness trackers). However, this data is generally fragmented, unstructured, dynamically changing, and complexly correlated. Existing technologies for processing this data have significant shortcomings in terms of dynamic annotation, deep correlation mining, and intelligent validation with evidence-based knowledge bases, making it difficult to efficiently integrate and utilize this data to generate personalized, evidence-based health insights and treatment recommendations. Existing technologies for medical data processing have the following major shortcomings: First, the data annotation process relies on manual or semi-automated tools, which is time-consuming and prone to errors; second, the training and analysis of AI models is complex, requiring significant computing resources and expertise; and third, the structured processing of data and the storage of analysis results are relatively simple, making them difficult to meet the needs of diverse application scenarios. Summary of the Invention

[0003] The purpose of the present invention is to provide a method and device for intelligent analysis of personal health data and evidence-based decision-making to solve the technical problems raised in the background technology.

[0004] To achieve the above objectives, the specific technical solutions of the present invention are as follows: a device for intelligent analysis of personal health data and evidence-based decision-making, including a data capture module, a data cleaning module, a structured data processing module, a dynamic annotation module, a graph modeling module, a MOCK verification module and a graph database module; wherein, The data capture module is used to collect the patient's traditional Chinese and Western medicine diagnosis and treatment and health data from various sources; The data cleaning module is used to pre-process the captured data; The structured data processing module is used to convert the cleaned data into a structured form; The dynamic annotation module is used to automatically adjust the annotation strategy according to data characteristics and application scenarios; The graph modeling module is used to construct a nonlinear association network of "constitution-allergy-disease"; The MOCK verification module is used to implement clinical evidence-based verification of individualized programs; The graph database module is used to store the final analysis results in the graph database of the ArangoDB distributed cluster by storing the health relationship graph with timestamps.

[0005] Preferably, the data capture module is connected to the data cleaning module, the data cleaning module is connected to the structured data processing module, the structured data processing module is connected to the dynamic annotation module, the dynamic annotation module is connected to the graph modeling module, the graph modeling module is connected to the graphic database module, and the graphic database module is bidirectionally connected to the MOCK verification module.

[0006] Preferably, the dynamic annotation module includes: A multi-source heterogeneous data access layer for deploying lightweight edge computing nodes, equipped with standardized API interfaces, and real-time access to physical examination reports, wearable devices, and electronic medical record data sources; A distributed annotation processing unit, comprising a node-level annotator and a collaborative annotation network. The node-level annotator is used to train a medical entity recognition model to parse raw data and generate initial labels. The collaborative annotation network is used to achieve cross-node semantic alignment using a federated learning architecture to ensure that allergens are consistently labeled in the home bracelet and hospital systems. A time series annotation database is used to store timestamped label change records.

[0007] Preferably, the lightweight edge computing node includes a home gateway and a hospital front-end server.

[0008] Preferably, the graph modeling module includes: Identity-driven data aggregators, used to generate unique health IDs based on ID numbers / biometrics, linking date of birth and baseline genetic disease history data; A graph database storage engine, used to build a property graph model using Neo4j, defining three types of entity nodes and relationship edges; Nonlinear analysis computing cluster, used for GPU-accelerated graph neural network training platform; Graph database cluster, used to build a full lifecycle graph.

[0009] Preferably, the MOCK verification module includes: Clinical knowledge base docking module, used to connect to the UpToDate clinical decision system, FDA adverse event database, and real-world research database; A high-performance simulation engine, based on the Monte Carlo method, supports 10,000+ parallel simulations. The risk-benefit assessment panel is used to visualize the output of the validation results.

[0010] The present invention also relates to a method for intelligent analysis of personal health data and evidence-based decision-making, comprising the following steps: First, we collect patients' Chinese and Western medicine diagnosis and treatment and health data from various sources, pre-process the captured data, convert the cleaned data into a structured form, automatically adjust the annotation strategy according to the data characteristics and application scenarios, extract and analyze the input data, identify key information and potential patterns in the data, and dynamically generate annotation rules and templates based on this information to guide the subsequent annotation process. We construct a "constitution-allergy-disease" nonlinear association network, perform entity recognition and relationship extraction on structured data, and generate an initial knowledge graph. Through graph fusion and reasoning technology, we integrate data from different sources and types into a unified knowledge graph to form a complete medical knowledge system. We apply the Monte Carlo simulator + FDA knowledge base interface to simulate real scenarios and implement clinical evidence-based verification of individualized solutions. We generate simulated data and scenarios based on application scenarios and needs as the basis for verification, and apply the analysis results to the simulated scenarios to observe their performance and effects. Finally, we store the health relationship graph with timestamps and the final analysis results in the graph database of the ArangoDB distributed cluster for subsequent queries and applications.

[0011] The method and device for intelligent analysis of personal health data and evidence-based decision-making of the present invention have the following advantages: 1. This invention uses dynamic annotation technology to automatically adjust annotation strategies based on data characteristics and application scenarios, avoiding the time-consuming and error-prone problems of manual annotation and significantly improving the efficiency and accuracy of annotation. Through graph modeling technology, it can construct a complete medical knowledge graph, revealing the deep correlations between data, providing a more comprehensive and in-depth reference for clinical decision-making, and improving the interpretability and application value of data. Through MOCK verification technology, the validity and reliability of analysis results can be verified in simulated real-world scenarios, ensuring the quality of the final output, improving the reliability and practicality of the results, and better meeting the needs of practical applications.

[0012] 2. This invention breaks through the three bottlenecks of static analysis, linear association, and lack of verification in existing technologies through the closed loop of "dynamic annotation-graph modeling-Mock verification", achieving a paradigm shift from passive treatment to active and precise intervention, and providing core technical support for AI family doctors. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0014] Figure 1 It is a schematic diagram of the overall structure of the present invention; Figure 2 This is a structural diagram of the dynamic annotation module in the present invention; Figure 3 This is a schematic diagram of the structure of the atlas modeling module in the present invention; Figure 4 It is a structural diagram of the MOCK verification module in the present invention. DETAILED DESCRIPTION

[0015] Hereinafter, only certain exemplary embodiments are briefly described. As will be appreciated by those skilled in the art, the described embodiments may be modified in various ways without departing from the spirit or scope of the embodiments of the present invention. Therefore, the drawings and description are to be regarded as illustrative in nature and not restrictive.

[0016] In the description of the embodiments of the present invention, it should be understood that the terms "length", "vertical", "horizontal", "top", "bottom", etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings. They are only for the convenience of describing the embodiments of the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, they cannot be understood as limitations on the embodiments of the present invention.

[0017] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of the present invention, "plurality" means two or more, unless otherwise specifically defined.

[0018] In the embodiments of the present invention, unless otherwise expressly specified or limited, the terms "installed," "connected," "connected," "fixed," etc. should be understood in a broad sense. For example, they may refer to fixed connection, detachable connection, or integration; mechanical connection, electrical connection, or communication; direct connection or indirect connection through an intermediate medium; internal communication between two components or interaction between two components. Those skilled in the art will understand the specific meanings of the above terms in the embodiments of the present invention based on specific circumstances.

[0019] The disclosure below provides many different embodiments or examples for implementing different structures of the embodiments of the present invention. In order to simplify the disclosure of the embodiments of the present invention, the components and configurations of specific examples are described below. Of course, these are merely examples and are not intended to limit the embodiments of the present invention. In addition, the embodiments of the present invention may repeat reference numerals and / or reference letters in different examples. Such repetition is for the purpose of simplicity and clarity and does not in itself indicate the relationship between the various embodiments and / or configurations discussed.

[0020] In order to better understand the purpose, structure and function of the present invention, the following is a further detailed description of a method and device for intelligent analysis of personal health data and evidence-based decision-making of the present invention in conjunction with the accompanying drawings.

[0021] like Figure 1-4 As shown, the present invention is a device for intelligent analysis and evidence-based decision-making of personal health data, including a data capture module, a data cleaning module, a structured data processing module, a dynamic annotation module, a graph modeling module, a MOCK verification module and a graph database module. The data capture module is used to collect the patient's Chinese and Western medicine diagnosis and treatment and health data from various sources; the data cleaning module is used to pre-process the captured data to ensure the quality and consistency of the data; the structured data processing module is used to convert the cleaned data into a structured form for subsequent analysis; the dynamic annotation module is used to automatically adjust the annotation strategy according to the data characteristics and application scenarios. Its function is to annotate new health events in real time (delay <10 seconds), and the device uses edge Computing nodes and a federated learning scheduling center improve the accuracy and efficiency of annotation. The graph modeling module is used to construct a nonlinear "constitution-allergy-disease" association network. The primary device utilizes a Neo4j cluster and the CROSS-Net algorithm engine to build a knowledge graph, revealing deep connections between data. The MOCK verification module implements evidence-based clinical validation of individualized plans. The device utilizes a Monte Carlo simulator and an FDA knowledge base interface to simulate real-world scenarios and verify the validity and reliability of analysis results. The graph database module stores time-stamped health relationship graphs and stores the final analysis results in the ArangoDB distributed cluster graph database for subsequent query and application. The data capture module is connected to the data cleaning module, which is then connected to the structured data processing module, which is then connected to the dynamic annotation module, which is then connected to the graph modeling module, which is then connected to the graph database module, and the graph database module is bidirectionally connected to the MOCK verification module.

[0022] The core function of the dynamic annotation module is to automatically adjust the annotation strategy based on the characteristics of the data and the application scenario. Specifically, the module first extracts and analyzes the input data to identify key information and potential patterns in the data. It then dynamically generates annotation rules and templates based on this information to guide the subsequent annotation process. The dynamic annotation module also has self-learning capabilities, continuously optimizing the annotation strategy based on the annotation results to improve the accuracy and efficiency of annotation. Furthermore, the module supports multi-user collaborative annotation, allowing multiple annotators to participate in the annotation task simultaneously. Intelligent scheduling algorithms rationally allocate tasks to further improve annotation speed and quality. The dynamic labeling module includes: a multi-source heterogeneous data access layer for deploying lightweight edge computing nodes (such as home gateways and hospital front-end servers), equipped with a standardized API interface (HL7 FHIR protocol compatible), and real-time access to more than 10 types of data sources such as physical examination reports, wearable devices, and electronic medical records; a distributed labeling processing unit, which includes a node-level labeler and a collaborative labeling network. The node-level labeler is used to train medical entity recognition models (such as BioBERT+CRF) to parse raw data and generate initial labels (such as "pollen allergy-severe"). The collaborative labeling network is used to use a federated learning architecture to achieve cross-node semantic alignment, ensuring that "allergens" are consistently labeled in home bracelets and hospital systems; a time series labeling database using a time series database (InfluxDB) to store timestamp label change records (such as "2025-06-10 09:00 | Abnormal heart rate | New").

[0023] The purpose of traditional labeling is to provide static sample labels for model training (e.g., labeling 100,000 CT scans as “lung cancer / normal”). One-time labels are generated in batches manually or through pre-trained models and then fixed in the training set. The purpose of the dynamic labeling of this invention is to create a real-time semantic mapping of an individual's health status. Time-series label streams are continuously generated through edge nodes. Traditional labeling is a preprocessing step for model training and serves static AI models; dynamic labeling, on the other hand, is a real-time status mapping system for health events, serving to build a health map for individuals throughout their lifecycle.

[0024] The logic and function of the distributed dynamic annotation engine collaborative mechanism: Dynamic incremental annotation, design T-DyAN algorithm (Time-aware Dynamic Annotation), when new data is input: 1) Extract change features (e.g., a sudden 20% increase in the standard deviation of a fitness tracker’s heart rate); 2) Triggering the local node annotator to generate a temporary label (“suspected arrhythmia”); 3) Verify label consistency through collaborative network (compared with historical ECG annotations); 4) Finally, a versioned tag (V2.3) is generated and synchronized to the central scheduler.

[0025] If there is an increment, add a note (executed on edge nodes): ```Python def real_time_annotation(data): if data.type == "sleep bracelet": # Dynamic threshold determination label = "Insufficient sleep" if data.duration < 360 else "Normal" return (label, time.now()) # Timestamp binding ``` Function: Achieve second-level annotation and update of new health events (traditional solutions have a delay of ≥6 hours).

[0026] Semantic conflict resolution: Construct label mapping rules based on the medical ontology library (SNOMED CT): ```Python def resolve_conflict(node_label, central_label): if node_label in SNOMED_CT.equivalent_class(central_label): return central_label # Adopt central label elif SNOMED_CT.subClassOf(node_label, central_label): return node_label # adopt a finer-grained label else: initiate_consensus_voting() # Start multi-node voting ``` Function: Eliminate the terminological ambiguity between "damp-heat constitution" in Traditional Chinese Medicine and "metabolic syndrome" in Western Medicine, and improve the accuracy of cross-domain data fusion.

[0027] The graph modeling module reveals the deep connections between medical data by constructing a knowledge graph, providing a more comprehensive reference for clinical decision-making. Specifically, the module first performs entity recognition and relationship extraction on structured data to generate an initial knowledge graph. Then, through graph fusion and reasoning technology, data from different sources and types are integrated into a unified knowledge graph to form a complete medical knowledge system. The graph modeling module also supports dynamic updating and maintenance of graphs, and can adjust the graph structure in a timely manner according to new data and information to maintain the timeliness and accuracy of the graph. In addition, the module also provides rich graph query and analysis functions, allowing users to explore the connections between data from multiple angles and dimensions and discover potential patterns and trends. The graph modeling module includes: an identity-driven data aggregator, which is used to generate a unique health ID based on ID number / biometric characteristics, and associate it with date of birth and baseline data on genetic disease history; a graph database storage engine, which is used to build an attribute graph model using Neo4j, defining three types of entity nodes (personal indicators, environmental factors, and medical events) and relationship edges ("induce", "inhibit", and "correlation > 0.8"); a nonlinear analysis computing cluster, which is used to train a GPU-accelerated graph neural network (GAT); and a graph database cluster, which is used to build a full life cycle graph.

[0028] Taking the personal health ID as the root node, expand the subgraph along the time axis: 1) Vertical association: Connect the birth season node ("born in winter") → the congenital constitution node ("yang deficiency constitution") → the abnormal physical examination node at age 45 ("hypothyroidism").

[0029] 2) Horizontal correlation: Merge real-time environmental data (such as PM2.5 > 100 μg / m³) with blood oxygen data from a fitness tracker.

[0030] Dynamic addition and deletion mechanism: When the labeling engine updates the "pollen allergy" label, it automatically creates a seasonal association edge between it and the "spring" node.

[0031] Deep Correlation Analysis: Developing the CROSS-Net algorithm (Cross-domain Nonlinear Interaction Solver): ```matlab function risk = CROSS_NET(physique, allergy, environment): # Build a multidimensional interaction tensor T = einsum('i,j,k->i jk', physique_embed, allergy_embed, env_embed) # Load the pre-trained GAT model to predict synergy effects risk = GAT_model.predict(T) return risk # Output respiratory disease risk coefficient ``` Example: Quantify the synergistic gain effect of "humid-heat constitution (0.7) + pollen allergy (0.9) + high humidity environment (0.6)" on the incidence of asthma (risk value = 0.92, while the traditional linear model is only 0.58).

[0032] Its function: to solve the problem of data silos, build a health evolution network spanning a 20+ year cycle, capture multi-factor nonlinear interactions, and warn of complex health risks.

[0033] The MOCK Validation Module simulates real-world scenarios to verify the validity and reliability of analytical results, ensuring the quality of the final output. Specifically, the module first generates simulated data and scenarios based on the application scenario and requirements as the basis for validation. The analytical results are then applied to the simulated scenarios to observe their performance and effectiveness. The MOCK Validation Module also features an intelligent evaluation and feedback mechanism that automatically assesses the accuracy and applicability of analytical results and generates detailed evaluation reports and improvement suggestions. Furthermore, the module supports multiple rounds of iterative validation, allowing users to continuously optimize analytical methods and parameters based on the evaluation results until satisfactory results are achieved. Application of the MOCK Validation Module effectively enhances the credibility and practicality of analytical results, providing a reliable basis for clinical decision-making. The MOCK Validation Module includes a clinical knowledge base integration module for connecting to the UpToDate clinical decision system, the FDA adverse event database, and real-world research repositories (such as the Flatiron HealthOncology dataset); a standardized simulation engine, based on Monte Carlo methods, capable of supporting over 10,000 parallel simulations; and a risk-benefit assessment dashboard for visualizing validation results (e.g., efficacy probability distribution plots and adverse reaction heatmaps).

[0034] Mock verification closed-loop optimization analysis method and function: Individualized solution verification: 1) Receive diagnosis and treatment recommendations output by graph analysis (e.g., "Metformin is recommended for glucose control"); 2) Extract matching conditions from the knowledge base: Clinical Guidelines: ADA 2025 Diabetes Medication Guidelines; Real cases: Medication records of 1,000 patients with "damp-heat constitution + obesity"; contraindication database: marked "Metformin is contraindicated: eGFR<45".

[0035] Start a Monte Carlo simulation: ```java MockResult runMock(PatientProfile pp, DrugPlan dp) { for (int i=0; i<10000; i++) { / / Random sampling individual variables (fluctuation values ​​of liver and kidney function) double eGFR = sampleNormal(pp.base_eGFR, pp.std_dev); / / Detect taboo trigger if (eGFR < 45 && dp.contains("Metformin")) outcome[i] = "ADVERSARIAL"; else outcome[i] = predictEfficacy(dp, pp.genotype); } return new MockResult(outcome); / / Returns 95% confidence interval result } ``` Output: "Metformin has an 82.3% chance of being effective for this patient, but there is a 12.7% chance of triggering contraindications when eGFR fluctuates."

[0036] Dynamic regimen optimization: When a contraindication risk >10% is detected, alternative regimens (such as "SGLT2 inhibitors") are automatically retrieved; Mock verification is repeated until the risk is <5%; and the final regimen is generated: "Empagliflozin is recommended (efficacy 78.5% | risk 4.2%)."

[0037] What it does: Reduces individual medication error rates from 15-20% with traditional regimens to <5%.

[0038] Through the collaborative innovation of three major modules, the present invention can realize the closed loop of dynamic labeling, in-depth association, evidence-based verification and scenario-based decision-making of personal health data, providing a complete technical path to break the core bottleneck of precision medicine.

[0039] The present invention also relates to a method for intelligent analysis of personal health data and evidence-based decision-making, comprising the following steps: First, we collect patients' Chinese and Western medicine diagnosis and treatment and health data from various sources, pre-process the captured data, convert the cleaned data into a structured form, automatically adjust the annotation strategy according to the data characteristics and application scenarios, extract and analyze the input data, identify key information and potential patterns in the data, and dynamically generate annotation rules and templates based on this information to guide the subsequent annotation process. We construct a "constitution-allergy-disease" nonlinear association network, perform entity recognition and relationship extraction on structured data, and generate an initial knowledge graph. Through graph fusion and reasoning technology, we integrate data from different sources and types into a unified knowledge graph to form a complete medical knowledge system. We apply the Monte Carlo simulator + FDA knowledge base interface to simulate real scenarios and implement clinical evidence-based verification of individualized solutions. We generate simulated data and scenarios based on application scenarios and needs as the basis for verification, and apply the analysis results to the simulated scenarios to observe their performance and effects. Finally, we store the health relationship graph with timestamps and the final analysis results in the graph database of the ArangoDB distributed cluster for subsequent queries and applications.

[0040] This invention utilizes dynamic annotation, graph modeling, and MOCK verification to achieve intelligent processing and analysis of medical data. Specifically, the dynamic annotation module automatically adjusts annotation strategies based on data characteristics and application scenarios, improving annotation accuracy and efficiency. The graph modeling module constructs a knowledge graph to reveal deep connections between data, providing a more comprehensive reference for clinical decision-making. The MOCK verification module simulates real-world scenarios to verify the validity and reliability of analysis results, ensuring the quality of the final output. Its application scenarios specifically target deep intelligent analysis of electronic medical records, precision upgrades to clinical decision support systems (CDSS), and the construction of dynamic medical knowledge bases. Its core value lies in significantly improving the integration, analytical depth, and evidence-based utilization efficiency of personal medical and health data, providing core technical support for the realization of a "personal AI smart family doctor." Furthermore, this solution can serve as a powerful auxiliary tool, enabling hospital physicians to quickly gain in-depth insights based on patient data throughout their lifecycle during consultations, significantly improving diagnostic efficiency and accuracy, and promoting the development of precision medicine and proactive health management.

[0041] It will be understood that the present invention is described by way of some embodiments, and it will be appreciated by those skilled in the art that various changes or equivalent substitutions may be made to these features and embodiments without departing from the spirit and scope of the present invention. In addition, under the teachings of the present invention, these features and embodiments may be modified to adapt to specific circumstances and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are intended to be protected by the present invention.

Claims

1. A device for intelligent analysis of personal health data and evidence-based decision-making, characterized by: It includes data capture module, data cleaning module, structured data processing module, dynamic annotation module, graph modeling module, MOCK verification module and graph database module; among them, The data capture module is used to collect the patient's traditional Chinese and Western medicine diagnosis and treatment and health data from various sources; The data cleaning module is used to pre-process the captured data; The structured data processing module is used to convert the cleaned data into a structured form; The dynamic annotation module is used to automatically adjust the annotation strategy according to data characteristics and application scenarios; The graph modeling module is used to construct a nonlinear association network of "constitution-allergy-disease"; The MOCK verification module is used to implement clinical evidence-based verification of individualized programs; The graph database module is used to store the final analysis results in the graph database of the ArangoDB distributed cluster by storing the health relationship graph with timestamps.

2. The device for intelligent analysis of personal health data and evidence-based decision-making according to claim 1, characterized in that: The data capture module is connected to the data cleaning module, the data cleaning module is connected to the structured data processing module, the structured data processing module is connected to the dynamic annotation module, the dynamic annotation module is connected to the graph modeling module, the graph modeling module is connected to the graphic database module, and the graphic database module is bidirectionally connected to the MOCK verification module.

3. The device for intelligent analysis of personal health data and evidence-based decision-making according to claim 1, characterized in that: The dynamic annotation module includes: A multi-source heterogeneous data access layer for deploying lightweight edge computing nodes, equipped with standardized API interfaces, and real-time access to physical examination reports, wearable devices, and electronic medical record data sources; A distributed annotation processing unit, comprising a node-level annotator and a collaborative annotation network. The node-level annotator is used to train a medical entity recognition model to parse raw data and generate initial labels. The collaborative annotation network is used to achieve cross-node semantic alignment using a federated learning architecture to ensure that allergens are consistently labeled in the home bracelet and hospital systems. Time series annotation database, used to store timestamped label change records.

4. The device for intelligent analysis of personal health data and evidence-based decision-making according to claim 3, characterized in that: The lightweight edge computing nodes include a home gateway and a hospital front-end server.

5. The device for intelligent analysis of personal health data and evidence-based decision-making according to claim 1, characterized in that: The graph modeling module includes: Identity-driven data aggregators, used to generate unique health IDs based on ID numbers / biometrics, linking date of birth and baseline genetic disease history data; A graph database storage engine, used to build a property graph model using Neo4j, defining three types of entity nodes and relationship edges; Nonlinear analysis computing cluster, used for GPU-accelerated graph neural network training platform; Graph database cluster, used to build a full lifecycle graph.

6. The device for intelligent analysis of personal health data and evidence-based decision-making according to claim 1, characterized in that: The MOCK verification module includes: Clinical knowledge base docking module, used to connect to the UpToDate clinical decision system, FDA adverse event database, and real-world research database; A high-performance simulation engine, based on the Monte Carlo method, supports 10,000+ parallel simulations. The risk-benefit assessment panel is used to visualize the output of the validation results.

7. A method for intelligent analysis of personal health data and evidence-based decision-making, characterized by: The steps include: First, we collect patients' traditional Chinese and Western medicine diagnosis and treatment, and health data from various sources. We preprocess the captured data and convert the cleaned data into a structured form. We automatically adjust the annotation strategy based on the data characteristics and application scenarios, extract and analyze the input data, identify key information and potential patterns in the data, and dynamically generate annotation rules and templates based on this information to guide the subsequent annotation process. We construct a "constitution-allergy-disease" nonlinear association network, perform entity recognition and relationship extraction on the structured data, and generate an initial knowledge graph. Through graph fusion and reasoning techniques, we integrate data from different sources and types into a unified knowledge graph to form a complete medical knowledge system. We apply the Monte Carlo simulator + FDA knowledge base interface to simulate real-world scenarios and implement clinical evidence-based verification of individualized solutions. We generate simulated data and scenarios based on application scenarios and needs as the basis for verification, and then apply the analysis results to the simulated scenarios to observe their performance and effectiveness. Finally, we store the health relationship graph with timestamps and the final analysis results in the graph database of the ArangoDB distributed cluster for subsequent query and application.

Images