Health management agent system based on dynamic space-time calibration and detection method

Abstract

The invention relates to the technical field of intelligent medical treatment and health management Internet of Things, in particular to a health management agent system based on dynamic space-time calibration and a detection method, and the method comprises the steps: obtaining health state data obtained based on user multi-modal physiological data analysis, and environment space-time data containing a geographic position; the personalized representation associated with the natural rhythm is generated based on the environmental spatio-temporal data, the representation and the health state data are fused, a personalized intervention strategy is generated in combination with gender and age, and dynamic optimization can be performed through user feedback; the system comprises a data acquisition module, an environment space-time processing module, a health state processing module, a strategy generation module, and a support federated learning and generative AI module. The equipment is integrated with a multi-mode physiological collector, an air pressure altitude sensor and other components. According to the invention, accurate health management of'time, place and people 'can be effectively achieved, the health conditioning effect is remarkably improved, and the whole-cycle health guarantee is formed.

Description

Technical Field

[0001] This invention belongs to the field of smart healthcare and health management Internet of Things technology, specifically relating to a health management intelligent agent system and detection method based on dynamic spatiotemporal calibration. Background Technology

[0002] The Twenty-Four Solar Terms are a precise summary by the ancient Chinese people of the periodic changes in environmental factors such as sunlight, heat, and precipitation caused by the Earth's revolution around the sun. Essentially, they are accurate representations of specific climatic and phenological stages and a core component of the traditional Chinese medicine theory of "harmony between man and nature," providing an important temporal dimension for traditional health practices. However, with the modernization and intelligentization of health management, the traditional concept of adapting to the seasons faces multiple technical bottlenecks in practical application, hindering the efficient transformation of its scientific value and limiting its widespread adoption. Specifically: (1) See Figure 1 The map shows latitude (°N) on the horizontal axis and altitude (m) on the vertical axis. It marks the three-dimensional correspondence of latitude-altitude-phenology for cities such as Zhengzhou (34.76°N, 100m, peach blossoms bloom around March 20th, before and after the Spring Equinox, serving as a reference point for the origin of the Twenty-Four Solar Terms), Beijing (39.93°N, 43m, peach blossoms bloom around April 5th, before and after Qingming Festival), Guangzhou (23.06°N, 26m, peach blossoms bloom around February 5th, before and after the Beginning of Spring), Chengdu (30.6°N, 500m, peach blossoms bloom between the Awakening of Insects and the Spring Equinox around March 10th), and Shanghai (31.23°N, 5m, peach blossoms bloom before March 10th, before the Awakening of Insects). It is evident that under the same standard solar term (such as the vernal equinox), phenological phenomena can differ by 15-20 days in different regions due to latitude and altitude (e.g., Beijing is about 50 days later than Guangzhou). This directly demonstrates that existing general recommendations based on fixed solar term dates are severely out of sync with users' actual phenological conditions, significantly reducing their effectiveness. Therefore, most existing health management applications provide general health advice based on fixed solar term dates, completely ignoring the vast regional climate differences across China. Significant time differences exist in phenological phenomena across different latitudes and altitudes, and the substantial differences in actual phenological phenomena (such as flowering and leaf fall) corresponding to the same standard solar term date clearly reveal the limitations of providing health advice solely based on fixed solar term dates. This directly leads to a severe mismatch between the pushed health advice and the actual phenological conditions in the user's location, reducing the effectiveness of the treatment.

[0003] (2) Current health management systems have two typical limitations: one type of system can identify a user's TCM constitution through technical means, but the generated conditioning plan remains static and cannot be adjusted according to the dynamic changes of the seasons and climate; the other type of system introduces the concept of seasons, but only provides general suggestions that are the same for everyone, without customizing them according to the individual differences of the user's constitution. Both types of systems fail to deeply integrate the dynamically changing information of the seasons and regional climate with the individual characteristics of the user's constitution, and cannot achieve true precise adaptation according to the time, place and person.

[0004] (3) The lack of convenient and objective physical data collection terminals and closed-loop feedback mechanisms makes it difficult to continuously and quantitatively track changes in users' physiological states, resulting in the inability to accurately verify and iteratively optimize diagnostic and treatment plans. Furthermore, there is a bottleneck in the disconnect between prevention and treatment. A significant gap exists between current health management platforms and clinical diagnosis and treatment systems; the former lacks the capability to evolve into the latter. When users progress from a sub-healthy state to a disease state, their long-accumulated health data and personalized models cannot seamlessly serve subsequent clinical decision support, leading to a disruption in the continuity of health management and preventing the formation of a closed-loop health protection system covering the entire lifecycle from "pre-disease to impending disease to existing disease."

[0005] In view of this, the present invention is hereby proposed. Summary of the Invention

[0006] To address the aforementioned technical problems in existing technologies, this invention provides a health management intelligent agent system and detection method based on dynamic spatiotemporal calibration. This solves the problem that existing technologies "provide general health advice based on fixed standards, without acquiring environmental spatiotemporal data such as the user's geographical location and generating a personalized environmental spatiotemporal representation that dynamically relates to the user's actual natural rhythm, and without integrating and analyzing this representation with the user's health status data, resulting in the inability to generate personalized health intervention strategies that adapt to the user's actual natural rhythm and individual health status, making it difficult to achieve precise health management."

[0007] To achieve the above objectives, the technical solution of the present invention is as follows: Firstly, a health management intelligent agent system based on dynamic spatiotemporal calibration includes: The data acquisition module is used to acquire users' health status data and environmental spatiotemporal data; An environmental spatiotemporal processing module is used to process the environmental spatiotemporal data to generate a personalized environmental spatiotemporal representation. A health status processing module is used to process the health status data to generate a health status representation; The strategy generation module is used to integrate the personalized environmental spatiotemporal representation with the health status representation and output personalized health intervention strategies. Terminal visualization module: Used to establish data interaction with data acquisition module, health status processing module, environmental spatiotemporal processing module and strategy generation module, realize the visualization guidance of associated data acquisition, the visualization display of associated strategy generation and support the feedback data reception and transmission function for closed-loop optimization.

[0008] Furthermore, the data acquisition module includes: Multimodal physiological sensing unit: configured to collect users' multimodal physiological data via an AI-powered intelligent diagnostic terminal; Environmental spatiotemporal perception unit: configured to acquire real-time latitude, longitude and altitude data of the user through a high-precision positioning module, and acquire real-time meteorological data, historical meteorological data and phenological observation data of the user's location through a meteorological data API interface.

[0009] Furthermore, the health status processing module includes: Multimodal health status fusion unit: used to construct a health status assessment system based on DS evidence theory and generate health status representation; Concept digitization mapping unit: used to handle ambiguity and subjectivity, transforming relevant concepts into quantifiable and calculable numerical parameters, providing digital support for data interaction and collaborative operation between various functional modules of the system; Closed-loop learning and optimization unit: used to build an adaptive learning mechanism based on Bayesian networks, combining historical health data and real-time user feedback information to optimize system model parameters and personalized health intervention strategies.

[0010] Furthermore, the strategy generation module includes: Intelligent health preservation knowledge base and compatibility engine unit: used for structured intelligent health preservation knowledge base of food and medicine homology; The "Food and Medicine from the Same Source" recommendation unit is used to intelligently match and fine-tune the combination of food and Chinese medicine based on the user's current constitution, degree of imbalance, solar term, local climate characteristics, gender, and age. Food and medicine homology cooking unit: used to match the appropriate cooking method according to the scheme provided by the food and medicine homology recommendation unit; Exercise and rest suggestion unit: It is used to create an exercise and rest plan that includes suitable daily routines, exercise plans and intensity, based on the user's local sunrise and sunset times, temperature and humidity changes and personal lifestyle habits; Natural Language Generation and Interaction Unit: Used to integrate generative artificial intelligence technology, it receives structured data output from the solution generation engine and transforms it into health guidance content.

[0011] Furthermore, it also includes: Gender Differentiation Generation Unit: Used to fine-tune the scheme based on gender differentiation; Age-Stage Adaptation Unit: Used to prioritize and fine-tune the treatment plan according to the physiological characteristics and core treatment principles of each stage; Closed-loop management and optimization unit: Regularly revisit users through AI terminals to collect updated physiological indicator data; analyze the interaction between the evolution trend of physical condition and changes in meteorological data through multi-dimensional algorithms, and dynamically optimize conditioning strategies and model parameters.

[0012] Furthermore, the terminal visualization module includes: Acquisition guidance unit: used to receive the acquisition request from the data acquisition module, provide acquisition process guidance when acquiring multimodal physiological data, and display the acquisition progress and data verification status in real time; Strategy Display Unit: Used to receive the personalized health intervention strategy output by the strategy generation module and the personalized environmental spatiotemporal representation generated by the environmental spatiotemporal processing module, present the health intervention strategy on the user's smart device client, and simultaneously display the calibrated spatiotemporal and phenological related information. Feedback interaction unit: used to receive user feedback data on the effect of health intervention strategies and transmit the feedback data to the system to support closed-loop optimization of the health status processing module and the strategy generation module; The data backtracking unit is used to receive the historical health status representation of the health status processing module and the historical personalized environmental spatiotemporal representation of the environmental spatiotemporal processing module, and to display the historical health status, historical personalized environmental spatiotemporal representation and execution records of past health intervention strategies on the user's smart device client. The natural language interpretation subunit is integrated into the strategy display unit to convert structured health intervention strategies into natural language interpretation content and present it on the user's smart device client. The solution interaction subunit is integrated into the strategy display unit. When presenting solutions related to medicine and food in health intervention strategies, it marks the role of food pairing and displays the efficacy descriptions of each pairing role to the user through interactive operations.

[0013] Furthermore, it also includes: The exercise-assisted display subunit is integrated into the strategy display unit to display visual content related to exercise movements when presenting exercise-related solutions in health intervention strategies, and to provide corresponding playback control functions. The retest reminder subunit is integrated into the feedback interaction unit. It receives the physical fitness retest cycle requirement from the health status processing module and displays a visual reminder and retest process guide on the user's smart device client when the preset time before the next physical fitness retest is approaching.

[0014] Secondly, a health management intelligent agent detection method based on dynamic spatiotemporal calibration includes: S1. Obtain the user's health status data; S2. Obtain the spatiotemporal data of the user's environment, wherein the spatiotemporal data includes at least the geographical location; S3. Based on the aforementioned environmental spatiotemporal data, generate a dynamic, personalized environmental spatiotemporal representation that is associated with the user's actual natural rhythm. S4. Based on the fusion analysis of the health status data and the personalized environmental spatiotemporal representation, a personalized health intervention strategy is generated.

[0015] Furthermore, it also includes: Receive feedback data from users after they have implemented the health intervention strategy, including updated physiological indicator data and / or user subjective experience data; Based on the feedback data, the analysis model of the health status data and / or the generation strategy of the health intervention strategy are dynamically adjusted through machine learning algorithms.

[0016] Furthermore, when generating the personalized health intervention strategy, at least one of the user's gender information and age stage information is further integrated, and the intervention element type, target point or execution intensity in the strategy is configured differently based on this.

[0017] Compared with existing technologies, the present invention provides a health management intelligent system and detection method based on dynamic spatiotemporal calibration. The system includes modules for data acquisition, environmental spatiotemporal processing, health status processing, and strategy generation, supporting federated learning and generative AI. The device integrates components such as a multimodal physiological data collector and a barometric altimeter sensor. By dynamically generating personalized environmental spatiotemporal representations related to the user's actual natural rhythms and deeply integrating them with the user's health status data, it achieves differentiated health interventions based on gender and age. It can also optimize strategies and connect with clinical diagnosis and treatment based on closed-loop feedback, effectively achieving precise health management "according to time, place, and person". The method includes: acquiring health status data derived from the user's multimodal physiological data and environmental spatiotemporal data containing geographical location; generating personalized representations related to natural rhythms based on environmental spatiotemporal data; integrating these representations with health status data; generating personalized intervention strategies based on gender and age; and dynamically optimizing them through user feedback. This significantly improves the health conditioning effect and forms a full-cycle health guarantee. Attached Figure Description

[0018] Figure 1 A schematic diagram illustrating the influence of latitude difference and altitude on phenology provided by this invention; Figure 2 This is an architecture diagram of the intelligent health monitoring and management system provided in an embodiment of the present invention; Figure 3 A flowchart of the intelligent health monitoring and management method provided in an embodiment of the present invention; Figure 4 The flowchart of the climate and solar term adaptive module provided in the embodiment of the present invention is shown. Figure 5 This is a schematic diagram of a user terminal application interface provided in an embodiment of the present invention; Figure 6 A flowchart illustrating the generation of the knowledge base and integrated solution provided in embodiments of the present invention; Figure 7 A flowchart of the intelligent health monitoring and management method provided in an embodiment of the present invention. Detailed Implementation

[0019] The technical solution of the present invention will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are not all embodiments of the present invention. All other embodiments obtained by those skilled in the art without creative effort are within the protection scope of the present invention.

[0020] It should be noted that, unless otherwise specifically stated, the relative arrangement and numerical expressions of the components and steps described in these embodiments should not be construed as limiting the scope of the invention.

[0021] The following description of exemplary embodiments is merely illustrative and is not intended to limit the invention or its application or use in any way. Techniques, methods, and apparatus known to those skilled in the art may not be discussed in detail herein, but where applicable, such techniques, methods, and apparatus should be considered part of this specification.

[0022] Example 1 See Figure 2 , Figure 2 This is an architecture diagram of a health management intelligent agent system based on dynamic spatiotemporal calibration proposed in this invention. The system is named "Dumu Zhikang". The system adopts a three-layer architecture design consisting of a terminal layer, a user layer, and a cloud platform layer. The correspondence between each module and the architecture layer is as follows: The terminal layer includes an AI-powered intelligent diagnostic terminal (integrating the M11 multimodal physiological sensing unit under the M1 data acquisition module, used to collect user tongue, pulse, facial, and acoustic data) and a barometric altimeter sensor (integrated into the M1 data acquisition module, used to collect real-time altitude data of the user's location). The user layer is responsible for implementing human-computer interaction functions, covering operations such as tongue diagnosis data upload, plan viewing, and feedback submission, corresponding to the user operation entry provided by the M5 terminal visualization module. The cloud platform layer is the core processing layer of the system, including a multimodal evidence processing unit (corresponding to the DS evidence fusion function of the M3 health status processing module), a climatology learning system (corresponding to the historical phenological data fitting work of the M2 environmental spatiotemporal processing module, used to determine the values ​​of latitude-phenological delay coefficient and altitude-phenological delay coefficient), a feedback F optimization data unit (corresponding to the model and parameter update function of the M47 closed-loop management and optimization unit under the M4 strategy generation module), and an intelligent plan generation unit (corresponding to the M4 strategy generation module). The terminal layer, user layer, and cloud platform layer interact through encrypted data transmission, jointly supporting the operation of the dynamically coupled model.

[0023] This is a "spatiotemporal-individual" dynamically coupled digital health model based on the integration of a dynamic spatiotemporal coordinate system of "four seasons-eight solar terms-sixteen solar terms-seventy-two pentads" and an individual health baseline of "nine types of TCM constitutions." The dynamic spatiotemporal benchmark allows the system to deepen the traditional twenty-four solar terms into a multi-level model. Four seasons: spring, summer, autumn, and winter; The eight solar terms (four beginnings, two equinoxes, and two solstices): Beginning of Spring, Spring Equinox, Beginning of Summer, Summer Solstice, Beginning of Autumn, Autumn Equinox, Beginning of Winter, and Winter Solstice serve as nodes of dramatic energy changes during seasonal transitions; The Sixteen Solar Terms: Rain Water, Awakening of Insects, Pure Brightness, Grain Rain, Grain Buds, Grain in Ear, Minor Heat, Major Heat, End of Heat, White Dew, Cold Dew, Frost's Descent, Minor Snow, Major Snow, Minor Cold, Major Cold, describe the gradual evolution of climate and phenology. The Seventy-Two Pentads: Each solar term is subdivided into three phenological periods (one pentad every 5 days), totaling seventy-two precise phenological phenomena, which serve as the core basis for the systematic implementation of refined health preservation based on the phenological conditions; Individual health baseline: Based on nine types of TCM constitutions, combined with gender and age, it forms the cornerstone of personalized plans; Dynamic Spatiotemporal Context: The system dynamically constructs a dynamic spatiotemporal context for each user based on environmental perception data. This context is a digital information field that encapsulates time-series information (such as phenological periods accurate to the seventy-two pentads) and regional climate characteristics, processed by a three-dimensional dynamic calibration algorithm and synchronized with the actual natural rhythms of the user's location. This enables health regimens to achieve dynamic and precise synchronization with local natural rhythms. Specifically, this includes: M1, the data acquisition module, is configured to acquire user health status data and environmental spatiotemporal data, and is integrated into the barometric altimeter sensor on the terminal device, configured to acquire real-time altitude data of the user's location. Specifically, it includes: M11, Multimodal Physiological Sensing Unit: Configured to collect users' multimodal physiological data through an AI-powered intelligent diagnostic terminal; The AI-powered intelligent diagnostic terminal integrates multiple sensors, such as a high-definition camera, pulse sensor, and voiceprint acquisition module, to non-invasively collect the user's tongue image (reflecting tongue coating and tongue body characteristics), pulse image (radial artery pulse signal), facial image (facial skin color and texture characteristics), and voice image (speech signal in a stable state). After the collection is completed, the terminal uploads the raw physiological data to the cloud server.

[0024] M12, Environmental Spatiotemporal Sensing Unit: Configured to acquire real-time latitude, longitude and altitude data of the user through a high-precision positioning module, and to acquire real-time meteorological data, historical meteorological data and phenological observation data of the user's location through a meteorological data API interface.

[0025] Real-time latitude and longitude (accuracy sufficient for positioning) is obtained through the BDS / GPS module built into the user's smartphone, and the digital elevation model (DEM) database is accessed to match the elevation data of the user's location. Third-party meteorological data APIs (such as the China Meteorological Administration's open platform) are called to obtain real-time meteorological data (temperature, humidity, air pressure) and historical meteorological data for the same period over the past three years for the user's location. Simultaneously, the publicly available phenological dataset from the Chinese Ecosystem Research Network (CERN) is accessed to obtain historical phenological observation records for the user's region (such as the correspondence between plant flowering periods and solar terms), providing foundational data for subsequent environmental spatiotemporal processing. M2, an environmental spatiotemporal processing module, is configured to process the environmental spatiotemporal data to generate a personalized environmental spatiotemporal representation; see [link / reference]. Figure 4 This module processes environmental spatiotemporal data using a three-dimensional dynamic calibration algorithm (integrating latitude difference, real-time weather, and altitude dimensions) to generate a personalized environmental spatiotemporal representation synchronized with the user's actual natural rhythms, thus solving the spatiotemporal disconnect problem of existing technologies. Specifically, it includes: M21. Reference location setting: Obtain the user's precise geographical location and use Zhengzhou City, Henan Province (34.76°N), which is the representative of the Central Plains region where the Twenty-Four Solar Terms originated, as the reference location and the origin point for calculating the solar term offset. M22, Base Offset Calculation: Calculate the latitude difference with the Central Plains region (Zhengzhou, Henan, 34.76°N, 100m), based on the latitude difference between the user's location and Zhengzhou, according to the formula:

[0026] in, This represents the number of days (in days) that the solar term is offset from. The latitude (degrees) of the user's location. The reference location is the average latitude (34.76°N). This is the latitude-phenological delay coefficient (days / degree). It is obtained by fitting historical phenological observation data, with an optimal value range of 0.6 to 1.2 days / degree, and a typical value of 0.8 days / degree. The coefficient k reflects the average number of days of phenological delay caused by each degree of latitude change, and is a regional empirical parameter obtained by fitting a large amount of historical phenological and meteorological data.

[0027] When the latitude of the user's location Higher than hour, A positive value indicates that the seasonal health advice should be postponed. |day;when Below hour, A negative value indicates that it should be done in advance. |Heaven.

[0028] The latitude-phenological lag coefficient k and the altitude-phenological lag coefficient m are regional empirical parameters obtained by fitting historical phenological observation data with corresponding latitude, longitude, altitude, and meteorological data through multiple regression analysis. They can be continuously optimized through the closed-loop management and optimization module.

[0029] M23. Real-time Weather Fine-tuning: Acquires data (such as temperature and humidity) through a real-time weather data interface, compares the current real-time data with the historical average for the same period, and calculates according to preset rules (e.g., add 1 day if the low temperature exceeds 3°C for 3 consecutive days). Fine-tune the base offset.

[0030] Fine-tuning rules can be preset based on the characteristics of different solar terms. For example, if the current average temperature for the three consecutive days is lower than the historical average for the same period by more than [a certain amount], [the rules can be adjusted accordingly]. Then the basic offset Add 1 day (postponed); if humidity is 15% higher than the historical average for this time of year, then... Reduce by 0.5 days (advance). The fine-tuning range can be set to an upper limit (e.g., ±2 days). The above fine-tuning rules and thresholds can be adaptively set according to the core climate characteristics of different solar terms (e.g., temperature for the Spring Equinox, humidity for Grain Rain), further increasing the flexibility of the model.

[0031] M24, Altitude Correction: Altitude is a key factor affecting local climate. Based on the difference between the user's location altitude and Zhengzhou's altitude, the altitude correction is calculated using the following formula:

[0032] in, Altitude correction offset (days). The altitude (m) of the user's location. The elevation (in meters) is for reference. The altitude-phenological delay coefficient (days / 100m); It can be sea level (0m) or the average elevation of the Central Plains region (e.g., 100m); The preferred value range is 0.5 to 1.5 days / 100m, with a typical value of 1.0 day / 100m.

[0033] M25. Final Offset Synthesis: The base offset, real-time meteorological fine-tuning, and altitude correction are added together to obtain the final solar term offset, expressed as follows:

[0034] in, The basic offset is based on the latitude difference. These are fine-tuned values ​​based on real-time meteorological data. The correction value is based on altitude; this three-dimensional calibration model can ensure that the output time point is highly consistent with the actual local phenological phenomena.

[0035] M26. Phenological Period Determination: Based on the user's geographical location, real-time meteorological data, and historical phenological data, the precise phenological period the user is currently in is determined by combining a phenological period calculation function; the specific expression is as follows:

[0036] This function can be implemented using a lightweight machine learning classification model (such as a decision tree or gradient booster model), which takes geolocation codes and real-time meteorological features (temperature, humidity, precipitation, etc.) as input and outputs the most likely phenological period classification.

[0037] M27. Generate personalized environmental spatiotemporal representation: Integrate calibrated solar terms (including final offset), determined phenological periods, regional climate characteristics (such as humidity and dryness), and spatiotemporal weight parameters (such as the adaptation weights for dehumidification and warming the sun) to form a personalized environmental spatiotemporal representation that can be called by subsequent strategy generation units.

[0038] M3, the health status processing module, is configured to process the health status data to generate a health status representation. The core function of this module is to overcome the technical bottleneck of multi-source physiological data fusion, construct a robust health status assessment system through DS evidence theory, and improve the accuracy and robustness of physical condition identification. Specifically, it includes: M31, Multimodal Health Status Fusion Unit: Used to fuse multi-source physiological data such as user tongue image, pulse image, facial image, and voice image to construct a health status assessment system and generate a health status representation that includes the user's dominant physical constitution and assessment uncertainties; specifically including: M311, Implementation methods based on DS evidence theory include: Construct an identification framework, assuming that the nine basic constitutions in Traditional Chinese Medicine constitute a mutually exclusive and complete identification framework, denoted as:

[0039] in, to These represent the balanced constitution, qi deficiency constitution, yang deficiency constitution, yin deficiency constitution, phlegm-dampness constitution, damp-heat constitution, blood stasis constitution, qi stagnation constitution, and special constitution, respectively; the power set of this framework. It includes all possible combinations of constitutions. The function is a mapping: And satisfy:

[0040] in, For this evidence, the proposition is correct. The degree of support (i.e., a certain single constitution or combination of constitutions).

[0041] For example, for tongue image evidence ( ), its basic probability allocation Possible:

[0042]

[0043]

[0044] Similarly, pulse evidence ( )Establish , as facial evidence ( )Establish And so on. These assignments are based on data mining of a large number of labeled medical records and expert knowledge bases.

[0045] M312, Evidence Fusion and Synthesis: The system employs Dempster's combination rule to pairwise fuse the basic probability assignments of multiple pieces of evidence. For two pieces of evidence... and The corresponding basic probability allocation and The synthesis rules are as follows:

[0046] in, This is a normalization constant used to measure the degree of conflict between pieces of evidence; its calculation formula is as follows:

[0047] The fusion computation steps include: calculating the collision quality. The specific formula is:

[0048] Calculate the normalization constant K: For each proposition in the identification framework Find all that satisfy Combine the evidence, multiply their masses, and add them together; then divide the sum by the normalization constant. That is, to obtain the new basic probability assignment after fusion. .

[0049] M313. The system combines information layer by layer (e.g., first integrating tongue and pulse findings, then integrating the results with facial findings) to ultimately obtain a comprehensive identification framework that integrates all the information from the four diagnostic methods. Final basic probability assignment .

[0050] Health status representation output: After fusion, the system calculates the constitution of each individual. Reliability function and similarity function ; confidence function , indicating that the proposition The minimum level of support.

[0051] A plausible function represents a statement that is true for a proposition. The maximum possible level of support is calculated using the following formula:

[0052] The system selects the constitution with the highest reliability function (Bel) as the user's dominant constitution. Simultaneously, the reliability value itself serves as the confidence level for constitution determination, and the interval formed by the similarity function and the reliability function... These indicators reflect the range of uncertainty in the judgment results and are used for fine-tuning of the subsequent scheme generation module.

[0053] M32, Concept Digital Mapping Unit: Used to handle the ambiguity and subjectivity of TCM concepts, transforming unstructured TCM-related concepts such as tongue color, pulse, and constitution degree into quantifiable numerical parameters, providing digital support for data interaction and collaborative operation between various functional modules of the system; This module addresses the ambiguity and subjectivity of TCM concepts (such as tongue color, pulse, and constitution level) by transforming unstructured TCM concepts into quantifiable and computable numerical parameters through a fuzzy logic system. This provides a unified digital foundation for collaborative computation across all modules of the system, specifically including: M321. Fuzzification of Traditional Chinese Medicine Concepts: For key TCM concepts in health management, define corresponding fuzzy variables and value sets, transforming qualitative descriptions into categorizable fuzzy categories. Specifically, this includes: Tongue color fuzzy variable: The value set is {pale white, light red, red, crimson, bluish-purple} (covering the main TCM classifications of tongue color); Fuzzy variables for pulse diagnosis: The set of values ​​is {floating, deep, slow, rapid, slippery, rough, wiry, tight} (covering the core TCM characteristics of pulse diagnosis). Fuzzy variable for physical condition severity: The value set is {mild, moderate, severe} (quantifying the severity of physical condition bias).

[0054] M322. Membership Function Design: Triangular and trapezoidal membership functions are used to quantify the continuous membership degree of fuzzy variables, achieving precise mapping between fuzzy categories and numerical values. Taking tongue color as an example (light red), the membership function is defined as follows:

[0055] in, The digital original parameters of tongue color (such as the normalized values ​​converted from color saturation, brightness, etc.) for The degree to which a value belongs to the light red category (value range [0,1], the closer the value is to 1, the higher the degree of membership).

[0056] when At that time, membership degree This indicates that the tongue color value does not fall into the pale red category; when When the membership degree increases linearly from 0 to 1, hour, ;exist hour, ; when When the membership degree decreases linearly from 1 to 0, hour, ;exist hour, ; when At that time, membership degree This indicates that the tongue color value is no longer light red.

[0057] exist Reaching peak This forms a symmetrical triangle membership function.

[0058] M323. Fuzzy Reasoning Rules: Based on traditional Chinese medicine theory and clinical experience, condition-conclusion type fuzzy reasoning rules are constructed to realize the logical mapping between fuzzy variable combinations and health management conclusions. Example rules include: Rule 1: If tongue color is pale and pulse is deep and slow, then the constitution is Yang deficiency. Rule 2: IF constitution IS phlegm-dampness constitution AND season IS long summer THEN intervention intensity IS increased; (In the rule, "IS" means "belongs to". The membership degree is used to determine whether the condition is true and then output the corresponding conclusion.) M324 Defuzzified Output: The centroid method is used to convert the fuzzy conclusions obtained from fuzzy inference into precise numerical parameters, ensuring that the output results can be directly used for calculations in subsequent modules (such as the strategy generation module). Specifically, the membership function curve of the fuzzy output is calculated, and the abscissa value corresponding to its geometric centroid is the precise value after defuzzification.

[0059] M33, Closed-Loop Learning and Optimization Unit: Used to build an adaptive learning mechanism based on Bayesian networks, combining historical health data and real-time user feedback to optimize system model parameters and personalized health intervention strategies, supporting closed-loop operation throughout the entire health management cycle. It also supports multi-node collaborative model optimization based on local data. Specifically, it includes: M331. Bayesian Network Modeling of Health Status Evolution: Construct a dynamic Bayesian network (including node associations in the time dimension), clarify the node types and relationships of key factors in the entire health management process, and form a visualized network structure of "factor-outcome". Node types include: Environmental factors: season, phenological period, regional climate (external variables affecting health status); Individual status nodes: physical condition, symptoms, and physiological indicators (core variables reflecting the user's health status); Intervention measures: Food and medicine homology program, exercise program (intervention variables output by the system); Effect feedback points: physiological improvement, subjective feelings (user feedback variables on the intervention plan).

[0060] M332, Conditional Probability Relationship Learning: Based on historical health data accumulated by the system, this method uses statistical learning algorithms (such as maximum likelihood estimation) to learn the conditional probability distribution between network nodes, quantifying the association strength between cause nodes and result nodes. Example probabilities include: P(Physical Change | Current Physical Condition, Intervention Plan, Spatiotemporal Environment): The probability that physical condition will change under the conditions of "current physical condition, implementation of a certain intervention plan, and specific spatiotemporal environment"; P (Symptom Improvement | Changes in Physical Condition, Intervention Compliance): The probability of symptom improvement under the condition of "changes in physical condition and user compliance with the intervention plan".

[0061] M333, Bayesian Inference Mechanism: Based on the conditional probability distribution of Bayesian networks, it implements three core inference modes to provide quantitative basis for health management decisions: Diagnostic reasoning: P(Constitution|Symptoms, Tongue Appearance, Pulse Appearance) - Given a user's symptoms, tongue appearance, and pulse appearance, the probability of inferring their constitution type; Predictive reasoning: P(health risk|current state, environmental change) — Given the user's current health status and future environmental changes, predict the probability of the user facing health risks; Decision reasoning: P(intervention effect|candidate solution, individual characteristics) — Given candidate intervention solutions and individual user characteristics (such as age and gender), predict the probability of the intervention effect of the solution.

[0062] M334. Online Parameter Update: A Bayesian update rule is adopted, combined with newly collected real-time observational data (such as user feedback on physiological improvements and new physical fitness test data) to dynamically adjust model parameters, ensuring the model adapts to the dynamic changes in the user's health status. The update formula is:

[0063] in, For model parameters, For new observational data, Let P(D|θ) be the prior probability of the parameter, and P(D|θ) be the likelihood probability. The posterior probability of the parameter.

[0064] This module, based on DS evidence theory, resolves conflicts in multi-source data, fuses uncertain information, and assesses confidence in health status. The mapping layer, relying on a fuzzy logic system, completes the digital mapping of TCM concepts, fuzzification of linguistic variables, and preliminary rule-based reasoning. The learning layer uses a Bayesian network to conduct probabilistic relationship modeling, personalized solution optimization, and long-term trend prediction and early warning. These three algorithms are not simply connected in series but deeply synergistic. DS evidence theory provides reliable input evidence for fuzzy logic, fuzzy logic provides structured feature representations for the Bayesian network, and the Bayesian network provides continuous learning and optimization capabilities for the entire system.

[0065] M4, Strategy Generation Module, see [link / reference] Figure 6 This is configured to integrate the personalized environmental spatiotemporal representation with the health status representation and output a personalized health intervention strategy. The user-side application interface diagram exemplifies the human-computer interaction interface through which the system pushes personalized solutions to the user. This module addresses the disconnect between human and policy in existing technologies by deeply integrating dynamic spatiotemporal context, individual constitution, and multi-dimensional information such as user gender and age. Based on the core theory of Traditional Chinese Medicine that "women's constitution is rooted in the liver and blood, while men's constitution is rooted in the kidneys and essence," this module generates highly personalized treatment plans for users of different genders with the same constitution and under the same spatiotemporal background, while maintaining consistent core treatment principles (treatment rules). These plans feature highly personalized details in execution, target points, and intensity of treatment.

[0066] Within the spatiotemporal framework and climate constraints set by a dynamic spatiotemporal context, precise strategy optimization is performed based on individual health status. Through deep fusion of multi-dimensional information and access to a structured knowledge base, highly personalized health intervention strategies (medicine and food homology solutions, exercise and lifestyle recommendations) are output, and the system possesses the capability to extend to diagnostic and treatment decision support. Specifically, this includes: M41, Intelligent Health Preservation Knowledge Base and Compatibility Engine Unit: Used for a structured intelligent health preservation knowledge base of medicinal and edible homologous materials, which is built upon 106 kinds of medicinal and edible homologous materials promulgated by the state. Each material is labeled with its digital attributes: Properties and Channels Entered (Cold, Hot, Warm, Cool; Sour, Bitter, Sweet, Spicy, Salty; Channels Entered); Core functions (such as strengthening the spleen, removing dampness, and warming yang); Spatiotemporal and individual adaptation weights (the strength of the correlation with solar terms, phenology, physical constitution, and region); Compatibility rules (mutual dependence, mutual assistance, mutual restraint, mutual opposition).

[0067] M42, Food and Medicine Homologous Recommendation Unit: This unit intelligently matches and fine-tunes the combination of food and Chinese medicine from the knowledge base based on the user's current constitution, degree of imbalance, solar term, local climate characteristics, gender, and age. For example, for users with Yang deficiency, teenagers are advised to focus on warming Yang to promote growth, middle-aged people are advised to focus on warming Yang to resist aging and prevent disease, and the elderly are advised to focus on warming Yang to strengthen the body and prevent external pathogens. Warming Yang foods are recommended for both men and women with Yang deficiency, but women are advised to add blood-nourishing products (such as angelica), and men are advised to add essence-benefiting products (such as wolfberry).

[0068] Based on the user's current constitution, degree of imbalance, current solar term (after adaptive adjustment), local climate characteristics, user gender, and age, the system intelligently matches and fine-tunes the combination of food and traditional Chinese medicine from the knowledge base. For example, for users with Yang deficiency, teenagers focus on warming Yang to aid growth, middle-aged people focus on warming Yang to resist aging and prevent disease, and the elderly focus on warming Yang to strengthen the body and prevent external pathogens. Warming Yang foods are recommended for both men and women with Yang deficiency, but women are advised to add blood-nourishing ingredients (such as angelica), while men are advised to add essence-boosting ingredients (such as goji berries).

[0069] Based on the health status representation output by the health status processing module, determine the core conditioning principles (such as strengthening the spleen and removing dampness, warming yang and tonifying qi, etc.) according to the main constitutions of the user (such as the nine traditional Chinese medicine constitutions like phlegm-dampness constitution, qi-deficiency constitution, etc.) and the constitution confidence level; introduce the personalized environmental time-space representation generated by the environmental time-space processing module, and convert the calibrated solar terms, phenological periods (one of the seventy-two pentads), and regional climate characteristics (such as humidity, dryness, low temperature, etc.) into scheme constraint conditions (such as "relaxing tendons" is required during the second pentad of the Tomb-Sweeping Festival, and "strengthening dampness elimination" is required in humid regions), so as to ensure that the strategy is synchronized with the actual natural rhythm of the user; Specifically, according to the user's current constitution, degree of deviation, and dynamic time-space context, intelligently match the combination of ingredients and traditional Chinese medicines from the knowledge base to generate daily dietary suggestions and health-preserving tea formulas; following the collaborative compatibility logic of "sovereign, minister, assistant, and courier", the decision-making process is as follows: The main effective element (sovereign): Target the core health contradictions of the user.

[0070] The auxiliary effective element (minister): Enhance the main effect or take into account the core concurrent syndromes.

[0071] The harmonizing element (assistant): Harmonize the properties of various elements, or target the secondary contradictions.

[0072] The guiding element (courier): Guide the direction of action, or take into account the implementation of the plan.

[0073] This unit intelligently matches and fine-tunes the combination of ingredients and traditional Chinese medicines from the knowledge base according to the user's current constitution, degree of deviation, solar terms, and local climate characteristics. Its core decision-making process follows the principle of "prioritize the main symptoms, and the number of medicines should not exceed five": Establish the sovereign medicine (1-2 flavors): First, the system locks in 1-2 flavors of "sovereign medicine" (such as "Poria cocos" for strengthening the spleen and removing dampness) targeting the core contradictions according to the user's core constitution (such as "phlegm-dampness constitution").

[0074] Match the ministerial and assistant medicines (1-3 flavors): Subsequently, the system matches the "ministerial medicine" to enhance the main effect (such as "Chinese yam" for strengthening the spleen and consolidating the foundation) according to the concurrent syndromes and time-space characteristics, or matches the "assistant and courier medicines" to take into account the secondary contradictions, harmonize the properties of medicines, and guide the action (such as "Coix seed" for relaxing tendons and responding to the phenological period, and "Adzuki bean" for promoting diuresis and removing dampness to respond to the region).

[0075] The five-flavor constraint and optimization algorithm: The system has a built-in five-flavor constraint device. When the candidate combination exceeds five flavors, the optimization algorithm will be activated and screened according to the following priorities: Priority one: The correlation strength with the core conditioning principle (treatment principle).

[0076] Priority two: The adaptation weight with the current solar terms and phenology.

[0077] Priority three: The effectiveness of the user's individual preferences and past feedback.

[0078] Priority four: the commonness and availability of ingredients.

[0079] The final output solution is a core combination of medicinal and food ingredients (no more than five ingredients), and can provide an explanation of the "role" (principal, assistant, adjuvant, guide) of each ingredient in this solution to help users understand the logic of the prescription.

[0080] Among them, the "no more than five herbs" principle is the best practice mode recommended by the system. In rare cases, such as when a user's constitution is highly complex and has multiple obvious biases, and the system algorithm evaluates that the benefit of adding a single herb far outweighs the simplified solution, the system can generate a "core solution (no more than five herbs) + optional auxiliary items" mode, and provide special explanations for any excess to ensure that the user can clearly understand.

[0081] M43, Food and Medicine Homologous Cooking Unit: Used to match appropriate cooking methods according to the schemes provided by the Food and Medicine Homologous Recommendation Unit, and to match cooking preparation schemes from the intelligent health knowledge base that are consistent with the user's regional food pairing habits, flavor and taste and daily eating habits. Based on the suggestions provided by the food and medicine homology recommendation unit, cooking and preparation schemes that match the local compatibility, flavor, taste and dietary habits are intelligently matched from the knowledge base.

[0082] M44, Exercise and Rest Suggestion Unit: This unit is used to develop an exercise and rest plan that includes suitable daily routines, exercise schedules and intensity, taking into account the sunrise and sunset times, temperature and humidity changes in the user's location, and personal lifestyle habits. Based on local sunrise and sunset times, temperature and humidity changes, and users' personal lifestyle habits, we can develop suitable daily routines, exercise plans (such as guided exercises and Baduanjin), and intensity.

[0083] M45, Natural Language Generation and Interaction Unit: Used to integrate generative artificial intelligence technology, receive structured data output by the solution generation engine, and transform it into health guidance content.

[0084] It receives structured data from the solution generation engine, including but not limited to combinations of food and medicine with similar properties, exercise suggestions, and conditioning principles; and based on a pre-trained large language model, transforms the structured data into fluent, natural, and encouraging personalized health guidance text. Its core functions include, but are not limited to, the following three categories: M451, Solution Description Generation: Structured solution information such as "15g Poria cocos, 10g Dioscorea opposita, 20g Coix lacryma-jobi" can be transformed into natural language descriptions such as "Dear user, today we recommend a spleen-strengthening and dampness-removing 'Poria cocos, Dioscorea opposita and Coix lacryma-jobi porridge' for you. Coix lacryma-jobi can also help soothe muscles and tendons in spring, which is very suitable for your current condition," making the solution easier for users to understand; M452, Interactive Q&A: Interacts with users through a dialogue interface, and can respond to various free-form questions from users about health intervention strategies. The questions include, but are not limited to, the principles of the plan, the methods of replacing ingredients, and the details of implementation. M453. Multimedia Content Generation: Based on the core elements of the health intervention strategy, automatically generate or recommend relevant multimedia health guidance information, including but not limited to instructional video scripts, illustrated cooking steps, and guided action diagrams.

[0085] By applying generative artificial intelligence technology, this module can transform the system's algorithm output into warm and easy-to-understand daily action guidelines, helping users to more easily receive and implement health intervention strategies.

[0086] M46, Gender-Differentiated Generation Unit: Based on the traditional Chinese medicine theory that women's innate constitution is rooted in the liver and blood, while men's is rooted in the kidneys and essence, this unit fine-tunes the plans. The female plan emphasizes nourishing blood and soothing the liver, while the male plan emphasizes benefiting essence and strengthening the kidneys. In the personalized plan generation, user gender is innovatively introduced as a key variable. The system fine-tunes the plans based on the following differentiation strategy model constructed from traditional Chinese medicine theory: M461, Female User Strategy: During solution generation, the system will automatically strengthen the dimensions of "nourishing blood" and "soothing the liver." In the recommendation of food and medicine from the same source, priority will be given to combining ingredients / medicinal herbs that have both conditioning effects (such as warming yang and removing dampness) and effects of nourishing blood and promoting blood circulation or soothing the liver and relieving depression; in the exercise and rest suggestions, priority will be given to guiding techniques and exercise intensity that help regulate qi and relax muscles and bones.

[0087] M462, Strategy for Male Users: During the plan generation process, the system will automatically strengthen the dimensions of "benefiting essence" and "strengthening the kidneys." In the recommendations of food and medicine from the same source, priority will be given to combining ingredients / medicinal herbs that have both conditioning effects and kidney-benefiting and essence-replenishing effects; in the exercise and lifestyle suggestions, exercises to strengthen the waist and leg functions can be appropriately increased, and reminders on abstinence and preservation of essence will be incorporated into daily life.

[0088] This strategy ensures that the health and wellness program adheres to core treatment principles while precisely matching the different physiological foundations of men and women, achieving a deeper level of "personalized care".

[0089] M47, Age-Stage Adaptation Unit: This unit matches the user's age to childhood, adolescence, middle age, or old age, prioritizing and fine-tuning the basic treatment plan according to the physiological characteristics and core principles of each stage. User age is used as a key variable to achieve "age-specific care." The system divides the human life cycle into several main stages and sets a core focus for each stage: M471. Childhood (0-14 years): Physiological characteristics include "delicate internal organs, underdeveloped physical form and energy, vigorous vitality, and rapid development." The core of treatment is "strengthening the spleen and kidneys, and promoting growth and development," avoiding strong tonics. The plan emphasizes light, nutritious, and easily absorbed foods, and recommends fun activities that are beneficial to both body and mind.

[0090] M472, Young Adulthood (15-45 years old): Physiological characteristics include "strong organ function and abundant Qi and blood." However, this stage is characterized by high social pressure, which easily depletes essence and blood. The core of conditioning is "soothing the liver and regulating Qi, harmonizing emotions, and strengthening acquired constitution (spleen and stomach)." The treatment plan needs to be tailored to the fast-paced lifestyle, providing convenient and efficient conditioning suggestions.

[0091] M473, Middle Age (46-65 years old): The physiological characteristics are "the decline of Yin and Yang Qi and Blood, and the decline of organ function." The *Neijing* states, "At forty, Yin Qi is halved, and daily life declines." The core of treatment is "tonifying the liver and kidneys, harmonizing Yin and Yang, and delaying aging." The plan needs to place greater emphasis on comprehensive conditioning to prevent the occurrence and development of chronic diseases.

[0092] M474, Old Age (65 years and older): The physiological characteristics are "deficiency of all five internal organs, insufficiency of qi and blood, and damage to both yin and yang." The core of treatment is "tonifying qi and blood, strengthening the foundation, and maintaining function." The plan follows the principle of gentle and gradual tonification, avoiding harm to the body's vital energy. All suggestions prioritize safety, and exercise should be gentle.

[0093] This unit automatically matches the user's age to the above-mentioned stage and, based on the core conditioning principles of that stage, prioritizes and fine-tunes the basic plan generated by factors such as physical condition and time and space, ensuring that the plan is synchronized with the user's physiological aging rhythm.

[0094] M48, Closed-Loop Management and Optimization Unit: Regularly revisits users via AI terminals to collect updated physiological indicator data; analyzes the interaction between physical condition evolution trends and meteorological data changes through multi-dimensional algorithms to dynamically optimize conditioning strategies and model parameters. The model optimization process of the closed-loop management and optimization module can adopt a federated learning paradigm.

[0095] In federated learning, multiple system instances (acting as clients) deployed across different institutions or serving different user groups train models locally using their user data (e.g., fine-tuning phenological delay coefficients k and m in a dynamic spatiotemporal calibration model, or network parameters in a physical fitness identification model). Only the updated model parameters are encrypted and uploaded to a central server. The central server aggregates model updates from multiple clients, generates a globally improved model, and then distributes it to each client.

[0096] This process does not require centralized sharing of any original user physiological data or sensitive information such as geographical location, fundamentally ensuring user data privacy and security. At the same time, it can continuously improve the accuracy and generalization ability of the system's core model by utilizing widely distributed data, achieving secure collaborative evolution of the model while the data remains stationary.

[0097] M5, Terminal Visualization Module: Configured to establish data interaction with the data acquisition module, health status processing module, environmental spatiotemporal processing module, and strategy generation module. This enables visualized guidance for associated data acquisition, visualized display of associated strategy generation, and feedback data reception and transmission functions supporting closed-loop optimization. Specifically, it includes: M51, Acquisition Guidance Unit: Used to receive the acquisition request from the data acquisition module, provide acquisition process guidance when acquiring multimodal physiological data, and display the acquisition progress and data verification status in real time; M52, Strategy Display Unit: Used to receive the personalized health intervention strategy output by the strategy generation module and the personalized environmental spatiotemporal representation generated by the environmental spatiotemporal processing module, present the health intervention strategy on the user's smart device client, and simultaneously display the calibrated spatiotemporal and phenological related information. M53, Feedback Interaction Unit: Used to receive user feedback data on the effectiveness of health intervention strategies and transmit the feedback data to the system to support closed-loop optimization of the health status processing module and the strategy generation module; M54, Data Retrospective Unit, is used to receive the historical health status representation of the health status processing module and the historical personalized environmental spatiotemporal representation of the environmental spatiotemporal processing module, and to display the historical health status, historical personalized environmental spatiotemporal representation and execution records of past health intervention strategies on the user's smart device client. M55, Natural Language Interpretation Subunit, is used to integrate into the strategy display unit to convert structured health intervention strategies into natural language interpretation content and present it on the user's smart device client; M56, Solution Interaction Subunit, is used to integrate with the strategy display unit, to mark the compatibility roles of ingredients when presenting solutions related to medicine and food in health intervention strategies, and to display the efficacy descriptions of each compatibility role to the user through interactive operations; M57, Exercise Assistance Display Subunit, is integrated into the strategy display unit to display visual content related to exercise movements when presenting exercise-related solutions in health intervention strategies, and provides corresponding playback control functions; M58, the retest reminder subunit, is integrated into the feedback interaction unit. It receives the physical fitness retest cycle requirement from the health status processing module and displays a visual reminder and retest process guide on the user's smart device client when the preset time before the next physical fitness retest is approaching.

[0098] See Figure 5 The terminal visualization module's functions are implemented through the user's smart device interface (such as a mobile app). The correspondence between interface elements and sub-units is as follows: The top of the page displays a "Friendly Reminder: Tomorrow at 17:52, we will enter the Autumn Equinox. Beijing is currently experiencing dry weather." The corresponding strategy display unit will simultaneously display calibrated spatiotemporal and phenological information. "Constitution: Lung Qi Deficiency (Tendency: 65%)": Corresponding health status indicator (data from M3 module); "Food and medicine share the same origin: lily porridge, snow pear and lily soup" and "Exercise and rest: teeth tapping and saliva swallowing, breathing guidance": corresponding to the "Health Intervention Strategy Presentation" in M52 (data from module M4); Feedback button: corresponds to the "Feedback Entry" of the M53 feedback interaction unit (used to collect user subjective ratings); See detailed interpretation: Corresponds to the M55 Natural Language Interpretation subunit (generating an explanation of the health preservation principles for "Autumn Dryness Prevention").

[0099] In summary, when generating food-medicine homology solutions, the principles of "principal, assistant, adjuvant, and guide" in traditional Chinese medicine and the essence of "simple yet potent" formulation are followed. Innovatively, it is stipulated that the number of core food-medicine homology substances in each treatment plan should not exceed five. This includes: aligning with the spirit of classic TCM formulas; classic formulas such as "Si Jun Zi Tang," "Si Wu Tang," and "Gui Zhi Tang" are characterized by refined medicinal ingredients and clear efficacy, directly targeting the disease site through the synergistic effect of a few drugs, representing a digital inheritance of this traditional thought; ensuring a clear core treatment direction; limiting the number of medicinal ingredients allows the system to focus on the most critical constitutional imbalances and temporal and spatial factors, extracting key treatment principles such as strengthening the spleen, dispelling dampness, and warming yang, avoiding the dilemma of overly comprehensive and vague solutions; and significantly improving the safety and feasibility of the solutions, as combinations of five or fewer ingredients greatly reduce the risk of adverse interactions such as "antagonism" or "mutual restraint" between ingredients. Anticipating the risks of interactions, the simplified solution reduces user procurement and cooking costs, greatly improving user compliance. This "no more than five medicines" principle is essentially a quantity constraint mechanism for core intervention elements. By pre-setting a maximum quantity of five, the solution ensures that it focuses on the core contradictions. Furthermore, any adaptive adjustments to this maximum quantity based on the same inventive concept by those skilled in the art are still within the scope of protection of this invention. In addition, the limited combination of five flavors facilitates efficient and accurate knowledge base matching and weight calculation by the algorithm. During the closed-loop optimization process, the actual effect of each ingredient can be analyzed more clearly, providing support for the continuous iteration of the solution.

[0100] Example 2 See Figure 3 , Figure 3 The flowchart of the health management intelligent agent detection method based on dynamic spatiotemporal calibration proposed in this invention includes the following steps: S1. Obtain user health status data; health status data is a classification result based on the analysis of user's multimodal physiological data. The health status data uses "nine TCM constitutions" (balanced constitution, qi deficiency constitution, yang deficiency constitution, yin deficiency constitution, phlegm-dampness constitution, damp-heat constitution, blood stasis constitution, qi stagnation constitution, and special constitution) as the classification criteria, obtained through multimodal physiological data collection and analysis; specific steps may include: S11. Multimodal physiological data collection: Users complete data collection through AI intelligent diagnostic terminals. The collected data includes tongue appearance (such as tongue coating thickness and tongue shape, with typical characteristics such as a swollen tongue and greasy coating), pulse appearance (such as pulse rate and pulse shape, with typical characteristics such as a weak and thready pulse), facial appearance (such as skin color and skin texture), and voice appearance (stable voice signal). After collection, the raw physiological data is uploaded to the cloud server for subsequent analysis.

[0101] S12. Health status classification analysis: The cloud server converts the received raw data into standardized feature vectors, such as normalizing tongue images into pixel matrices and pulse signals into spectrum diagrams. Deep learning model determination: The standardized feature vector is input into the pre-trained deep learning model (which is trained based on a large amount of physiological data labeled with constitution by professional TCM doctors). The model outputs a 9-dimensional probability vector, with each 1-dimensional vector corresponding to a score of a TCM constitution. Fusion of DS Evidence Theory: Constructing an Identification Framework Based on Nine Mutually Exclusive and Complete Sets of Traditional Chinese Medicine Constitutions The study treats tongue appearance, pulse appearance, facial appearance, and voice appearance as independent evidence and assigns basic probabilities to each evidence. The Dempster combination rule is used to calculate the fusion confidence of multiple evidences, and finally the constitution with the highest confidence is selected as the user's health status data.

[0102] S2. Acquire the user's environmental spatiotemporal data, which includes at least geographical location; the altitude data within the environmental spatiotemporal data is collected in real time by a barometric altimeter sensor integrated into the user terminal. Specific steps may include: S21. Geographic location acquisition: The accurate latitude is obtained through the BDS / GPS module of the user's smartphone. The positioning accuracy meets the requirements of phenological calibration. For example, the positioning result for a user in Beijing is 39.93°N, 116.40°E. Altitude is collected in real time by a barometric altimeter integrated into an AI-powered diagnostic terminal or wearable device. If the sensor malfunctions, the altitude data is matched using a digital elevation model (DEM) database. For example, the altitude for a user in Shanghai is 4m, while the altitude for a user in Chengdu is 500m.

[0103] S22. Environmental Data Supplement: Real-time meteorological data (temperature, humidity, air pressure, etc.) and historical meteorological data for the same period in the past 3 years are obtained from the meteorological data API interface for the user's location. Phenological data are obtained from the phenological dataset of the Long-Term Ecological Observation Network (CERN), such as "peach blossoms bloom around Qingming Festival in Beijing" and "peach blossoms bloom around the Spring Equinox in Zhengzhou".

[0104] S3. Based on the aforementioned environmental spatiotemporal data, the personalized environmental spatiotemporal representation is a dynamically calibrated time-series model correlated with the user's actual natural rhythms; specific steps may include: S31. Reference location and core parameter settings: Based on Zhengzhou City, Henan Province, the representative of the Central Plains region where the Twenty-Four Solar Terms originated, its geographical parameters are 34.76° north latitude and about 100m altitude. Key coefficients: latitude-phenological lag coefficient (range 0.6-1.2 days / degree, typical value 0.8 days / degree) and altitude-phenological lag coefficient m (range 0.5-1.5 days / 100 meters, typical value 1.0 day / 100 meters). Both coefficients were obtained by fitting historical phenological data with meteorological data.

[0105] S32. Latitude offset: Calculated based on the latitude difference between the user's location and Zhengzhou. For example, if the latitude difference between a user in Shanghai (31.23°N) and Zhengzhou is 3.53°, then the phenological events in Shanghai will be about 2.8 days earlier than those in Zhengzhou. Altitude offset: Calculated based on the altitude difference between the user's location and Zhengzhou. For example, if the altitude difference between a user in Chengdu (500m) and Zhengzhou is 400m, then the phenology in Chengdu will be about 4 days later than in Zhengzhou. Base offset: The base offset is obtained by adding the latitude offset and the altitude offset.

[0106] S33. Real-time meteorological dynamic fine-tuning: Compare the real-time meteorological data of the user's location with the historical data for the same period, and fine-tune the basic offset: If the average temperature for 3 consecutive days is 3°C lower than the historical average for the same period, the basic offset is increased by 1 day (indicating a delay in phenology); if the humidity is 15% higher than the historical average for the same period, the basic offset is decreased by 0.5 days (indicating an advance in phenology), and the upper limit of the fine-tuning range is ±2 days; add the basic offset and the meteorological fine-tuning amount to obtain the final offset.

[0107] S34. Generation of the time series model of phenology school standard: Based on the final offset, the solar term time of the user's location is calibrated. For example, the autumnal equinox in Zhengzhou is September 23, and the final offset of the Chengdu user is 4 days, so the "local autumnal equinox" in Chengdu is September 27. Through function (Combining geographical location and real-time meteorological data) Determine the current phenological period, such as "the second phase of the Autumn Equinox, when thunder begins to subside" and "the second phase of the Qingming Festival, when field mice turn into quails"; Integrate and calibrate the solar terms, phenological periods, and regional climate characteristics (such as "dry autumnal equinox" in Beijing and "humid awakening of insects" in Chengdu) to generate personalized spatiotemporal environmental representations, such as "September 27, 2025 (autumn equinox calibrated in Chengdu) - second pentad of autumnal equinox - altitude 500m - humidity 65%".

[0108] S4. Based on the fusion analysis of health status data and personalized environmental spatiotemporal representation, generate personalized health intervention strategies. The generation of personalized environmental spatiotemporal representation is based on the latitude difference and / or altitude difference between the stated geographical location and the reference geographical location; the personalized environmental spatiotemporal representation is a time series model dynamically calibrated based on phenological theory. It also includes dynamically fine-tuning the base spatiotemporal offset calculated from the latitude difference and / or altitude difference based on real-time meteorological data of the user's location. When generating the personalized health intervention strategy, at least one of the user's gender information and age stage information is further integrated, and based on this, the intervention element type, target point, or execution intensity in the strategy are configured differently. Specific steps include: S41. Data Acquisition for Personalized Health Intervention Strategies: Acquire the three core data types required to generate personalized health intervention strategies, specifically including: Collect user health status related data: including dominant constitution type (one of nine TCM constitutions such as balanced constitution, qi deficiency constitution, yang deficiency constitution, etc.), degree of constitution imbalance and constitution confidence level. This data is obtained by analyzing multimodal physiological data such as user tongue appearance, pulse appearance, facial appearance, and voice appearance. Collect spatiotemporal data of the user's environment: including the solar terms after dynamic calibration, the current phenological period (one of the seventy-two pentads), the climate characteristics of the user's region (such as humidity, dryness, low temperature), real-time temperature, humidity, and sunrise and sunset times; Collect basic user information, including user gender (male / female) and specific age. The information is obtained from user registration or authorized terminal devices.

[0109] S42. Generate basic food and medicine homology combinations: Following the principles of "prioritizing primary symptoms, combining principal, assistant, adjuvant, and guide herbs, and limiting the number of herbs to five," match and fine-tune the combination of food ingredients and traditional Chinese medicine, specifically including: S421. Determine the core conditioning principle: Based on the user's dominant constitution type and degree of imbalance, identify the conditioning direction corresponding to the core health contradiction. For example, users with phlegm-dampness constitution correspond to the principle of "strengthening the spleen and removing dampness", while users with yang deficiency constitution correspond to the principle of "warming yang and replenishing qi". S422. Transform environmental spatiotemporal constraints: Transform the calibrated solar terms, current phenological period, and regional climate characteristics into constraints for food and medicine combinations. For example, during the second phase of Qingming, ingredients with "muscle-relieving" effects should be added, and in humid regions, the combination of "dampness-removing" ingredients should be strengthened. S423. Matching the principal medicine (1-2 herbs): Based on the core conditioning principle, select 1-2 ingredients / medicinal herbs from the intelligent health knowledge base that directly address the core health contradictions as the principal medicine. For example, users with phlegm-dampness constitution are matched with "Poria cocos" (whose core function is to strengthen the spleen and eliminate dampness), and users with qi deficiency constitution are matched with "Astragalus membranaceus" (whose core function is to replenish qi and strengthen the exterior). S424, Matching Assistant and Adjuvant Herbs (1-3 herbs): Based on the user's concurrent symptoms and environmental and temporal constraints, match assistant, adjuvant, and guiding herbs respectively: Assistant herbs: Select ingredients / medicinal materials that can enhance the core efficacy of the principal herb, such as pairing "Poria cocos" (principal herb) with "Dioscorea opposita" (assistant herb, which helps to strengthen the spleen and consolidate the foundation, and enhance the effect of removing dampness); Adjuvant herbs: Select ingredients / medicinal materials that can harmonize the medicinal properties and adapt to the characteristics of the environment and time. For example, "coix seed" is used to meet the need for "relaxing muscles" during the second phase of Qingming, and "red adzuki bean" is used to meet the needs of humid areas (to assist in mild diuresis and dampness removal). S425. Perform Five-Flavor Constraint and Optimization Screening: If the initially matched medicinal and food combinations exceed five flavors, activate the optimization algorithm to screen according to the following priority (from high to low): The order of importance is: the degree of relevance to the core conditioning principles, the weight of the adaptation to the current solar term and phenology, the validity of users' past feedback on similar ingredients, and the order of personal dietary preferences and the commonness and availability of ingredients. S426. Special Case Handling: Only when a user's constitution is highly complex (combining 3 or more obvious biases), and the algorithm evaluates that the conditioning benefit of adding 1 medicinal herb is significantly greater than the benefit of the simplified solution, will a pattern combining a core medicinal food combination with optional auxiliary ingredients be generated, and the role, dosage, and applicable scenarios of the optional auxiliary ingredients be specifically explained.

[0110] S43. Matching and Adapting Cooking Methods: Based on the basic food-medicine homology combination generated in step S42, and combined with the user's regional habits and dietary preferences, determine the specific cooking plan: S431. Extract user regional characteristics: Based on the climate characteristics of the user's region (such as the south, the north, the plateau, etc.), prioritize matching cooking methods that conform to local traditional pairing habits. For example, in humid southern regions, prioritize "soup" and "stew"; in dry northern regions, prioritize "porridge" and "steaming". S432. Adapt to ingredient characteristics and user habits: Adjust cooking methods according to the texture of ingredients (such as root vegetables, leafy vegetables, and dried goods). For example, root vegetables such as "Poria cocos" and "yam" are suitable for "stewing", while flower and herb ingredients such as "rose" and "chrysanthemum" are suitable for "stewing". At the same time, adjust cooking details based on user feedback on dietary preferences (such as whether they like oily food or avoid spicy food). For example, for users who do not like oily food, change "deep-frying" to "stir-frying" or "steaming". S433. Output specific cooking steps: Clearly indicate the amount of each ingredient, the pre-treatment method (such as soaking, slicing), the cooking temperature (such as boiling over high heat, simmering over low heat) and the cooking time. For example, "Poria and Yam Porridge" should indicate "Poria 15g (soak for 30 minutes in advance), yam 20g (peeled and sliced), cook with rice, put in cold water, bring to a boil over high heat, then simmer over low heat for 30 minutes until the porridge thickens".

[0111] S44. Develop a basic exercise and lifestyle plan: Based on the user's environmental and temporal data and personal lifestyle habits, develop a daily routine and exercise plan, specifically including: S441. Determine your daily routine: Adjust your schedule according to the sunrise and sunset times in your location. For example, if the sunrise is early in summer (e.g., 5:30 am), it is recommended to get up at 6:00 am; if the sunset is early in winter (e.g., 5:00 pm), it is recommended to go to bed before 9:30 pm to ensure that your daily routine is in line with the natural rhythm. S442. Develop an exercise plan: Choose the type of exercise: Prioritize guided exercises that match the current season, such as "Baduanjin" (for liver regulation) in spring and "breathing exercises" (for lung health) in autumn; Determine the intensity and duration of exercise: Adjust according to real-time temperature and humidity. For example, when the temperature is >30℃ and the humidity is >70%, it is recommended to exercise for ≤30 minutes with an intensity of "moderate to low" (such as slow walking and gentle stretching); when the temperature is 15-25℃ and the humidity is 40-60%, the intensity can be increased to "moderate" (such as brisk walking + the complete set of Baduanjin), and the duration can be controlled at 40-60 minutes. S443. Adapt to personal lifestyle habits: Adjust exercise time based on user feedback on daily routines (such as whether they are office workers or have a habit of morning / evening exercise). For example, for office workers, schedule exercise time for "15 minutes after waking up in the morning" or "30 minutes after getting off work in the evening" to avoid conflicts with work and commuting time.

[0112] S45. Gender-based fine-tuning: Based on the traditional Chinese medicine theory that "women's constitution is rooted in the liver and blood, while men's constitution is rooted in the kidneys and essence," the basic scheme generated in steps S42-S44 is fine-tuned, specifically including: S451, Minor Adjustments to the Female User Plan Food and medicine combinations: These combinations enhance the effects of "nourishing blood" and "soothing the liver" in the basic formula. For example, for women with Yang deficiency, in addition to "longan (warming Yang)," add "5g of angelica" (to help nourish blood) or "3g of rose" (to help soothe the liver and relieve depression); for women with Qi deficiency, in addition to "astragalus (tonifying Qi)," add "5 red dates" (to help nourish blood and calm the mind). Exercise and rest plan: Prioritize exercises that can regulate Qi and relax muscles and bones, such as "yoga stretching" (focusing on stretching the rib area and helping to soothe the liver) and "Tai Chi Cloud Hands" (slow-paced Qi regulation); at the same time, add "emotion regulation reminders", such as "10 minutes of meditation before bed every night to help calm emotions and soothe the liver and regulate Qi".

[0113] S452, Minor Adjustments to the Male User Plan Food and medicine homology combination: The "nourishing essence" and "strengthening the kidney" effects are enhanced in the basic combination. For example, for men with yang deficiency, "10g of wolfberry" (to help nourish the kidney and replenish essence) or "15g of gorgon fruit" (to help strengthen the kidney and astringe essence) are added to the "longan (warming yang)" combination; for men with qi deficiency, "5g of eucommia bark" (to help strengthen the kidney and waist) are added to the "astragalus (tonifying qi)" combination. Exercise and lifestyle plan: Appropriately increase exercises to strengthen the waist and leg functions, such as focusing on strengthening the "hands grasping the feet to strengthen the kidneys and waist" movement in the "Eight Pieces of Brocade" exercise; add suggestions for "conserving essence through sexual activity" to the daily routine reminders, such as "avoid staying up late (going to sleep after 11:00 PM) to reduce the depletion of kidney essence."

[0114] S46. Age-based fine-tuning: Match the user's age to the corresponding life cycle stage, and further optimize the plan after fine-tuning in step S45 according to the physiological characteristics and core treatment principles of each stage. Specifically, this includes: S461. Childhood (0-14 years old) Food and medicine combinations: Focus on "strengthening the spleen and kidneys, light and easily absorbed", avoid using strong tonics. For example, for children with phlegm and dampness constitution, "Poria cocos (chief herb)" is combined with "5g of hawthorn" (to help stimulate appetite and digestion) instead of "Atractylodes lancea" (which is too dry and not suitable for children) which is commonly used in adults. Exercise and rest schedule: We recommend fun activities such as "children's version of Tai Chi" and "outdoor jump rope", with exercise time controlled at 15-20 minutes to avoid boring guided exercises; add "ensure 8-10 hours of sleep every day to support growth and development" to the daily routine reminders.

[0115] S462, Young adulthood (15-45 years old) Food and medicine combination: Focusing on "soothing the liver and regulating qi, and protecting the spleen and stomach", it is suitable for fast-paced life, such as providing "convenient tea" (such as "rose and goji berry tea", which can be brewed) to replace complicated soup-making methods; Exercise and lifestyle plan: We recommend efficient and short-duration exercises, such as "15 minutes of Baduanjin in the morning" and "20 minutes of brisk walking in the evening", which are suitable for busy work scenarios; add "avoid prolonged sitting (get up and move around for 5 minutes every hour) to the daily routine reminders to prevent stagnation of Qi".

[0116] S463, Middle Age (46-65 years old) Food and medicine combination: focuses on "tonifying the liver and kidneys and harmonizing yin and yang". For example, for middle-aged users with qi deficiency, "goji berries 10g" (tonifying the kidneys) and "lily bulbs 10g" (nourishing yin) are added to "Astragalus membranaceus (chief herb)" to prevent yin and yang imbalance. Exercise and rest plan: Based on the principle of "gentle and injury prevention", for example, change "brisk walking" to "slow walking + joint rotation exercises" to avoid joint damage caused by strenuous exercise; add "20-30 minutes of afternoon nap every day to help restore energy" to the daily routine reminders.

[0117] S464, Old Age (65 years and older) Food and medicine combinations: These focus on "tonifying qi and blood, nourishing the original qi and strengthening the foundation", with mild medicinal properties. For example, elderly patients with yang deficiency can use "20g of yam + 5 red dates" (mildly tonifying yang) instead of "cinnamon" (which has a stronger medicinal property and can easily damage the body's vital energy). Exercise and rest schedule: With "relaxation and safety" as the core, we recommend "seated Baduanjin" and "15-minute afternoon walk (sunbathing, to help with calcium supplementation)" to avoid the risk of falls; "warm-up reminders" should be added before exercise (such as moving wrists and ankles for 5 minutes); "soaking feet in warm water for 10 minutes before bed to help improve sleep" should be added to the daily routine reminders.

[0118] S47. Output a complete personalized health intervention strategy: When outputting a complete personalized health intervention strategy, it is necessary to integrate the optimized content from the previous steps to form a structured plan, specifically including: The plan is divided into two main modules: "Medicine and Food from the Same Source (including ingredient combinations and cooking steps)" and "Exercise and Rest (including daily routines and exercise plans)". Each part is clearly labeled with its "suitability basis" (e.g., "Angelica sinensis: nourishes blood in women, suitable for women aged 32 and above"). The roles of the medicinal and food combinations are also labeled (e.g., "Poria cocos (principal: strengthens the spleen and removes dampness), Dioscorea opposita (assistant: strengthens the spleen), Coix lacryma-jobi (assistant: relaxes muscles and responds to the Qingming Festival)"). This helps users understand the logic of the prescription. Generative artificial intelligence technology can also be used to output natural language interpretation text, interactive Q&A content (e.g., "Why is this porridge recommended?" "Can the exercise intensity be adjusted?"), or multimedia guidance information (e.g., cooking videos, exercise demonstration animations) to improve users' understanding and ease of execution of the plan.

[0119] S5. Receive feedback data from the user after implementing the health intervention strategy, the feedback data including updated physiological indicator data and / or user subjective experience data; specifically including: S51. Feedback Data Collection: Objective physiological data is collected by retesting multimodal physiological data of users every 7 days through AI intelligent diagnostic terminal, and synchronously uploading sleep duration, heart rate and other data recorded by wearable devices; Subjective experience data is collected through APP questionnaires for "diet palatability (1-5 points)", "post-exercise fatigue (1-5 points)" and "changes in sleep quality".

[0120] S52. Model and Strategy Adjustment: The health status analysis model is adjusted so that if the feedback shows a deviation in the constitution judgment (such as the user's self-report of being afraid of cold, the original model judged it as "damp-heat constitution"), the probability allocation weight of the DS evidence theory is optimized through machine learning (such as increasing the weight of the consultation data). For example, a user in Beijing with lung qi deficiency reported "decreased sleep quality": "Lily calming tea" was added, and "indoor humidification at night (50%-60%)" was supplemented; a user in Guangzhou with damp-heat constitution reported "reduced steps and irritability": "Jiaodiao music (to soothe the liver and regulate qi)" was added, and the medicinal diet was adjusted to "Five-finger peach and Poria soup".

[0121] Based on the feedback data, the analysis model of the health status data and / or the generation strategy of the health intervention strategy are dynamically adjusted through machine learning algorithms; the adjustment frequency is once every 14 days, and real-time adjustment is triggered by extreme weather (such as continuous rainstorms) or sudden changes in physical condition (such as aggravation of qi deficiency after a cold).

[0122] See Figure 7 This method focuses on the specific work steps and system judgment nodes in the four stages of "testing-judgment-adjustment-evaluation," clarifying how the system initiates a new cycle and performs self-optimization based on retest data. It strengthens the disclosure of the technical characteristics of closed-loop optimization, specifically including: Measurement: AI terminals collect multimodal physiological data and store it in the core database (to provide a historical baseline for subsequent analysis). Judgment: The cloud-based fusion analysis module calls the core database and combines user tags (age, gender, physique, climate, altitude) to complete spatiotemporal calibration and physique identification; Adjustment: Based on the judgment results, the action subsystem generates and pushes personalized solutions; Comment: The learning subsystem receives user feedback data, optimizes the analysis model of judgment (such as adjusting the weight of DS evidence) and the strategy for generating adjustment solutions (such as replacing food ingredients that users cannot tolerate), updates the core database, and starts the next closed loop.

[0123] Example 3 This embodiment uses an intelligent health monitoring and management system provided by the present invention, taking the autumnal equinox of 2025 as an example to perform intelligent health monitoring, specifically including: For users in Beijing (North China): The system detected that its latitude is higher than that of Henan, and real-time weather data showed that the air humidity had dropped below 40%. Before the traditional autumnal equinox, the system pushed a "lung-nourishing and dryness-preventing" plan to users with lung qi deficiency: recommending pear and lily bulb soup, the morning teeth-tapping and saliva-swallowing method, and suggesting adding layers of clothing in the morning and evening. The plan was accompanied by Shang-style music.

[0124] Guangzhou users (South China region): The system determined that the local area is still experiencing damp heat and pushed a "strengthening the spleen and resolving dampness" plan to users with damp-heat constitution: reminding them to reduce the intake of greasy foods, recommending Five-Finger Peach soup, and adjusting the guided exercises.

[0125] Closed-loop optimization: A week later, the system discovered through the AI ​​terminal that the sleep quality of Beijing users had declined, so it added suggestions for calming tea and nighttime humidification; for Guangzhou users, due to their reduced steps and mood fluctuations (monitored by facial or vocal image analysis through the AI ​​terminal), the system incorporated Jiaodiao music and dietary therapy suggestions for soothing the liver and regulating Qi into the solution.

[0126] Example 4 This embodiment uses an intelligent health monitoring and management system provided by the present invention, taking Chengdu, Sichuan Province (latitude 30.67°N, altitude 500m) and Shanghai (latitude 31.23°N, altitude 4m) as examples, specifically including: Both cities are located at similar latitudes, so their phenological periods should be almost identical according to the latitude formula. However, Chengdu's altitude is about 500 meters higher than Shanghai's. According to the model, Chengdu's phenological period shift will be greater than Shanghai's by: ΔH = 1.0 × (500 - 4) / 100 ≈ 5 days. This means that during the Spring Equinox, the system's health reminders for Chengdu users will be about 5 days later than those for Shanghai users, which aligns with actual climate conditions.

[0127] Example 5 This embodiment uses an intelligent health monitoring and management system provided by the present invention. Taking four users who all have the "phlegm-dampness constitution," all reside in "Shanghai," and are under the "Awakening of Insects" solar term background after system calibration as an example, the gender-differentiated scheme under the same constitution and time period specifically includes: Child user (6 years old, boy): System analysis: Phlegm-dampness constitution, in the "childhood stage", the core problem is "frequent spleen deficiency".

[0128] Personalized solutions: Food and medicine share the same origin: We recommend "Hawthorn and Job's Tears Porridge," which not only strengthens the spleen and removes dampness, but also adds hawthorn to stimulate appetite and aid digestion. Completely avoid herbs with strong drying properties commonly used by adults, such as Atractylodes lancea.

[0129] Exercise routine: It is recommended to increase physical activity through fun activities such as outdoor games and rope skipping, and avoid monotonous guided exercises.

[0130] Young adult user (35 years old, male, Mr. Li): System analysis: Phlegm-dampness constitution, combined with the physiological characteristic of men that "the kidneys are the foundation of life", the dampness and stickiness easily damage the yang qi.

[0131] Personalized solutions: Food and medicine share the same origin: "Poria and Euryale seed soup" is recommended to strengthen the spleen and remove dampness (treating the root cause of phlegm and dampness). At the same time, "dried tangerine peel" is added to regulate qi and dry dampness. It is also emphasized that the diet should not be too heavy to reduce the burden on the spleen, stomach and kidney qi.

[0132] Exercise and rest: It is recommended to practice "Eight Pieces of Brocade", especially strengthening the movements of "Holding up the sky with both hands to regulate the three jiaos" and "Climbing the feet with both hands to strengthen the kidneys and waist" to regulate water metabolism and strengthen the waist and kidneys.

[0133] This embodiment demonstrates that the present invention can generate precise solutions for users with identical physical constitutions and spatiotemporal backgrounds, based on their gender differences, using different methods from the same source and different approaches from the same root, showcasing the system's unparalleled depth of personalization.

[0134] Young adult user (32 years old, female, Ms. Wang): System analysis: Phlegm-dampness constitution, combined with the physiological characteristic of women that "the liver is the foundation of life", is prone to qi stagnation.

[0135] Personalized solution: Food and medicine from the same source: We recommend "Job's tears and yam porridge" to strengthen the spleen and remove dampness (treating the root cause of phlegm and dampness), while adding a small amount of "rose petals" to make tea to soothe the liver and relieve depression, and prevent qi stagnation from aggravating dampness obstruction.

[0136] Exercise and lifestyle: It is recommended to practice "yoga stretching", focusing on stretching the rib area to soothe the liver qi; and remind people to regulate their emotions by participating in social activities.

[0137] Elderly user (70 years old, female, Ms. Zhang): System analysis: Phlegm-dampness constitution, combined with the characteristics of elderly women "deficiency of all five internal organs and insufficiency of qi and blood".

[0138] Personalized solutions: Food and medicine share the same origin: We recommend "Astragalus and Yam Porridge". On the basis of strengthening the spleen and removing dampness (yam), we use more "Astragalus" to replenish qi and consolidate the exterior, support the body's vital energy, and prevent damage to the already depleted vital energy during the process of expelling pathogens (dampness).

[0139] Exercise and rest schedule: It is strongly recommended to take an afternoon walk to get some sun and practice the "seated Baduanjin" exercise, with absolute safety as the core principle.

[0140] This embodiment comprehensively demonstrates that the present invention can generate multi-dimensional, highly customized, and precise solutions for users with the same physical condition and spatiotemporal background, based on their cross-differences in gender and age, showcasing the unparalleled depth and breadth of the system in achieving "personalized solutions".

[0141] Example 6 This embodiment uses the core of the preset "Lixia" (Start of Summer) solar term health preservation plan as "protecting the heart's yang and avoiding irritability and anger." Taking a user in Beijing with a "yin deficiency constitution" as an example, the system monitors the local area for continuous low temperatures and rainy weather (average temperature 5°C lower than the historical average) a week before Lixia via a real-time meteorological interface, and detects the dynamic response under sudden climate change; specifically including: System dynamic response: The climate and solar term adaptive module calculates... The timeframe was increased by 2 days (significantly delayed), which updated the user's dynamic spatiotemporal context and determined that the actual phenology was still in the damp and cold state of "late spring".

[0142] Adjustment of the plan: Based on the updated dynamic spatiotemporal context, the system has temporarily adjusted the plan, strengthening the dimension of "strengthening the spleen and removing dampness" on the basis of "protecting the heart". The original recommended "lily and lotus seed soup" has been slightly adjusted to "coix seed, poria and yam porridge", and the exercise suggestion has been adjusted from "slightly strenuous morning run" to "indoor Baduanjin". The standard start of summer plan will be restored after the weather warms up.

[0143] Results: This embodiment demonstrates that the system not only calibrates the time, but also dynamically adjusts the program content according to the actual climate, possessing true environmental adaptability.

[0144] Example 7 This embodiment, based on the DS evidence theory, addresses the identification of constitution based on contradictory physical signs. It demonstrates how this invention utilizes DS evidence theory for rational integration and decision-making when a user's four diagnostic methods (inspection, auscultation and olfaction, palpation, and olfaction) information is contradictory. Specifically, it includes: D1. Implementation Scenarios and Data Acquisition In this embodiment, the user is Mr. Zhang. He completes the collection of multimodal physiological data through the system's AI intelligent diagnostic terminal. The collected data includes tongue appearance, pulse appearance, facial appearance, and voice appearance. At the same time, the user inputs consultation information (usually feeling cold and having cold hands and feet) through the system's interactive interface. The above-mentioned collected data and consultation information are uploaded to the system's cloud server for subsequent constitution identification and analysis.

[0145] D2. Construction of Multimodal Evidence and Basic Probability Assignment The system treats the collected multimodal physiological data and consultation information as independent evidence, and assigns basic probabilities to each evidence based on the TCM constitution identification knowledge base and historical labeled data, as follows: Tongue image evidence ( The user's tongue appearance is described as "red tongue with a yellow, greasy coating," and the corresponding basic probability distribution is as follows:

[0146] Where Θ represents uncertainty, i.e., the combination of physical constitutions that the evidence cannot distinguish; Pulse evidence body ( The user's pulse is deep, thready, and weak, with the corresponding basic probability distribution as follows:

[0147] Medical history evidence ( The user reports feeling cold and having cold hands and feet; the corresponding basic probability distribution is as follows:

[0148] D3. Multi-evidence fusion calculation based on Dempster's evidence theory: The system uses Dempster's combination rule to fuse the above evidence bodies layer by layer. The specific process is as follows: D31, First Stage: Tongue Image Evidence ( ) and pulse evidence ( Fusion: Calculating conflict quality: Sum of the basic probabilities that the intersection of propositions in two pieces of evidence is an empty set.

[0149] Calculate the normalization constant : ; Calculate the base probability after fusion ( For each proposition in the identification framework Θ, the Dempster combination rule is used for calculation, and the specific formula is as follows:

[0150] because ,only The effective probability of contribution is calculated. And uncertainty The proportion is extremely high; Integration characteristics: Due to the significant differences in constitution indicated by tongue appearance (supporting damp-heat constitution and yin deficiency constitution) and pulse appearance (supporting qi deficiency constitution and yang deficiency constitution), the confidence level of each constitution is low after integration, reflecting strong conflict.

[0151] D32, Second Stage: Fusion Results ( ) and the body of medical history evidence ( ) fusion Medical history evidence ( It has extremely high support for those with Yang deficiency constitution. ),and During fusion, the confidence level of Yang deficiency constitution is enhanced by Dempster's combination rule; Final fusion result ( ): , .

[0152] D4. Systemic constitution assessment and generation of personalized health intervention strategies D41. Constitutional Assessment Result: Based on the above DS evidence fusion results, the system determined that Mr. Zhang's dominant constitution is Yang deficiency, with a confidence level of 0.78. At the same time, it identified that the damp-heat characteristics shown in the tongue appearance were a false appearance caused by short-term inducing factors (such as short-term dietary influences), and that the pulse and consultation information together pointed to Yang deficiency as the essence of his constitution.

[0153] D42. Personalized Health Intervention Strategy Generation: The system combines the personalized environmental spatiotemporal representation generated for the user (including real-time weather, calibrated solar terms, and phenological information of the user's location) with the core conditioning principle of warming yang and strengthening the spleen to generate a health intervention strategy: recommending the consumption of warm foods such as ginger, mutton, and longan, and avoiding cold drinks and raw foods; at the same time, matching exercise and rest suggestions that are suitable for the current spatiotemporal characteristics to avoid using heat-clearing and dampness-removing herbs due to misjudgment of damp-heat constitution, and to prevent damage to the user's yang qi.

[0154] This embodiment demonstrates that even when the information from the four diagnostic methods is contradictory, the system can effectively avoid misjudgment and improve clinical safety by using the DS evidence theory for robust diagnosis and combining it with dynamic spatiotemporal context to generate a treatment plan.

[0155] Example 8 This example demonstrates the pathway from "phlegm-dampness constitution" management to "metabolic syndrome" assisted intervention, showcasing how the system, based on long-term accumulated data, facilitates a natural transition for users from health management to disease early warning and assisted intervention.

[0156] Phase 1 (Health Management Period): User Mr. Wang was diagnosed with "phlegm-dampness constitution" by the system for a year and was continuously provided with dietary therapy and exercise plans to strengthen the spleen and remove dampness. Through closed-loop management, the system found that his "phlegm-dampness" imbalance was slowly but continuously worsening and that he was beginning to show concurrent characteristics of "blood stasis".

[0157] Phase Two (Risk Warning Period): The system, combining the user's physical examination data (entered with user authorization), discovers that the user's blood lipid and blood sugar levels are consistently approaching critical values. At this point, the system no longer simply outputs a health regimen plan, but generates a "Health Risk Warning Report," clearly indicating a high risk of "metabolic syndrome," recommending that the user seek further examination from medical professionals, and providing long-term tracked physical data and trend charts to the user and doctor for reference.

[0158] Phase Three (Assisted Intervention Period): The user is clinically diagnosed with "metabolic syndrome" (phlegm-stasis syndrome). After the doctor's confirmation, the system switches to "Assisted Diagnosis and Treatment Mode": Syndrome Confirmation: Based on the user's latest tongue, pulse, facial, and symptom data, the system confirms that the user meets the criteria for "phlegm and blood stasis syndrome".

[0159] Solution generation: The system intelligently matches a modified prescription based on "Wen Dan Tang combined with Tao Hong Si Wu Tang" from the "Therapeutic Prescriptions Knowledge Base", lists the core ingredients and dosage ranges, and clearly marks "This solution is an auxiliary treatment suggestion. Please adjust and use it according to your doctor's advice".

[0160] Lifestyle management: At the same time, the system generates more stringent medicinal diet therapy plans (such as recommending hawthorn and cassia seed tea) and exercise suggestions (such as emphasizing the duration and frequency of aerobic exercise) that are coordinated with the treatment plan.

[0161] This embodiment demonstrates that the data foundation and model architecture constructed in this invention have the powerful potential to naturally transition from "prevention of disease" to "treatment of existing disease" in decision support, achieving seamless integration of health management throughout the user's entire life cycle.

[0162] In summary, the present invention has the following advantages: 1. By achieving a multi-dimensional, three-dimensional, and precise matching of time (solar terms), space (region), and individual (physique, gender, age) throughout the entire life cycle, health and wellness advice is transformed from a generalized form to a customized form for each unique individual, effectively improving the effectiveness and safety of health conditioning. 2. By responding to sudden climate changes, changes in the user's place of residence, and changes in the user's physical condition, the system can adjust the health conditioning plan in real time. It has a high degree of intelligence and operational flexibility and can continuously adapt to the user's actual dynamic environment and individual condition. 3. By introducing a data-driven closed-loop feedback mechanism, and through the continuous collection and analysis of user physiological data, program execution data, and effect data, the intervention effect of TCM health preservation programs can be quantified and tracked, enabling scientific verification of the program's effectiveness. 4. By pushing health warning information to users in advance and providing personalized health guidance, the system transforms complex TCM health preservation knowledge into simple and feasible daily action guidelines, significantly improving users' compliance with health conditioning plans and optimizing the user experience. 5. By integrating personalized environmental spatiotemporal representations, health status representations, and user gender and age information to generate personalized solutions, the disconnect between human and policy is resolved. The disconnect between diagnosis and treatment is resolved through closed-loop management and optimization modules, thereby systematically overcoming the core bottlenecks of existing technologies.

[0163] The above specific embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to examples, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A health management intelligent agent system based on dynamic spatiotemporal calibration, characterized in that, include: The data acquisition module is used to acquire users' health status data and environmental spatiotemporal data; An environmental spatiotemporal processing module is used to process the environmental spatiotemporal data to generate a personalized environmental spatiotemporal representation. A health status processing module is used to process the health status data to generate a health status representation. The strategy generation module is used to integrate the personalized environmental spatiotemporal representation with the health status representation and output personalized health intervention strategies. Terminal visualization module: Used to establish data interaction with data acquisition module, health status processing module, environmental spatiotemporal processing module and strategy generation module, realize the visualization guidance of associated data acquisition, the visualization display of associated strategy generation and support the feedback data reception and transmission function for closed-loop optimization.

2. The health management intelligent agent system based on dynamic spatiotemporal calibration according to claim 1, characterized in that, The data acquisition module includes: Multimodal physiological sensing unit: configured to collect users' multimodal physiological data via an AI-powered intelligent diagnostic terminal; Environmental spatiotemporal perception unit: configured to acquire real-time latitude, longitude and altitude data of the user through a high-precision positioning module, and acquire real-time meteorological data, historical meteorological data and phenological observation data of the user's location through a meteorological data API interface.

3. The health management intelligent agent system based on dynamic spatiotemporal calibration according to claim 1, characterized in that, The health status processing module includes: Multimodal health status fusion unit: used to construct a health status assessment system based on DS evidence theory and generate health status representation; Concept digitization mapping unit: used to handle ambiguity and subjectivity, transforming relevant concepts into quantifiable and calculable numerical parameters, providing digital support for data interaction and collaborative operation between various functional modules of the system; Closed-loop learning and optimization unit: used to build an adaptive learning mechanism based on Bayesian networks, combining historical health data and real-time user feedback information to optimize system model parameters and personalized health intervention strategies.

4. The health management intelligent agent system based on dynamic spatiotemporal calibration according to claim 1, characterized in that, The strategy generation module includes: Intelligent health preservation knowledge base and compatibility engine unit: used for structured intelligent health preservation knowledge base of food and medicine homology; The "Food and Medicine from the Same Source" recommendation unit is used to intelligently match and fine-tune the combination of food and Chinese medicine based on the user's current constitution, degree of imbalance, solar term, local climate characteristics, gender, and age. Food and medicine homology cooking unit: used to match the appropriate cooking method according to the scheme provided by the food and medicine homology recommendation unit; Exercise and rest suggestion unit: It is used to create an exercise and rest plan that includes suitable daily routines, exercise plans and intensity, based on the user's local sunrise and sunset times, temperature and humidity changes and personal lifestyle habits; Natural Language Generation and Interaction Unit: This unit integrates generative artificial intelligence technology, receives structured data output from the solution generation engine, and transforms it into health guidance content.

5. The health management intelligent agent system based on dynamic spatiotemporal calibration according to claim 4, characterized in that, Also includes: Gender Differentiation Generation Unit: Used to fine-tune the scheme based on gender differentiation; Age-Stage Adaptation Unit: Used to prioritize and fine-tune the treatment plan according to the physiological characteristics and core treatment principles of each stage; Closed-loop management and optimization unit: Regularly revisit users through AI terminals to collect updated physiological indicator data; analyze the interaction between the evolution trend of physical condition and changes in meteorological data through multi-dimensional algorithms, and dynamically optimize conditioning strategies and model parameters.

6. The health management intelligent agent system based on dynamic spatiotemporal calibration according to claim 1, characterized in that, The terminal visualization module includes: Acquisition guidance unit: used to receive the acquisition request from the data acquisition module, provide acquisition process guidance when acquiring multimodal physiological data, and display the acquisition progress and data verification status in real time; Strategy Display Unit: Used to receive the personalized health intervention strategy output by the strategy generation module and the personalized environmental spatiotemporal representation generated by the environmental spatiotemporal processing module, present the health intervention strategy on the user's smart device client, and simultaneously display the calibrated spatiotemporal and phenological related information. Feedback interaction unit: used to receive user feedback data on the effect of health intervention strategies and transmit the feedback data to the system to support closed-loop optimization of the health status processing module and the strategy generation module; The data backtracking unit is used to receive the historical health status representation of the health status processing module and the historical personalized environmental spatiotemporal representation of the environmental spatiotemporal processing module, and to display the historical health status, historical personalized environmental spatiotemporal representation and execution records of past health intervention strategies on the user's smart device client. The natural language interpretation subunit is integrated into the strategy display unit to convert structured health intervention strategies into natural language interpretation content and present it on the user's smart device client. The solution interaction subunit is integrated into the strategy display unit. When presenting solutions related to medicine and food in health intervention strategies, it marks the role of food pairing and displays the efficacy descriptions of each pairing role to the user through interactive operations.

7. The health management intelligent agent system based on dynamic spatiotemporal calibration according to claim 6, characterized in that, Also includes: The exercise-assisted display subunit is integrated into the strategy display unit to display visual content related to exercise movements when presenting exercise-related solutions in health intervention strategies, and to provide corresponding playback control functions. The retest reminder subunit is integrated into the feedback interaction unit. It receives the physical fitness retest cycle requirement from the health status processing module and displays a visual reminder and retest process guide on the user's smart device client when the preset time before the next physical fitness retest is approaching.

8. A method for detecting intelligent agents in health management based on dynamic spatiotemporal calibration, characterized in that, include: S1. Obtain the user's health status data; S2. Obtain the spatiotemporal data of the user's environment, wherein the spatiotemporal data includes at least the geographical location; S3. Based on the aforementioned environmental spatiotemporal data, generate a dynamic, personalized environmental spatiotemporal representation that is associated with the user's actual natural rhythm. S4. Based on the fusion analysis of the health status data and the personalized environmental spatiotemporal representation, a personalized health intervention strategy is generated.

9. The health management intelligent agent detection method based on dynamic spatiotemporal calibration according to claim 8, characterized in that, Also includes: Receive feedback data from users after they have implemented the health intervention strategy, including updated physiological indicator data and / or user subjective experience data; Based on the feedback data, the analysis model of the health status data and / or the generation strategy of the health intervention strategy are dynamically adjusted through machine learning algorithms.

10. The health management intelligent agent detection method based on dynamic spatiotemporal calibration according to claim 8, characterized in that, When generating the personalized health intervention strategy, at least one of the user's gender information and age stage information is further integrated, and the intervention element type, target point or execution intensity in the strategy are configured differently based on this.

Images