Personalized health management scheme generation method and device, equipment and medium

Abstract

The invention relates to a personalized health management scheme generation method and device, equipment and a medium. The method comprises the following steps: fusing multi-source health data to form a unified user state; generating an initial candidate scheme set based on the user target and the nutritional requirement; calculating a decision energy consumption value by quantifying historical habit deviation, context convenience and resource complexity cost of each health action in each scheme; comprehensively weighing the physiological benefits and the total decision energy consumption of the schemes, and generating a health scheme list sorted according to the comprehensive benefits; and identifying a high-energy-consumption link in the list and replacing the high-energy-consumption link with a replacement action which is easier to execute, and outputting an optimal scheme after re-evaluation. According to the method, the limitation that only the physiological indexes are optimized and the execution burden of the user is neglected in the prior art is broken through, and by quantifying and optimizing the'decision-making energy consumption ', a personalized scheme which meets the health target and is convenient for the user to persist can be generated, and the actual compliance and long-term effect of health management are effectively improved.

Description

Technical Field

[0001] This invention relates to the fields of computer application technology and intelligent health management technology, specifically to a method, apparatus, equipment and medium for generating personalized health management plans. Background Technology

[0002] In recent years, with the increasing public awareness of health, personalized health management plan generation technology has received widespread attention. This method aims to analyze a user's personal physiological data, lifestyle habits, and health goals, and then use algorithmic models to automatically plan a customized health plan, including diet and exercise, to help users achieve goals such as weight loss, muscle gain, or maintaining health more scientifically and efficiently. Its core is to filter and combine the most suitable action sequences from a vast amount of health knowledge for a specific individual, theoretically providing users with digital guidance from goal setting to daily execution.

[0003] However, existing technologies still have a significant drawback in practice. Most current solution generation methods focus on optimizing the achievement of physiological indicators, such as the precise calculation of calorie deficits or the balance of nutrient ratios, striving to provide an "optimal solution" at the theoretical level. But these "optimal" solutions often overlook non-physiological obstacles faced by users during actual implementation, such as the ease of obtaining the required ingredients, the complexity of the cooking steps, and whether the exercise schedule conflicts with the user's daily routine. While these factors do not directly affect the theoretical health benefits of the solution, they greatly increase the cognitive burden and difficulty of implementation for users, causing many seemingly perfect solutions to be abandoned by users due to "difficulty in sticking to them," ultimately failing to produce any real health benefits. Summary of the Invention

[0004] Based on this, the purpose of the present invention is to provide a method, apparatus, equipment and medium for generating personalized health management plans that can fully consider and optimize the convenience of actual implementation of the plan while ensuring health benefits.

[0005] The objective of this invention is achieved through the following solution:

[0006] In a first aspect, the present invention provides a method for generating a personalized health management plan, comprising the following steps:

[0007] S1: Integrate multi-source heterogeneous health data from user terminals, health devices and third-party services to generate a unified user state vector that includes user physiological data, real-time status and health goals;

[0008] S2: Based on the health goals and real-time nutritional needs of the unified user state vector, retrieve and combine multiple feasible action sequences that meet the basic nutritional constraints from the predefined action knowledge base to generate an initial solution set;

[0009] S3: Based on the unified user state vector, perform personalized decision energy consumption quantification on each candidate scheme in the initial scheme set. By calculating the historical habit deviation cost, context convenience cost, and execution resource complexity cost required for atomic health actions in each candidate scheme, generate a set of labeled schemes with quantified decision energy consumption values.

[0010] S4: Perform multi-objective comprehensive evaluation and ranking on the set of labeled solutions, comprehensively weigh the physiological benefit evaluation value of each candidate solution against its total decision energy consumption value, calculate the comprehensive benefit score that balances health benefits and ease of execution, and generate a list of health solutions based on the comprehensive benefit score.

[0011] S5: Perform final optimization and selection processing on the health plan list based on energy consumption perception, identify the high decision-making energy consumption bottleneck actions of the top-ranked plans in the list, replace the high decision-making energy consumption bottleneck actions based on preset action substitution rules and knowledge base, recalculate the comprehensive benefit score of the updated plans, select the plan with the highest comprehensive benefit score, and generate a health management plan.

[0012] In one embodiment, S1 of the personalized health management plan generation method provided by the present invention specifically includes the following steps:

[0013] S11: Verify the source credibility and fill in missing values ​​for dietary text records, exercise descriptions and manually entered body indicators from user terminals to generate standardized user proactive health data;

[0014] S12: Perform timestamp alignment and data format conversion on physiological signal streams from health devices and food order records from third-party services to generate time-synchronized passive monitoring and consumption data;

[0015] S13: Perform semantic parsing and key information extraction processing on standardized user proactive health data and time-synchronized passive monitoring and consumption data, mapping unstructured behavioral descriptions into structured action, nutrition, and time tags to generate a unified user state vector.

[0016] In one embodiment, step S2 of the personalized health management plan generation method provided by the present invention specifically includes the following steps:

[0017] S21: Perform health goal and real-time nutritional gap analysis on the unified user state vector, transform the user's abstract exercise goal into specific numerical constraints on protein, carbohydrates and calories that need to be met in the current time period, and generate a quantitative nutritional demand vector.

[0018] S22: Based on the quantitative nutritional requirement vector, perform multi-condition retrieval processing on the predefined action knowledge base, and use the nutritional components of a single atom health action as the matching condition to filter out all candidate actions that can partially meet the quantitative nutritional requirement vector, and generate a candidate action pool.

[0019] S23: Perform combination optimization and sequence arrangement processing on the actions in the candidate action pool. Based on the dietary matching rules, combine multiple atomic actions into a complete meal-exercise plan, ensuring that the overall nutritional data of each combination plan falls within the feasible range specified by the quantitative nutritional requirement vector, and generate an initial plan set.

[0020] In one embodiment, step S3 of the personalized health management plan generation method provided by the present invention specifically includes the following steps:

[0021] S31: Based on the user's historical behavior records in the unified user state vector, calculate the historical habit deviation degree for each atomic health action in the candidate scheme, analyze the execution frequency of the atomic health action in similar times and situations in the past. The lower the frequency, the higher the degree of deviation of the atomic health action from the user's habits. Generate the historical habit deviation cost for each action.

[0022] S32: Combining the real-time location, time, and device availability context information in the unified user state vector, perform contextual convenience assessment processing on each atomic health action to determine whether the location, tools, or ingredients required to perform the atomic health action match the user's current state. The lower the matching degree, the higher the convenience cost, and generate the contextual convenience cost of each action.

[0023] S33: Based on the standard steps, estimated time, and required special skills of the atomic health actions extracted from the action knowledge base, perform execution resource complexity analysis on each atomic health action. The more steps, the longer the time, or the more special the required skills, the higher the complexity. Generate the execution resource complexity cost for each action.

[0024] S34: The cost of deviation from historical habits, the cost of contextual convenience, and the cost of execution resource complexity are weighted and integrated. Based on the preset weight coefficients in the user profile, the three are combined into a comprehensive quantitative decision energy consumption value, which is then associated with the corresponding atomic health action to generate a set of labeled schemes with quantitative decision energy consumption values.

[0025] In one embodiment, step S4 of the personalized health management plan generation method provided by the present invention specifically includes the following steps:

[0026] S41: Sum and aggregate the quantitative decision energy consumption values ​​of all atomic health actions contained in each candidate scheme in the labeled scheme set to obtain the total decision energy consumption value required to execute the candidate scheme;

[0027] S42: Extract physiological benefit data from the set of labeled schemes, obtain the calorie and nutrient compliance data from the pre-associated nutritional model of each candidate scheme, and convert them into standardized physiological benefit evaluation values.

[0028] S43: Based on a preset scoring function, perform multi-objective balance scoring on the total decision energy consumption value and physiological benefit evaluation value of each candidate scheme to generate a comprehensive benefit score;

[0029] S44: Based on the comprehensive benefit score, sort all candidate schemes in the labeled scheme set, generate an ordered scheme sequence according to the score from high to low, and generate a list of healthy schemes.

[0030] In one embodiment, step S5 of the personalized health management plan generation method provided by the present invention specifically includes the following steps:

[0031] S51: Identify and process high decision-making energy consumption bottleneck actions in the list of health plans. For at least one health plan with a high comprehensive benefit score in the list, analyze the quantified decision-making energy consumption value of each atomic health action contained in the health plan one by one. Mark the actions whose decision-making energy consumption value is significantly higher than the average energy consumption level of all actions in the health plan as high decision-making energy consumption bottleneck actions.

[0032] S52: Perform low-energy alternative action retrieval processing for high decision-making energy bottleneck actions, query the health action knowledge base according to the preset action alternative rules, match low-decision energy alternative solutions with similar nutritional function characteristics and behavioral execution characteristics, and select the one with the highest matching degree as the health alternative action.

[0033] S53: Update and re-evaluate the original plan that includes bottleneck actions with high decision-making energy consumption, replace the corresponding bottleneck actions in the original plan with health alternative actions, form an updated health plan, and recalculate the total decision-making energy consumption value and physiological benefit evaluation value of the updated health plan to generate an updated comprehensive benefit score.

[0034] S54: Perform a final screening of all original and updated health plans in the health plan list, compare the overall benefit scores of all plans, and select the plan with the highest score as the health management plan. The health management plan is used to instruct users to implement a personalized health plan that includes specific meals and exercise.

[0035] Secondly, the present invention provides a personalized health management plan generation system, which is configured with the following modules:

[0036] The multi-source health data fusion module is used to fuse multi-source heterogeneous health data from user terminals, health devices and third-party services to generate a unified user state vector that includes user physiological data, real-time status and health goals.

[0037] The initial solution generation module is used to retrieve and combine multiple feasible action sequences that meet basic nutritional constraints from a predefined action knowledge base based on the health goals and real-time nutritional needs of a unified user state vector, and generate an initial solution set.

[0038] The decision energy consumption quantification and labeling module is used to perform personalized decision energy consumption quantification processing on each candidate scheme in the initial scheme set based on a unified user state vector. By calculating the historical habit deviation cost, context convenience cost, and execution resource complexity cost required for atomic health actions in each candidate scheme, a set of labeled schemes with quantified decision energy consumption values ​​is generated.

[0039] The multi-objective comprehensive evaluation and ranking module is used to perform multi-objective comprehensive evaluation and ranking on the set of labeled solutions. It comprehensively weighs the physiological benefit evaluation value of each candidate solution with its total decision energy consumption value, calculates a comprehensive benefit score that balances health benefits and ease of execution, and generates a list of health solutions based on the comprehensive benefit score.

[0040] The energy consumption perception scheme optimization module is used to perform final optimization and selection processing on the health scheme list based on energy consumption perception. It identifies the high decision-making energy consumption bottleneck actions of the top-ranked schemes in the list, replaces the high decision-making energy consumption bottleneck actions based on preset action substitution rules and knowledge base, and recalculates the comprehensive benefit score of the updated scheme. It then selects the scheme with the highest comprehensive benefit score and generates a health management scheme.

[0041] Thirdly, this application provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement any of the above-mentioned personalized health management scheme generation methods.

[0042] Fourthly, this application provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements any of the above-mentioned personalized health management scheme generation methods.

[0043] In summary, the personalized health management plan generation method provided in this application achieves the goal of constructing a comprehensive and dynamic user health status profile through the fusion processing of multi-source heterogeneous data, laying a precise data foundation for the generation of personalized plans. Secondly, this method innovatively introduces and quantifies the dimension of "decision-making energy consumption," transforming previously overlooked abstract obstacles such as user habit deviations, ease of execution scenarios, and complexity of action resources into calculable and comparable explicit technical parameters. Furthermore, by establishing a multi-objective comprehensive evaluation model integrating physiological benefits and decision-making energy consumption, a paradigm shift can be achieved from simply pursuing the optimal nutritional solution to actively balancing "health benefits" and "execution convenience," thereby generating a list of candidate plans that better reflects users' real-life scenarios. Finally, through bottleneck identification and substitution optimization mechanisms based on energy consumption perception, the implementation threshold of the plan can be dynamically lowered, ensuring that the final selected health management plan is not only scientifically sound but also easily accepted and adhered to by users in practice, thus significantly improving the actual compliance and final effect of health interventions.

[0044] To better understand and implement this invention, the following detailed description is provided in conjunction with the accompanying drawings. Attached Figure Description

[0045] Figure 1 A flowchart illustrating a method for generating a personalized health management plan, as provided in an embodiment of this application;

[0046] Figure 2 This is a schematic diagram of a personalized health management plan generation device provided in another embodiment of this application. Detailed Implementation

[0047] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Preferred embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a thorough and complete understanding of the disclosure of the invention.

[0048] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0049] In one embodiment, such as Figure 1As shown, a method for generating personalized health management plans is provided. This embodiment illustrates the method using a terminal as an example. It is understood that this method can also be applied to a server, and furthermore, to a system including both a terminal and a server, and is implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0050] S1: Integrate multi-source heterogeneous health data from user terminals, health devices, and third-party services to generate a unified user state vector that includes user physiological data, real-time status, and health goals.

[0051] Specifically, the system, based on multi-source data fusion technology and vector normalization theory, collects heterogeneous health data from user terminals, health devices, and third-party services through preset interfaces. User terminal data includes basic physiological data, dietary behavior data, exercise behavior data, preference and constraint data, and natural language description data entered through the interactive interface. Health device data is acquired synchronously via communication protocols such as Bluetooth or WiFi, and third-party service data is acquired synchronously via API interfaces. The system standardizes the collected heterogeneous data, eliminating differences in data format and reliability. Numerical data is uniformly converted to standard units and outliers are removed. Textual data undergoes keyword extraction and standardized mapping using natural language processing technology. Missing data is supplemented using interpolation based on historical user data. The system sets dynamic weight allocation rules based on the reliability characteristics of different data sources, assigning corresponding weights to various types of standardized data to achieve credibility calibration. The system maps the processed standardized data into a fixed-dimensional vector structure, which includes physiological data sub-vectors, real-time status sub-vectors, health goal sub-vectors, habit constraint sub-vectors, context sub-vectors, and resource constraint sub-vectors. After normalization, the data of each dimension is integrated to form a unified user status vector. The generated user status vector is stored in the user health record database in real time and updated at a preset frequency.

[0052] Each sub-vector of the unified user state vector carries health information in different dimensions. The physiological data sub-vector covers basic physiological parameters of the user; the real-time state sub-vector covers current calorie intake, total calorie expenditure, calorie deficit, and the matching status of various nutrient intake with the target; the health goal sub-vector covers target type, target period, and target quantification; the habit constraint sub-vector covers taste preferences, dietary restrictions, allergen information, historical dietary pattern characteristics, and historical exercise pattern characteristics; the context sub-vector covers the distribution of active time periods in the current time period and recent continuous achievement of targets; and the resource constraint sub-vector covers the availability of food channels, accessibility of sports venues, and the daily available time for health management. The integration of these sub-vectors ensures a comprehensive and structured expression of user health information.

[0053] S2: Based on the health goals and real-time nutritional needs of the unified user state vector, retrieve and combine multiple feasible action sequences that meet the basic nutritional constraints from the predefined action knowledge base to generate an initial solution set.

[0054] Specifically, based on rule-based reasoning and knowledge base retrieval technology, the system analyzes the health goals and real-time nutritional needs in a unified user state vector, extracting basic nutritional constraints. The system calls a predefined action knowledge base, which includes a diet action sub-base, an exercise action sub-base, and a combination rule sub-base. The diet action sub-base stores standardized data related to ingredients and dishes, the exercise action sub-base stores standardized data related to exercise types, and the combination rule sub-base stores the logical rules for combining action sequences. Using the analyzed basic nutritional constraints as the retrieval basis, the system retrieves candidate sets of diet actions from the diet action sub-base that meet the requirements for calorie nutrients and suitability, and candidate sets of exercise actions from the exercise action sub-base that meet the requirements for calorie consumption and suitability. Based on the logical rules in the combination rule sub-base, the system arranges and combines the candidate sets of diet and exercise actions to form complete action sequences containing both diet and exercise action sequences. The system filters the formed complete action sequences, eliminating action sequences that do not meet the basic nutritional constraints or conflict with the combination rules, retaining valid sequences as initial solutions, and integrating them to form an initial solution set.

[0055] The analysis of basic nutritional constraints is based on health goals and physiological data in the user's state vector, clarifying the calorie deficit or balance range and the quantitative standards of various nutrients. It also covers the matching requirements between dietary actions and user taste preferences, the avoidance requirements of dietary actions against dietary restrictions and allergens, and the suitability requirements of exercise actions for venue accessibility and fitness level. The logical rules in the combination rule sub-library clarify the minimum number of daily dietary actions, the time interval requirements between exercise and dietary actions, and the coordination requirements of exercise and diet under different health goals, ensuring that the combined action sequence has basic rationality and feasibility. The initial set of solutions must be large enough to meet the needs of subsequent quantitative processing and screening.

[0056] S3: Based on the unified user state vector, perform personalized decision energy consumption quantification processing on each candidate scheme in the initial scheme set. By calculating the historical habit deviation cost, context convenience cost, and execution resource complexity cost required for atomic health actions in each candidate scheme, a set of labeled schemes with quantified decision energy consumption values ​​is generated.

[0057] Specifically, based on path dependence theory and decision cost theory in behavioral economics, the system first decomposes the complete action sequence of each candidate solution in the initial solution set into independent atomic health actions. Each atomic health action corresponds to a unique identifier in the action knowledge base to associate relevant attribute data. The system quantifies three types of costs for each atomic health action: historical habit deviation cost, context convenience cost, and execution resource complexity cost. The historical habit deviation cost is calculated based on the degree of difference between the atomic action and the user's historical health behavior patterns, determined by statistically analyzing the frequency of similar actions performed in the same time period within a recent period. The context convenience cost is calculated based on the degree of fit between the atomic action and the user's current context environment, integrating time cost coefficient, geographic location cost coefficient, and scene adaptation coefficient for comprehensive quantification. The time cost coefficient is determined based on the matching relationship between the time required for action execution and the user's current available time. The geographic location cost coefficient is determined based on the adaptability of the delivery range for food actions and the accessibility of the venue for exercise actions. The scene adaptation coefficient is determined based on the adaptability of the action to the user's current scene.

[0058] The calculation of resource complexity cost is based on the difficulty and operational complexity of acquiring external resources required for atomic actions. It integrates resource acquisition coefficients and operational complexity coefficients for comprehensive quantification. The resource acquisition coefficient is determined based on the ease of obtaining ingredients for the food action and the equipment requirements for the sports action, while the operational complexity coefficient is determined based on the complexity of the cooking steps for the food action and the operational difficulty for the sports action. The system uses a preset weighting rule to perform a weighted summation of the three types of costs to obtain the decision energy consumption value for each atomic action. The system then sums the decision energy consumption values ​​of all atomic actions included in each candidate scheme to obtain the total decision energy consumption value for each candidate scheme. The system associates the total decision energy consumption value with the corresponding candidate scheme to generate a set of labeled schemes with quantified decision energy consumption values. The labeled scheme set retains the core action sequence information of each candidate scheme in the initial scheme set and supplements it with the total decision energy consumption value label to support subsequent evaluation and ranking.

[0059] S4: Perform multi-objective comprehensive evaluation and ranking on the set of labeled solutions, comprehensively weigh the physiological benefit evaluation value of each candidate solution against its total decision energy consumption value, calculate the comprehensive benefit score that balances health benefits and ease of execution, and generate a list of health solutions based on the ranking of comprehensive benefit scores.

[0060] Specifically, the system performs multi-objective comprehensive evaluation and ranking on the labeled scheme set to generate a list of healthy schemes. This process is based on multi-objective decision theory and a weighted summation model. The system constructs a physiological benefit evaluation system, calculating the physiological benefit evaluation value for each candidate scheme. The calculation of the physiological benefit evaluation value covers three core dimensions. The first dimension is the calorie balance fit, where the system calculates the degree of fit between the total calories of the scheme and the target calorie range to obtain a corresponding score. The second dimension is the nutrient balance, where the system calculates the achievement rate of each core nutrient in the scheme and obtains a corresponding score through weighted summation. The third dimension is the goal fit, where the system evaluates the relevance of the scheme based on the type of health goal and obtains a corresponding score. The system integrates the scores of the three dimensions through a weighted summation algorithm to obtain the overall physiological benefit evaluation value for each candidate scheme.

[0061] Furthermore, the system normalizes the total decision energy consumption value of each candidate scheme in the labeled scheme set, mapping the total decision energy consumption value to a specified interval. The system dynamically adjusts the health benefit weight and execution convenience weight based on user profiles, with the sum of the health benefit weight and execution convenience weight being a fixed value. The system uses a weighted summation algorithm, combining the total physiological benefit evaluation value and the normalized total decision energy consumption value, to calculate the comprehensive benefit score for each candidate scheme. The system sorts the candidate schemes in descending order of comprehensive benefit score. For candidate schemes with the same comprehensive benefit score, the system performs a secondary sorting in descending order of total physiological benefit evaluation value. The system limits the length of the sorted scheme list and extracts and displays the core information of each scheme, including the comprehensive benefit score, physiological benefit score, total decision energy consumption value, core action summary, and nutritional composition highlights. The system outputs the sorted health scheme list in a specified format, which can be filtered by the front end according to user needs.

[0062] S5: Perform final optimization and selection processing on the health plan list based on energy consumption perception, identify the high decision-making energy consumption bottleneck actions of the top-ranked plans in the list, replace the high decision-making energy consumption bottleneck actions based on preset action substitution rules and knowledge base, recalculate the comprehensive benefit score of the updated plans, select the plan with the highest comprehensive benefit score, and generate a health management plan.

[0063] Specifically, based on bottleneck optimization theory and action substitution mechanisms, the system performs final optimization and selection processing on the list of healthy solutions based on energy consumption perception. The system first selects top-ranked candidate solutions from the list and identifies high-decision-energy-consumption bottleneck actions for each selected candidate solution. This identification process is achieved by calculating the proportion of decision-making energy consumption values ​​for each atomic action in each solution. Atomic actions exceeding a preset bottleneck threshold are identified as high-decision-energy-consumption bottleneck actions. These high-decision-energy-consumption bottleneck actions are key obstacles affecting the ease of solution execution, and their identification accuracy directly impacts subsequent optimization results. Based on preset action substitution rules and an action knowledge base, the system replaces the identified high-decision-energy-consumption bottleneck actions.

[0064] Preferably, the action substitution rules specify three core requirements: nutritional equivalence, adaptability consistency, and energy consumption reduction. The nutritional equivalence requirement ensures that the caloric value and core nutrient content of the substitution action are reasonably matched with the original bottleneck action. The adaptability consistency requirement ensures that the substitution action remains consistent with the habitual constraints in the user's state vector. The energy consumption reduction requirement ensures that the decision-making energy consumption of the substitution action is lower than that of the original bottleneck action. The system retrieves candidate substitution actions that meet the substitution rules from the action knowledge base, calculates the synergistic adaptability of the candidate substitution action with other atomic actions in the original plan, and selects the candidate substitution action with the highest synergistic adaptability to replace the original bottleneck action. The system recalculates the comprehensive benefit score of the plan after replacing the bottleneck action. The recalculation process includes updating the total decision-making energy consumption value and the physiological benefit evaluation value to ensure that the optimized plan still meets the health target requirements.

[0065] The system selects the plan with the highest overall benefit score after the update as the final health management plan. The final plan must meet the following requirements: an overall benefit score higher than all unoptimized plans and other optimized plans; no high-energy-consumption decision-making bottleneck actions; and physiological benefit evaluation values ​​meeting preset standards. The health management plan generated by the system includes a complete action sequence, decision-making energy values ​​for each action, physiological benefit details, and execution suggestions. The complete action sequence clearly defines the name, portion size, acquisition method or cooking steps of specific dietary actions, as well as the type, duration, intensity, and execution time of specific exercise actions. Physiological benefit details cover core health information such as calorie deficit and nutrient attainment. The execution suggestions provide practical guidance such as ordering and consuming dietary actions and the execution time of exercise actions. The system outputs the final health management plan to the user through a preset display module and supports the association of the plan with functions such as ordering from the online store and exercise check-in, enabling the plan to be implemented effectively.

[0066] In summary, the personalized health management plan generation method provided in this application achieves the goal of constructing a comprehensive and dynamic user health status profile through the fusion processing of multi-source heterogeneous data, laying a precise data foundation for the generation of personalized plans. Secondly, this method innovatively introduces and quantifies the dimension of "decision-making energy consumption," transforming previously overlooked abstract obstacles such as user habit deviations, ease of execution scenarios, and complexity of action resources into calculable and comparable explicit technical parameters. Furthermore, by establishing a multi-objective comprehensive evaluation model integrating physiological benefits and decision-making energy consumption, a paradigm shift can be achieved from simply pursuing the optimal nutritional solution to actively balancing "health benefits" and "execution convenience," thereby generating a list of candidate plans that better reflects users' real-life scenarios. Finally, through bottleneck identification and substitution optimization mechanisms based on energy consumption perception, the implementation threshold of the plan can be dynamically lowered, ensuring that the final selected health management plan is not only scientifically sound but also easily accepted and adhered to by users in practice, thus significantly improving the actual compliance and final effect of health interventions.

[0067] This application constructs a complete closed loop for health management, from data collection to solution implementation, ensuring that personalized health plans are both scientifically accurate and highly executable. Specifically, after new users register and log in through the software on their user terminals, they are guided to select health goals, enter body data, and set dietary preferences and restrictions. The system verifies and standardizes this actively input data, synchronizes data such as basal metabolic rate from health devices, transforms abstract health goals into quantifiable nutritional needs, and then retrieves suitable atomic action combinations from the software's meal database to form an initial plan. At the same time, it quantifies the historical habit deviation cost of users' first encounter with health plans, ensuring that the plans pushed to new users are aligned with their dietary foundation and lowering the entry barrier.

[0068] In daily use, users input natural language descriptions through the nutrition intake analysis or exercise consumption analysis on the software's homepage. The software's AI semantic parsing technology transforms unstructured descriptions into standardized data. The system then fills in missing values ​​and verifies the reliability of this data, aligning it with real-time exercise consumption data from smart wearable devices to achieve timestamp alignment. Based on dynamically updated data, the system calculates the total decision-making energy consumption value and physiological benefit evaluation value of the plan in real time. If the user's actual diet deviates from the plan, the software's health tips can identify high-energy-consumption bottleneck actions, search for low-energy-consumption alternative meals in the store, and update the overall benefit score of the plan. Users can view the optimized plan and specific action suggestions on the homepage, achieving a real-time closed loop of "recording-analysis-optimization-feedback".

[0069] Preferably, when a user has a need to dine out or purchase food, the order data from the software store will be automatically synchronized as passive monitoring and consumption data. The nutritional data of these standardized meals is accurate and does not require users to purchase and cook them themselves, significantly reducing the complexity and cost of execution resources. The patented technology will prioritize including these directly purchasable meals when combining initial solutions. If the original solution contains actions with high execution complexity, it will identify them as high-energy-consuming bottleneck actions and match the same meals in the store as alternatives. Users can complete the purchase with "one-click ordering," which not only meets nutritional needs but also solves the pain points of difficult nutritional recording and high execution costs associated with dining out.

[0070] In the scenario of periodic health review, the software's health record module gathers multi-source information such as users' diet records, exercise data, online store orders, and community check-ins. Patented technology performs semantic analysis and key information extraction on this data, updates a unified user state vector, calculates the total decision energy consumption value and physiological benefit evaluation value based on periodic data, identifies nutritional deficiencies and matches suitable meals from the online store, and generates a structured health report and personalized adjustment plan. Users can obtain professional answers to the plan iteration through the AI ​​nutritionist dialogue function, clarify the optimization direction for the next cycle, and realize the continuous iteration and upgrading of the health plan.

[0071] In one embodiment, S1 of the personalized health management plan generation method provided by the present invention specifically includes the following steps:

[0072] S11: Verify the source credibility and fill in missing values ​​for dietary text records, exercise descriptions, and manually entered body indicators from user terminals to generate standardized user-initiated health data.

[0073] Specifically, the system performs source credibility verification and missing value imputation on dietary text records, exercise descriptions, and manually entered body indicators from user terminals, ultimately generating standardized user-initiated health data. During the credibility verification process, for dietary text records and exercise descriptions, the system verifies the rationality of the content through semantic consistency analysis, and combines this with historical user behavior data to determine the degree of consistency between the description information and past behavioral patterns, eliminating records with obvious logical contradictions. For manually entered body indicators, the system performs validity verification based on preset reasonable ranges of physiological parameters, filtering out indicator data that conforms to common physiological knowledge.

[0074] In the missing value imputation stage, the system fills in missing dietary text records by using trend extrapolation based on the user's historical dietary data for similar time periods; for missing exercise descriptions, it supplements the data by combining the user's recent exercise preferences and frequency characteristics; and for missing body indicator data, it uses interpolation to fill in the missing data based on the user's historical indicator trends and the indicator distribution characteristics of similar populations. After imputation, the system standardizes the format of all verified proactive health data, unifying data representation specifications and units to form structured, standardized user proactive health data.

[0075] S12: Perform timestamp alignment and data format conversion on physiological signal streams from health devices and food order records from third-party services to generate time-synchronized passive monitoring and consumption data.

[0076] Specifically, the system performs timestamp alignment and data format conversion on physiological signal streams from health devices and meal order records from third-party services to generate time-synchronized passive monitoring and consumption data. The system first extracts the raw timestamp information from the physiological signal streams transmitted by the health devices, and simultaneously parses the transaction timestamp information from the meal order records of the third-party services, converting both types of timestamps into a standard time format. During timestamp alignment, the system establishes a time matching window based on the standard timestamp, associating and matching the physiological signal stream data and meal order records within the same time window to ensure the synchronization of the two types of data in the time dimension. In the data format conversion stage, the system performs signal parsing and feature extraction on the physiological signal streams output by the health devices, converting continuous physiological signals into discrete structured monitoring data; and performs field parsing on the meal order records of the third-party services, extracting core fields such as meal name, purchase time, and related nutritional information, converting them into structured consumption data that the system can recognize.

[0077] S13: Perform semantic parsing and key information extraction processing on standardized user proactive health data and time-synchronized passive monitoring and consumption data, mapping unstructured behavioral descriptions into structured action, nutrition, and time tags to generate a unified user state vector.

[0078] Specifically, the system performs semantic parsing and key information extraction on standardized user-initiated health data and time-synchronized passive monitoring and consumption data. It maps unstructured behavioral descriptions into structured action, nutrition, and time tags, ultimately generating a unified user state vector. During semantic parsing, the system uses natural language processing technology to perform semantic analysis on dietary text records and exercise descriptions in the user-initiated health data, identifying core behavioral information within the text. It also semantically associates structured information in the passive monitoring and consumption data, establishing semantic mapping relationships between information from different data sources.

[0079] In the key information extraction phase, the system extracts core information such as user body indicators, dietary behavior, and exercise behavior from the parsed proactive health data; and extracts physiological monitoring indicators, meal consumption information, and corresponding time information from passive monitoring and consumption data. The system converts the extracted unstructured behavioral descriptions into structured action tags, nutrition tags, and time tags according to preset mapping rules. Action tags identify specific dietary and exercise behaviors, nutrition tags associate corresponding nutrition-related attributes, and time tags mark the specific time period in which the behavior occurred. The system integrates all structured tags and corresponding quantitative data to construct a unified user state vector encompassing physiological data sub-vectors, real-time state sub-vectors, health goal sub-vectors, habit constraint sub-vectors, context sub-vectors, and resource constraint sub-vectors. Each sub-vector integrates corresponding information according to preset dimensions, ensuring a comprehensive and structured representation of user health information.

[0080] In one embodiment, step S2 of the personalized health management plan generation method provided by the present invention specifically includes the following steps:

[0081] S21: Perform health goal and real-time nutritional gap analysis on the unified user state vector, transform the user's abstract exercise goal into specific numerical constraints on protein, carbohydrates and calories required in the current time period, and generate a quantitative nutritional demand vector.

[0082] Specifically, the system performs health goal and real-time nutritional gap analysis on the unified user state vector, transforming the user's abstract exercise goals into specific numerical constraints on protein, carbohydrates, and calories required for the current time period, ultimately generating a quantified nutritional demand vector. The system first extracts the health goal sub-vector and real-time state sub-vector from the unified user state vector, analyzing the abstract exercise goals and the matching of current calorie intake, nutrient intake, and corresponding target amounts to identify the real-time nutritional gap. For the abstract exercise goals, based on preset nutritional demand conversion rules and combined with basic physiological parameters in the user's physiological data sub-vector, the system decomposes the abstract goals into quantifiable nutritional indicator constraints. Based on the differences in nutritional needs for different exercise goals, the system determines the priority of protein, carbohydrate, and calorie requirements, establishing a mapping relationship between goals and nutritional indicators. By integrating the real-time nutritional gap and the decomposed nutritional indicator constraints, the system generates a quantified nutritional demand vector containing protein, carbohydrate, and calorie requirement constraints. Each constraint in this vector clearly defines the specific nutritional intake standards to be achieved within the current time period.

[0083] S22: Based on the quantitative nutritional requirement vector, perform multi-condition retrieval processing on the predefined action knowledge base, and use the nutritional components of a single atom health action as the matching condition to filter out all candidate actions that can partially meet the quantitative nutritional requirement vector, and generate a candidate action pool.

[0084] Specifically, the system performs multi-condition retrieval processing on a predefined action knowledge base based on a quantified nutritional requirement vector. Using the nutritional components of individual atom-level health actions as matching conditions, it filters out all candidate actions that partially satisfy the quantified nutritional requirement vector, generating a candidate action pool. The system first parses the nutritional component data of each atom-level health action stored in the action knowledge base, establishing an association index between atom-level actions and nutritional components. During the multi-condition retrieval process, the system uses protein, carbohydrate, and calorie constraints from the quantified nutritional requirement vector as core retrieval conditions, while also incorporating habit constraint sub-vector information from the unified user state vector to supplement auxiliary retrieval conditions such as taste preferences, dietary restrictions, and allergens.

[0085] Preferably, the system traverses all atomic health actions in the action knowledge base, determining whether the nutritional components of each atomic health action partially match at least one nutritional constraint in the quantified nutritional requirement vector, and simultaneously verifying whether the action meets the auxiliary retrieval conditions. For atomic health actions that meet the matching requirements, the system includes them in the candidate pool, and finally integrates all eligible atomic health actions to form a candidate action pool. The candidate action pool is divided into subsets according to action type, corresponding to dietary actions and exercise actions respectively, and each action retains complete nutritional component data and suitability-related information.

[0086] S23: Perform combination optimization and sequence arrangement processing on the actions in the candidate action pool. Based on the dietary matching rules, combine multiple atomic actions into a complete meal-exercise plan, ensuring that the overall nutritional data of each combination plan falls within the feasible range specified by the quantitative nutritional requirement vector, and generate an initial plan set.

[0087] Specifically, the system performs combination optimization and sequence arrangement processing on actions in the candidate action pool. Based on dietary matching rules, it combines multiple atomic actions into complete meal-exercise plans, ensuring that the overall nutritional data of each combination plan falls within the feasible range specified by the quantified nutritional requirement vector, thus generating an initial set of plans. During the combination optimization process, the system first determines the meal division method and the nutritional allocation ratio of each meal based on dietary matching rules, clarifying the nutritional intake range corresponding to different meals. The system selects actions from the dietary action subset and exercise action subset of the candidate action pool respectively, and combines them according to the meal division requirements to form a complete plan framework that includes multiple meal dietary actions and exercise actions.

[0088] In the sequence arrangement phase, the system determines the execution order of each atomic action according to preset action timing rules, ensuring a reasonable connection between dietary and exercise actions in the time dimension. The system performs overall nutritional data verification on each combined plan, summarizing the nutritional composition data of all atomic health actions in the plan, and determining whether the summarized results fall within the feasible range specified by the quantified nutritional requirement vector. For plans with compliant overall nutritional data, the system retains them as valid initial plans; for plans that do not meet the requirements, the system readjusts the action combinations and arrangements until a sufficient number of valid initial plans are generated. Finally, the system integrates all valid initial plans to form an initial plan set, where each plan contains a complete action sequence and corresponding overall nutritional data.

[0089] In one embodiment, step S3 of the personalized health management plan generation method provided by the present invention specifically includes the following steps:

[0090] S31: Based on the user's historical behavior records in the unified user state vector, calculate the historical habit deviation degree for each atomic health action in the candidate scheme, analyze the execution frequency of the atomic health action in similar times and situations in the past. The lower the frequency, the higher the degree of deviation of the atomic health action from the user's habits. Generate the historical habit deviation cost for each action.

[0091] Specifically, based on user historical behavior records in a unified user state vector, the system calculates the historical habit deviation for each atomic health action in the candidate schemes, thereby generating the historical habit deviation cost for each action. The system first extracts complete user historical behavior records from the habit constraint sub-vector of the unified user state vector. It then filters historical action data belonging to the same category as the current atomic health action to be evaluated using action type classification rules, and simultaneously extracts the time and contextual information corresponding to each historical action. The system divides time intervals according to preset time segmentation rules, determines the similar time range corresponding to the current evaluation time, and then defines the core elements of similar contexts through contextual feature matching rules, completing the accurate positioning of similar times and similar contexts.

[0092] The system performs frequency statistics on historical action data after location tracking to determine the execution frequency of the current atomic health action under similar times and circumstances. The system invokes a preset frequency-deviation mapping rule to convert the statistically obtained execution frequency into a corresponding historical habit deviation, with execution frequency and deviation showing a negative correlation. Based on a deviation-cost conversion rule, the system further converts historical habit deviation into historical habit deviation costs. After calculation, the system establishes a correlation between historical habit deviation costs and corresponding atomic health actions.

[0093] S32: Combining the real-time location, time, and device availability context information in the unified user state vector, perform contextual convenience assessment processing on each atomic health action to determine whether the location, tools, or ingredients required to perform the atomic health action match the user's current state. The lower the matching degree, the higher the convenience cost, and generate the contextual convenience cost of each action.

[0094] Specifically, the system combines real-time location, time, and device availability context information from the unified user state vector to perform contextual convenience assessment on each atomic health action, generating a contextual convenience cost for each action. The system extracts real-time location information, current time information, and device availability information from the context sub-vectors of the unified user state vector. Simultaneously, it obtains the location conditions, tool requirements, and food needs required to execute the action by parsing the attribute information of the atomic health action. The system establishes a multi-dimensional matching analysis system to analyze the matching between location conditions and real-time location, the adaptability between tool requirements and device availability, and the fit between food needs and currently available food.

[0095] The system quantifies and scores the matching results for each dimension using preset matching score rules. Then, it weights and sums the individual dimension scores according to the weighting rules for each dimension to obtain a comprehensive matching score. The system calls preset matching score and cost mapping rules to convert the comprehensive matching score into context convenience cost; the comprehensive matching score and context convenience cost are negatively correlated.

[0096] S33: Based on the standard steps, estimated time, and required special skills of the atomic health actions extracted from the action knowledge base, perform execution resource complexity analysis for each atomic health action. The more steps, the longer the time, or the more special the required skills, the higher the complexity, and generate the execution resource complexity cost for each action.

[0097] Specifically, based on the standard step definitions, estimated time consumption, and required special skills of atomic health actions extracted from the action knowledge base, the system performs execution resource complexity analysis on each atomic health action, generating the execution resource complexity cost for each action. Through the action knowledge base access interface, the system accurately extracts the standard step definitions, estimated time consumption data, and required special skill descriptions corresponding to the atomic health action to be evaluated. The system performs hierarchical parsing of the standard step definitions, breaks them down into step units according to the execution order, and counts the number of steps required to execute the action; it directly extracts and records the estimated time consumption data; and it extracts keywords from the required special skill descriptions, matches them with preset skill complexity level classification rules, and determines the complexity level of the special skills.

[0098] The system invokes preset multi-dimensional evaluation rules to convert the number of steps, estimated time, and special skill complexity level into corresponding quantitative scores. Then, based on preset weighting rules, it performs a weighted sum of the quantitative scores for each dimension to obtain a comprehensive execution resource complexity score. Based on preset comprehensive score and cost mapping rules, the system converts the comprehensive execution resource complexity score into execution resource complexity cost.

[0099] S34: The cost of deviation from historical habits, the cost of contextual convenience, and the cost of execution resource complexity are weighted and integrated. Based on the preset weight coefficients in the user profile, the three are combined into a comprehensive quantitative decision energy consumption value, which is then associated with the corresponding atomic health action to generate a set of labeled schemes with quantitative decision energy consumption values.

[0100] Specifically, the system performs weighted fusion processing on the costs of historical habit deviation, contextual convenience, and execution resource complexity, generating a set of labeled schemes with quantified decision-making energy consumption values. The system extracts preset weight coefficients from user profile data. These weight coefficients are configured based on user behavioral characteristics, execution capabilities, and health need priorities, and can be dynamically adjusted based on historical user execution feedback data to balance the influence of the three types of costs in the overall decision-making energy consumption value. The system calls a preset weighted fusion algorithm to multiply the three types of costs corresponding to each atomic health action by their respective weight coefficients, and then sums the results to obtain the comprehensive quantified decision-making energy consumption value for that atomic health action.

[0101] Furthermore, the system iterates through all atomic health actions included in each candidate solution, summarizing the overall quantitative decision energy consumption value for each candidate solution. It then associates the energy consumption values ​​of individual atomic actions and the overall energy consumption value of the solution with the corresponding candidate solutions, completing the energy consumption labeling of the candidate solutions. The system integrates all candidate solutions with quantitative decision energy consumption values, constructing a labeled solution set according to a preset data organization format. This set retains the action sequences and nutritional data information of the original candidate solutions, while supplementing the energy consumption labeling information for each atomic action and the overall solution.

[0102] In one embodiment, step S4 of the personalized health management plan generation method provided by the present invention specifically includes the following steps:

[0103] S41: Sum and aggregate the quantitative decision energy consumption values ​​of all atomic health actions contained in each candidate scheme in the labeled scheme set to obtain the total decision energy consumption value required to execute the candidate scheme.

[0104] Specifically, the system sums and aggregates the quantified decision energy consumption values ​​of all atomic health actions contained in each candidate scheme in the labeled scheme set, accumulating the total decision energy consumption value required to execute the candidate scheme. The system first traverses all candidate schemes in the labeled scheme set, extracting all atomic health actions associated with each candidate scheme and their corresponding quantified decision energy consumption values, establishing a one-to-one correspondence between candidate schemes and atomic action energy consumption values. To ensure computational accuracy, the system performs validity checks after data extraction. By comparing the completeness of the association between atomic health action identifiers and energy consumption values, atomic actions with incompletely labeled energy consumption values ​​are removed. If the scheme still contains valid atomic actions after removal, the calculation continues; if the proportion of invalid atomic actions exceeds a preset range, the scheme is marked as invalid and excluded from the calculation.

[0105] For a single valid candidate solution, the system invokes a preset summation and aggregation algorithm to calculate the total decision energy consumption using a formula:

[0106]

[0107] in, This represents the total energy consumption of the candidate solutions. This indicates the number of valid atomic health actions included in the candidate scheme. This represents the energy consumption value of the quantitative decision for the health action of the i-th atom.

[0108] S42: Extract physiological benefit data from the set of labeled schemes, obtain the calorie and nutrient compliance data from the pre-associated nutritional model of each candidate scheme, and convert them into standardized physiological benefit evaluation values.

[0109] Specifically, the system extracts physiological benefit data from the set of labeled schemes. It obtains the calorie and nutrient compliance data from the pre-associated nutritional model of each candidate scheme and converts it into standardized physiological benefit evaluation values. The system iterates through each candidate scheme in the set of labeled schemes, calls the pre-associated nutritional model of each scheme through a preset interface, and extracts the calorie compliance data and various core nutrient compliance data recorded in the model. The calorie compliance data reflects the degree of fit between the actual calorie intake of the scheme and the calorie constraints in the quantified nutritional requirement vector, while the core nutrient compliance data covers the matching status between the actual intake of key nutrients such as protein and carbohydrates and their corresponding constraints.

[0110] The system first normalizes the extracted multidimensional compliance data, converting the original compliance data from different dimensions into quantitative scores within a unified range using a formula:

[0111]

[0112] in, This represents the standardized score of the k-th class of data. This represents the original qualifying data for the k-th data category. This represents the minimum possible value of the k-th class of data. This represents the maximum possible value of the k-th data category. The system then calls the preset weighted integration rule, using the formula... The standardized physiological benefit evaluation value of each candidate protocol was obtained, where To standardize the physiological benefit evaluation value, The number of data dimensions involved in the integration. represents the weight coefficient for the k-th class of data.

[0113] S43: Based on the preset scoring function, perform multi-objective balance scoring on the total decision energy consumption value and physiological benefit evaluation value of each candidate scheme to generate a comprehensive benefit score.

[0114] Specifically, the system performs multi-objective balanced scoring on the total decision energy consumption value and physiological benefit evaluation value of each candidate solution based on a preset scoring function, generating a comprehensive benefit score. The system extracts the preset scoring function from the data configuration area. This scoring function predefines the integration logic between the total decision energy consumption value and the physiological benefit evaluation value, with the core being the balancing of the weight relationship between health benefits and ease of execution. The system first normalizes the total decision energy consumption value, converting it into a quantitative indicator within a unified range using a formula:

[0115]

[0116] in, This represents the normalized total decision energy consumption value. This represents the maximum total decision energy consumption for all candidate schemes in the labeled scheme set.

[0117] Furthermore, the system substitutes the normalized total decision energy consumption value and standardized physiological benefit evaluation value of each candidate solution into the preset scoring function, and completes the multi-objective balance integration calculation through the formula:

[0118]

[0119] in, For comprehensive benefit scoring, The weighting coefficients for the physiological benefit evaluation value. The weighting coefficients are used to normalize the total decision energy consumption value. and All are dynamic weight parameters, configured based on the intensity of execution intent in the user profile. The intensity of execution intent is determined by continuous achievement records in a unified user state vector; different intensities of execution intent correspond to different... and The values ​​are chosen to balance the priorities of different users regarding health benefits and ease of execution.

[0120] S44: Based on the comprehensive benefit score, sort all candidate schemes in the labeled scheme set, generate an ordered scheme sequence according to the score from high to low, and generate a list of healthy schemes.

[0121] Specifically, the system sorts all candidate solutions in the labeled solution set based on the comprehensive benefit score, generating an ordered sequence of solutions from highest to lowest score, and producing a list of healthy solutions. The system extracts the comprehensive benefit score and corresponding solution information for all candidate solutions in the labeled solution set, establishes a mapping relationship between the comprehensive benefit score and the candidate solutions, and uses a preset sorting algorithm to sort the candidate solutions from highest to lowest comprehensive benefit score. The core logic of the sorting algorithm is implemented through comparison rules, which are as follows:

[0122]

[0123] in, and Candidate solutions With candidate solutions The overall benefit score, and These are the ranking positions of the two items, with higher scores resulting in a higher ranking.

[0124] If two or more candidate solutions have the same comprehensive benefit score during the sorting process, the system will call the preset secondary sorting rules and determine the sorting order by comparing the rules:

[0125]

[0126] in, and Candidate solutions and The normalized total decision energy consumption value, and These are the standardized physiological benefit evaluation values ​​for both options. After sorting, the system generates an ordered sequence of plans, then selects high-quality candidate plans from the sequence according to preset quantity filtering rules, and integrates them to form a list of healthy plans. The list retains the action sequence, total decision energy consumption value, physiological benefit evaluation value, and comprehensive benefit score for each plan.

[0127] In one embodiment, step S5 of the personalized health management plan generation method provided by the present invention specifically includes the following steps:

[0128] S51: Identify and process high decision-making energy consumption bottleneck actions in the list of health plans. For at least one health plan with a high overall benefit score in the list, analyze the quantified decision-making energy consumption value of each atomic health action contained in the health plan. Mark the actions whose decision-making energy consumption value is significantly higher than the average energy consumption level of all actions in the health plan as high decision-making energy consumption bottleneck actions.

[0129] Specifically, the system identifies high-decision-energy-consumption bottleneck actions in the health plan list. First, the system selects at least one health plan from the list with a high overall benefit score, designating it as a key optimization target. For each selected health plan, the system extracts all atomic health actions contained within the plan and their quantified decision-making energy consumption values. The system calculates the average of the decision-making energy consumption values ​​of all atomic health actions within that health plan, using the following formula:

[0130]

[0131] in, This represents the average decision-making energy consumption value for all actions within the plan. This indicates the number of atomic health actions included in the plan. This represents the decision energy consumption value of the i-th atomic health action. The system compares the decision energy consumption value of each atomic health action with the average energy consumption value. Based on the preset deviation judgment rules, actions with decision energy consumption values ​​significantly higher than the average energy consumption level are marked as high decision energy consumption bottleneck actions. This marking process ensures accurate location of key resistance points that affect the ease of implementation of the plan.

[0132] S52: For high-energy-consumption decision-making bottleneck actions, perform low-energy-consumption alternative action retrieval processing, query the health action knowledge base according to the preset action substitution rules, match low-energy-consumption decision-making alternative solutions with similar nutritional function characteristics and behavioral execution characteristics, and select the one with the highest matching degree as the health alternative action.

[0133] Specifically, the system performs low-energy alternative action retrieval processing for high-energy-consumption decision-making bottleneck actions. The system first extracts the core features of the labeled high-energy-consumption decision-making bottleneck actions, including nutritional function features and behavioral execution features. The nutritional function features cover key information such as the corresponding calorie supply and nutrient composition, while the behavioral execution features cover relevant information such as the execution type and required scenario conditions of the action.

[0134] Preferably, the system calls the retrieval interface of the health action knowledge base and initiates a retrieval request based on preset action substitution rules. These rules specify the core conditions that the alternative action must meet, including having similar nutritional functional characteristics to the original bottleneck action, matching the user's habitual constraints and contextual conditions, and having a lower decision-making energy consumption value than the original bottleneck action. The system evaluates the matching degree of the retrieved candidate alternative actions. By comprehensively comparing nutritional functional similarity, behavioral execution adaptability, and energy consumption reduction, the system selects the candidate alternative action with the highest matching degree as the health alternative action, thus achieving precise matching of the alternative action.

[0135] S53: Update and re-evaluate the original plan that includes bottleneck actions with high decision-making energy consumption, replace the corresponding bottleneck actions in the original plan with health alternative actions, form an updated health plan, and recalculate the total decision-making energy consumption value and physiological benefit evaluation value of the updated health plan to generate an updated comprehensive benefit score.

[0136] Specifically, the system updates and re-evaluates the original plan containing high-decision-energy-consumption bottleneck actions. The system replaces the corresponding high-decision-energy-consumption bottleneck actions in the original plan with selected health alternative actions, forming an updated health plan. During the replacement process, the system ensures that the action sequence of the updated health plan is logically coherent, and that the temporal and nutritional relationships between actions conform to preset rules. Preferably, the system recalculates the total decision-making energy consumption value and physiological benefit evaluation value for the updated health plan. The total decision-making energy consumption value is calculated using the following formula:

[0137]

[0138] in, The updated total decision energy consumption value. This represents the total energy consumption value for the original decision-making scheme. The energy consumption value for decision-making in the original bottleneck action. The system calculates the energy consumption value for health alternative actions. The physiological benefit evaluation value is obtained by re-extracting the compliance data from the updated plan's associated nutrition model and calculating it according to preset standardization and weighted integration rules. Based on the new total decision energy consumption value and physiological benefit evaluation value, the system generates an updated comprehensive benefit score for the updated health plan.

[0139] S54: Perform a final screening of all original and updated health plans in the health plan list, compare the overall benefit scores of all plans, and select the plan with the highest score as the health management plan. The health management plan is used to instruct users to implement a personalized health plan that includes specific meals and exercise.

[0140] Specifically, the system performs a final screening of all original and updated health plans in the health plan list to determine the health management plan. The system extracts the comprehensive benefit score of all plans to be screened and establishes a mapping relationship between the score and the corresponding plan. The system sorts and compares all scores, ranking all plans in descending order of comprehensive benefit score. Based on preset screening rules, the system selects the plan with the highest score from the sorted plans as the health management plan. The selected health management plan must fully include a personalized health plan with specific meals and exercise, clearly defining the specific implementation requirements of each atom health action, including the name, quantity, acquisition method or cooking steps of the dietary action, and the type, duration, intensity, and execution time of the exercise action. This health management plan is used to directly instruct the user to perform the corresponding health actions. The system organizes the plan according to a preset format and presents it to the user through a specified output method, while also supporting the association of the plan with related execution assistance functions to ensure the plan's effective implementation.

[0141] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0142] Based on the same inventive concept, this application also provides a personalized health management plan generation device for implementing the personalized health management plan generation method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more embodiments of the personalized health management plan generation device provided below can be found in the limitations of the personalized health management plan generation method described above, and will not be repeated here.

[0143] Preferably, such as Figure 2 As shown, the present invention provides a personalized health management plan generation system 600, which is configured with the following modules:

[0144] The multi-source health data fusion module 610 is used to fuse multi-source heterogeneous health data from user terminals, health devices and third-party services to generate a unified user state vector that includes user physiological data, real-time status and health goals.

[0145] The initial solution generation module 620 is used to retrieve and combine multiple feasible action sequences that meet basic nutritional constraints from a predefined action knowledge base based on the health goals and real-time nutritional needs of a unified user state vector, and generate an initial solution set.

[0146] The decision energy consumption quantification and labeling module 630 is used to perform personalized decision energy consumption quantification processing on each candidate scheme in the initial scheme set based on a unified user state vector. By calculating the historical habit deviation cost, context convenience cost, and execution resource complexity cost required for atomic health actions in each candidate scheme, a set of labeled schemes with quantified decision energy consumption values ​​is generated.

[0147] The multi-objective comprehensive evaluation and ranking module 640 is used to perform multi-objective comprehensive evaluation and ranking on the set of labeled schemes. It comprehensively weighs the physiological benefit evaluation value of each candidate scheme with its total decision energy consumption value, calculates the comprehensive benefit score that balances health benefits and execution convenience, and generates a list of health schemes based on the ranking of the comprehensive benefit scores.

[0148] The energy consumption sensing optimization module 650 is used to perform final optimization and selection processing of the health plan list based on energy consumption sensing. It identifies the high decision-making energy consumption bottleneck actions of the top-ranked plans in the list, replaces the high decision-making energy consumption bottleneck actions based on preset action substitution rules and knowledge base, and recalculates the comprehensive benefit score of the updated plan. It then selects the plan with the highest comprehensive benefit score and generates a health management plan.

[0149] Preferably, the multi-source health data fusion module 610 provided in this application is configured with the following units:

[0150] The Active Health Data Standardization Unit is used to verify the source credibility and fill in missing values ​​of dietary text records, exercise descriptions and manually entered body indicators from user terminals, and generate standardized user active health data.

[0151] The passive monitoring data time synchronization unit is used to perform timestamp alignment and data format conversion processing on the physiological signal stream from health devices and the food order records from third-party services to generate time-synchronized passive monitoring and consumption data.

[0152] The user state vector generation unit is used to perform semantic parsing and key information extraction processing on standardized user active health data and time-synchronized passive monitoring and consumption data, mapping unstructured behavioral descriptions into structured action, nutrition and time tags to generate a unified user state vector.

[0153] Preferably, the initial scheme generation module 620 provided in this application is configured with the following units:

[0154] The nutritional requirement quantification and analysis unit is used to analyze and process the health goals and real-time nutritional gaps of a unified user state vector, transforming the user's abstract exercise goals into specific numerical constraints on protein, carbohydrates and calories that need to be met in the current time period, and generating a quantified nutritional requirement vector.

[0155] The candidate action multi-condition retrieval unit is used to perform multi-condition retrieval processing on a predefined action knowledge base based on the quantified nutritional requirement vector. Using the nutritional components of a single atom health action as the matching condition, it filters out all candidate actions that can partially meet the quantified nutritional requirement vector and generates a candidate action pool.

[0156] The action combination scheme generation unit is used to combine and optimize the actions in the candidate action pool and process the sequence. Based on the dietary matching rules, it combines multiple atomic actions into a complete meal-exercise scheme, ensuring that the overall nutritional data of each combination scheme falls within the feasible range specified by the quantitative nutritional requirement vector, and generates an initial scheme set.

[0157] Preferably, the decision energy consumption quantification and labeling module 630 provided in this application is configured with the following units:

[0158] The historical habit deviation cost calculation unit is used to calculate the historical habit deviation degree of each atomic health action in the candidate scheme based on the user's historical behavior records in the unified user state vector. It analyzes the execution frequency of the atomic health action in similar times and situations in the past. The lower the frequency, the higher the degree of deviation of the atomic health action from the user's habits is determined, and the historical habit deviation cost of each action is generated.

[0159] The context convenience cost assessment unit is used to combine the real-time location, time and device availability context information in the unified user state vector to perform context convenience assessment on each atomic health action, and to determine whether the location, tools or ingredients required to perform the atomic health action match the user's current state. The lower the matching degree, the higher the convenience cost, and the context convenience cost of each action is generated.

[0160] The execution resource complexity cost analysis unit is used to perform execution resource complexity analysis on each atomic health action based on the standard steps, estimated time consumption, and required special skills extracted from the action knowledge base. The more steps, the longer the time consumption, or the more special the required skills, the higher the complexity, and the execution resource complexity cost of each action is generated.

[0161] The decision energy consumption weighted fusion annotation unit is used to perform weighted fusion processing on the cost of historical habit deviation, the cost of context convenience, and the cost of execution resource complexity. Based on the preset weight coefficients in the user profile, the three are combined into a comprehensive quantitative decision energy consumption value, which is then associated with the corresponding atomic health action to generate a set of annotation schemes with quantitative decision energy consumption values.

[0162] Preferably, the multi-objective comprehensive evaluation ranking module 640 provided in this application is configured with the following units:

[0163] The decision energy consumption aggregation calculation unit is used to sum and aggregate the quantitative decision energy consumption values ​​of all atomic health actions contained in each candidate scheme in the labeled scheme set, and accumulate the total decision energy consumption value required to execute the candidate scheme.

[0164] The standardized physiological benefit assessment unit is used to extract physiological benefit data from the set of labeled schemes, obtain the calorie and nutrient compliance data of each candidate scheme from the pre-associated nutritional model, and convert it into a standardized physiological benefit evaluation value.

[0165] The multi-objective comprehensive scoring calculation unit is used to perform multi-objective balance scoring on the total decision energy consumption value and physiological benefit evaluation value of each candidate scheme based on a preset scoring function, and generate a comprehensive benefit score.

[0166] The scheme comprehensive ranking unit is used to sort all candidate schemes in the labeled scheme set based on the comprehensive benefit score, generate an ordered scheme sequence according to the score from high to low, and generate a list of healthy schemes.

[0167] Preferably, the energy consumption sensing scheme optimization module 650 provided in this application is configured with the following units:

[0168] The bottleneck action identification unit is used to identify high decision-making energy consumption bottleneck actions in the list of health plans. For at least one health plan with a high comprehensive benefit score in the list, the unit analyzes the quantified decision-making energy consumption value of each atomic health action contained in the health plan. Actions whose decision-making energy consumption value is significantly higher than the average energy consumption level of all actions in the health plan are marked as high decision-making energy consumption bottleneck actions.

[0169] The low-energy alternative action retrieval unit is used to retrieve low-energy alternative actions for high-decision energy bottleneck actions. It queries the health action knowledge base according to preset action substitution rules, matches low-decision energy alternative solutions with similar nutritional function characteristics and behavioral execution characteristics, and selects the one with the highest matching degree as the health alternative action.

[0170] The scheme update and re-evaluation unit is used to update and re-evaluate the original scheme that includes bottleneck actions with high decision-making energy consumption. It combines health alternative actions to replace the corresponding bottleneck actions in the original scheme, forms an updated health scheme, and recalculates the total decision-making energy consumption value and physiological benefit evaluation value of the updated health scheme, generating an updated comprehensive benefit score.

[0171] The final screening unit for health plans is used to perform a final screening of all original plans and all updated health plans in the health plan list, compare the comprehensive benefit scores of all plans, and select the plan with the highest score as the health management plan. The health management plan is used to instruct users to implement a personalized health plan that includes specific meals and exercise.

[0172] In one embodiment, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described method for generating a personalized health management plan.

[0173] In one embodiment, this application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the above-described method for generating a personalized health management plan.

[0174] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of those different embodiments or examples.

[0175] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0176] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various variations or substitutions within the technical scope disclosed in this application, and these should all be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for generating a personalized health management plan, characterized in that, Includes the following steps: S1: Integrate multi-source heterogeneous health data from user terminals, health devices and third-party services to generate a unified user state vector that includes user physiological data, real-time status and health goals; S2: Based on the health goals and real-time nutritional needs of the unified user state vector, retrieve and combine multiple feasible action sequences that meet the basic nutritional constraints from the predefined action knowledge base to generate an initial solution set; S3: Based on the unified user state vector, perform personalized decision energy consumption quantification processing on each candidate scheme in the initial scheme set. By calculating the historical habit deviation cost, context convenience cost, and execution resource complexity cost required for atomic health actions in each candidate scheme, generate a set of labeled schemes with quantified decision energy consumption values. S4: Perform multi-objective comprehensive evaluation and ranking on the set of labeled schemes, comprehensively weigh the physiological benefit evaluation value of each candidate scheme against its total decision energy consumption value, calculate the comprehensive benefit score that balances health benefits and execution convenience, and generate a list of health schemes based on the comprehensive benefit score. S5: Perform final optimization and selection processing on the health plan list based on energy consumption perception, identify the high decision-making energy consumption bottleneck actions of the top-ranked plans in the list, replace the high decision-making energy consumption bottleneck actions based on preset action substitution rules and knowledge base, recalculate the comprehensive benefit score of the updated plans, select the plan with the highest comprehensive benefit score, and generate a health management plan.

2. The method according to claim 1, characterized in that, S1 includes: S11: Verify the source credibility and fill in missing values ​​for dietary text records, exercise descriptions and manually entered body indicators from user terminals to generate standardized user proactive health data; S12: Perform timestamp alignment and data format conversion on physiological signal streams from health devices and food order records from third-party services to generate time-synchronized passive monitoring and consumption data; S13: Perform semantic parsing and key information extraction processing on the standardized user active health data and the time-synchronized passive monitoring and consumption data, map the unstructured behavior descriptions into structured action, nutrition and time tags, and generate a unified user state vector.

3. The method according to claim 1, characterized in that, S2 includes: S21: Perform health goal and real-time nutritional gap analysis on the unified user state vector, transform the user's abstract exercise goal into specific numerical constraints on protein, carbohydrates and calories that need to be met in the current time period, and generate a quantitative nutritional demand vector. S22: Based on the quantified nutritional requirement vector, perform multi-condition retrieval processing on the predefined action knowledge base, and use the nutritional components of a single atom health action as the matching condition to filter out all candidate actions that can partially satisfy the quantified nutritional requirement vector, and generate a candidate action pool. S23: Perform combination optimization and sequence arrangement processing on the actions in the candidate action pool, combine multiple atomic actions into a complete meal-exercise plan based on dietary matching rules, ensure that the overall nutritional data of each combination plan falls within the feasible range specified by the quantified nutritional requirement vector, and generate an initial plan set.

4. The method according to claim 1, characterized in that, S3 includes: S31: Based on the user's historical behavior records in the unified user state vector, calculate the historical habit deviation degree for each atomic health action in the candidate scheme, analyze the execution frequency of the atomic health action in similar times and situations in the past, the lower the frequency, the higher the degree of deviation of the atomic health action from the user's habits, and generate the historical habit deviation cost for each action. S32: Combining the real-time location, time, and device availability context information in the unified user state vector, perform contextual convenience assessment processing on each atomic health action, determine whether the location, tools, or ingredients required to perform the atomic health action match the user's current state, the lower the matching degree, the higher the convenience cost, and generate the contextual convenience cost of each action. S33: Based on the standard steps, estimated time, and required special skills of the atomic health actions extracted from the action knowledge base, perform execution resource complexity analysis on each atomic health action. The more steps, the longer the time, or the more special the required skills, the higher the complexity. Generate the execution resource complexity cost for each action. S34: Perform weighted fusion processing on the historical habit deviation cost, the context convenience cost, and the execution resource complexity cost, and combine the three into a comprehensive quantitative decision energy consumption value according to the preset weight coefficient in the user profile, associate it with the corresponding atomic health action, and generate a set of annotation schemes with quantitative decision energy consumption values.

5. The method according to claim 1, characterized in that, S4 includes: S41: The quantitative decision energy consumption values ​​of all atomic health actions contained in each candidate scheme in the labeled scheme set are summed and aggregated to obtain the total decision energy consumption value required to execute the candidate scheme. S42: Extract physiological benefit data from the set of labeled schemes, obtain the calorie and nutrient compliance data from the pre-associated nutritional model of each candidate scheme, and convert them into standardized physiological benefit evaluation values. S43: Based on a preset scoring function, perform multi-objective balance scoring on the total decision energy consumption value and the physiological benefit evaluation value of each candidate scheme to generate a comprehensive benefit score; S44: Based on the comprehensive benefit score, sort all candidate schemes in the labeled scheme set, generate an ordered scheme sequence according to the score from high to low, and generate a healthy scheme list.

6. The method according to any one of claims 1-5, characterized in that, S5 includes: S51: Perform high decision-making energy consumption bottleneck action identification processing on the health plan list. For at least one health plan with a high comprehensive benefit score in the list, analyze the quantified decision-making energy consumption value of each atomic health action contained in the health plan one by one, and mark the action whose decision-making energy consumption value is significantly higher than the average energy consumption level of all actions in the health plan as a high decision-making energy consumption bottleneck action. S52: Perform low-energy alternative action retrieval processing on the high decision-making energy bottleneck action, query the health action knowledge base according to the preset action alternative rules, match low decision-making energy alternative solutions with similar nutritional function characteristics and behavioral execution characteristics, and select the one with the highest matching degree as the health alternative action. S53: Update and re-evaluate the original plan containing high decision-making energy consumption bottleneck actions, combine the health alternative actions to replace the corresponding bottleneck actions in the original plan, form an updated health plan, and recalculate the total decision-making energy consumption value and physiological benefit evaluation value of the updated health plan to generate an updated comprehensive benefit score. S54: Perform a final screening of all original plans and all updated health plans in the health plan list, compare the comprehensive benefit scores of all plans, and select the plan with the highest score as the health management plan. The health management plan is used to instruct users to implement a personalized health plan that includes specific meals and exercise.

7. A personalized health management plan generation system, characterized in that, The system includes: The multi-source health data fusion module is used to fuse multi-source heterogeneous health data from user terminals, health devices and third-party services to generate a unified user state vector that includes user physiological data, real-time status and health goals. The initial solution generation module is used to retrieve and combine multiple feasible action sequences that meet basic nutritional constraints from a predefined action knowledge base based on the health goals and real-time nutritional needs of the unified user state vector, and generate an initial solution set. The decision energy consumption quantification and labeling module is used to perform personalized decision energy consumption quantification processing on each candidate scheme in the initial scheme set based on the unified user state vector. By calculating the historical habit deviation cost, context convenience cost, and execution resource complexity cost required for atomic health actions in each candidate scheme, a set of labeled schemes with quantified decision energy consumption values ​​is generated. The multi-objective comprehensive evaluation and ranking module is used to perform multi-objective comprehensive evaluation and ranking on the set of labeled schemes. It comprehensively weighs the physiological benefit evaluation value of each candidate scheme with its total decision energy consumption value, calculates a comprehensive benefit score that balances health benefits and execution convenience, and generates a list of health schemes based on the comprehensive benefit score. The energy consumption sensing scheme optimization module is used to perform final optimization and selection processing on the health scheme list based on energy consumption sensing. It identifies the high decision-making energy consumption bottleneck actions of the top-ranked schemes in the list, replaces the high decision-making energy consumption bottleneck actions based on preset action substitution rules and knowledge base, and recalculates the comprehensive benefit score of the updated scheme. It then selects the scheme with the highest comprehensive benefit score and generates a health management scheme.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 6.

Images