A method, system, device and medium for optimizing nutritional balance of a child's diet

Abstract

This application relates to a method, system, device, and medium for optimizing the nutritional balance of children's diets. The method includes: acquiring children's growth status data and comparing the data with standard growth curve data to obtain standard growth indicators; inputting the standard growth indicators into a classification diagnostic machine learning model to generate children's health profile data; calculating daily nutritional target ranges based on the children's health profile data and growth status data; matching ingredients that meet the daily nutritional target ranges in a food database to generate a draft diet plan; identifying combinations of ingredients with antagonistic effects from the draft diet plan, generating nutritional conflict result information, and optimizing the draft diet plan based on the nutritional conflict result information to obtain a nutritionally balanced optimized children's diet. This method can avoid antagonistic effects in diets that lead to poor absorption and improve the reliability of the diet plan.

Description

Technical Field

[0001] This invention belongs to the field of data processing, and in particular relates to a method, system, device and medium for optimizing the nutritional balance of children's recipes. Background Technology

[0002] With the deep integration of artificial intelligence and digital health technologies, personalized nutrition recommendations have become a cutting-edge direction in children's health. Traditional solutions typically rely on static nutrient databases and rule-based linear computation models, accumulating the tagged nutritional values ​​of ingredients to match standard recommended intakes, thereby achieving a theoretically balanced diet. However, current methods have significant limitations. Essentially, they are a nutritional accounting model, focusing only on the surface balance of total intake while generally neglecting the complex bioavailability of nutrients in the body. For example, antagonistic effects between ingredients can lead to a significant decrease in actual absorption efficiency, but traditional models cannot quantify or mitigate this. Furthermore, this method treats children as a homogeneous group, and its algorithms lack the dynamic response to individual growth stages, physiological states, and absorption differences, and cannot flexibly adjust to real-time environment and dietary preferences. Therefore, the generated recipes are often theoretically balanced but fail to ensure effective nutrient supply in practice, making it difficult to meet the core needs of refined children's health management. Summary of the Invention

[0003] Therefore, it is necessary to provide a method, system, device, and medium for optimizing the nutritional balance of children's diets by capturing the interrelationships of nutrients in the human body, in order to address the aforementioned technical problems.

[0004] Firstly, this application provides a method for optimizing the nutritional balance of children's diets, including:

[0005] Acquire children's growth status data and compare the children's growth status data with standard growth curve data to obtain standard growth indicators;

[0006] Standard growth indicators are input into a classification diagnostic machine learning model to assess growth status and generate children's health profile data.

[0007] Based on children's health profile data and children's growth status data, the daily nutritional target range is calculated; the daily nutritional target range includes the daily calorie target range, the daily macronutrient target range, and the daily key nutrient target range.

[0008] Load the ingredient database, match ingredients that meet the daily nutritional goals, and generate a recipe draft;

[0009] Based on the nutrient interaction map, food combinations with antagonistic effects are identified from the draft recipe, nutrient conflict result information is generated, and the draft recipe is optimized and adjusted based on the nutrient conflict result information to obtain a nutritionally balanced optimized children's recipe.

[0010] Furthermore, the children's growth status data includes parameters of children's absorption capacity, and the nutrient interaction map is constructed using the following methods:

[0011] Based on the nutritional relationship data table, entity recognition is performed to identify nutritional entities, and the nutritional relationships between nutritional entities are extracted from the nutritional relationship data table. Among them, nutritional entities include food entities, nutrient entities and anti-nutrient entities, and nutritional relationships include biochemical relationships, physical relationships and conditional relationships.

[0012] Using nutrient entities as graph nodes and nutrient relationships as edges, a preliminary knowledge graph is constructed, and a logical consistency check is performed on the preliminary knowledge graph to obtain a general knowledge graph.

[0013] Physiological parameter vectors are constructed based on children's absorption capacity parameters, and then the general knowledge graph is personalized based on the physiological parameter vectors to obtain a nutrient interaction graph.

[0014] Furthermore, based on physiological parameter vectors, the general knowledge graph is personalized to obtain a nutrient interaction graph, including:

[0015] Based on the physiological parameter vector, the regulatory rules are matched from the nutritional pathway regulation rule table to obtain the nutritional pathway regulation instructions;

[0016] Based on nutrient pathway regulation instructions, the basic weights of edges in a general knowledge graph are adjusted to obtain personalized weights.

[0017] Traverse the personalized weights, remove edges whose personalized weights are less than the weight threshold, obtain highly relevant edges, and perform necessity verification on the removed edges based on the highly relevant edges to obtain the core relationship set.

[0018] Based on the core relationship set, the original knowledge subgraph is extracted from the general knowledge graph;

[0019] Based on the daily nutritional target range, nodes in the original knowledge subgraph corresponding to key nutrients are defined as the organization core, resulting in a nutrient interaction map.

[0020] Furthermore, ingredients that align with daily nutritional goals are matched against the ingredient database to generate a draft recipe, including:

[0021] Allocate the proportion of the daily nutrient target range to each meal to obtain the daily nutrient supply target plan;

[0022] Based on a food database, the objective function value is calculated with the goal of making the nutritional composition of the food in each meal combination as close as possible to the daily nutrient supply target plan.

[0023] The formula for the objective function value is:

[0024]

[0025] in, Let be the objective function, i be the nutrient index, j be the ingredient index, N be the nutrient set, and J be the ingredient set. Let be the weighting coefficient of the i-th nutrient. Let j be the amount of food ingredient of type j. Let be the nutrient content coefficient of the i-th nutrient in the j-th food group. A daily nutrient supply target plan for nutrient type i;

[0026] By injecting constraints into the objective function, an optimization proposition is obtained. Solving the optimization proposition yields a draft recipe.

[0027] Furthermore, based on the information regarding nutritional conflict outcomes, the draft recipe was optimized and adjusted to obtain a nutritionally balanced and optimized children's recipe, including:

[0028] Based on the nutritional conflict results, solutions with corresponding antagonistic relationships are matched from the nutritional antagonism adjustment strategy library to obtain nutritional antagonism adjustment schemes;

[0029] Based on the nutritional antagonism adjustment plan, the draft recipe was adjusted to obtain the optimized recipe;

[0030] The optimized recipe is input into the acceptance prediction model to predict acceptance and obtain the predicted acceptance. The acceptance prediction model is trained using dietary preference data.

[0031] Based on the predicted acceptance, identify the dishes in the optimized recipe whose predicted acceptance is lower than the acceptance threshold, and obtain the dishes to be adjusted.

[0032] Based on dietary preference data, the meals to be adjusted were fine-tuned to obtain a nutritionally balanced and optimized children's menu.

[0033] Furthermore, based on children's health profile data and children's growth status data, the daily nutritional target range is calculated, including:

[0034] Based on children's health profile data, the corresponding energy consumption and nutritional adjustment parameters for each nutrient can be obtained by querying the nutritional adjustment parameter comparison table.

[0035] Based on children's growth status data, basal energy expenditure, daily total energy requirement, nutrient baseline and micronutrient values ​​are calculated to obtain a list of basic nutrient values; among which, children's growth status data includes basic information and daily activity level;

[0036] Based on the nutritional adjustment parameters, the basic nutritional value list is adaptively adjusted to obtain a draft of personalized nutritional goals.

[0037] The draft personalized nutrition goals were subjected to a physiological range test to obtain the physiological test results. Based on the physiological test results, the draft personalized nutrition goals were adjusted to obtain the daily nutrition goal range.

[0038] Secondly, this application also provides a system for optimizing the nutritional balance of children's recipes, including:

[0039] The growth status module is used to acquire children's growth status data and compare the children's growth status data with standard growth curve data to obtain standard growth indicators;

[0040] The health profile module is used to input standard growth indicators into the classification and diagnostic machine learning model to assess the growth status type and generate children's health profile data.

[0041] The nutrition target module is used to calculate the daily nutrition target range based on children's health profile data and children's growth status data; the daily nutrition target range includes the daily calorie target range, the daily macronutrient target range, and the daily key nutrient target range.

[0042] The recipe draft module is used to load the ingredient database, match ingredients that meet the daily nutritional goals, and generate a recipe draft.

[0043] The optimization and adjustment module is used to identify food combinations with antagonistic effects from the recipe draft based on the nutrient interaction map, generate nutrient conflict result information, and optimize and adjust the recipe draft based on the nutrient conflict result information to obtain a nutritionally balanced optimized children's recipe.

[0044] Thirdly, this application also provides a computer device including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement any step of the method provided in the first aspect of this application.

[0045] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any step of the method provided in the first aspect of this application.

[0046] The aforementioned method, system, equipment, and medium for optimizing the nutritional balance of children's diets involve acquiring children's growth status data and comparing it with standard growth curve data to obtain standard growth indicators. These standard growth indicators are then input into a classification and diagnostic machine learning model to assess growth status types and generate children's health profile data. Based on the children's health profile data and growth status data, daily nutritional target ranges are calculated, including daily calorie target ranges, daily macronutrient target ranges, and daily key nutrient target ranges. A food database is loaded, and ingredients that meet the daily nutritional target ranges are matched to generate a draft diet plan. Based on a nutrient interaction map, combinations of ingredients exhibiting antagonistic effects are identified from the draft diet plan, generating nutrient conflict result information. Based on this nutrient conflict result information, the draft diet plan is optimized and adjusted to obtain a nutritionally balanced optimized children's diet. It can integrate discrete nutritional knowledge, biochemical reaction principles, and group dietary effect data into a structured cause-and-effect network. This network can not only identify simple combinations of ingredients, but also quantitatively simulate the synergistic and antagonistic effects of key nutrients in the digestion and absorption process, thereby predicting the comprehensive bioeffective nutritional output of a meal, rather than a rough estimate of the total intake. It enables dynamic and personalized nutritional needs mapping, transforming children's continuous growth and development indicators, real-time physiological states, and dietary preferences into a series of dynamic parameters. It can calibrate the general knowledge network in real time, generating a unique personal nutritional metabolic context model, so that the recommended focus and intervention intensity are closely matched with the child's current real needs. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 This is a schematic diagram of a method for optimizing the nutritional balance of a children's diet according to an embodiment of the present invention;

[0049] Figure 2 This is a schematic diagram of the structure of a children's recipe nutritional balance optimization system provided in an embodiment of the present invention. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0051] In one embodiment, such as Figure 1 As shown, a method for optimizing the nutritional balance of children's diets is provided. This embodiment illustrates the application of this method to a terminal. It is understood that this method can also be applied to a server, and to a system including both a terminal and a server, and implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0052] Step 101: Obtain children's growth status data and compare the children's growth status data with standard growth curve data to obtain standard growth indicators.

[0053] Child growth status data refers to measured data reflecting a child's current physical condition, typically including basic information such as height, weight, age, and gender, and may include more specific body composition data. Standard growth curve data is a reference standard published by authoritative health institutions, established based on data from a large-scale healthy children population. It usually exists in the form of a graph or data table, showing the normal distribution range of height, weight, and other indicators for children of different ages and genders. Standard growth indicators are a quantitative assessment result obtained by comparing a child's measured growth status data with standard growth curve data; the indicator represents the child's growth level relative to their peers of the same age and gender.

[0054] The terminal collects children's growth data, including but not limited to age, gender, height, and weight, through user input or connection to health devices. This data forms the basis for growth assessment. Based on the child's age and gender, the terminal selects the corresponding standard growth curve(s) from the built-in standard growth curve database. The child's actual measurement value is used as a data point and mapped onto the selected standard curve. Mathematical calculations determine the precise position of the measured value within the distribution represented by the standard curve and calculate the percentile corresponding to that value. This transforms the child's absolute growth value into a statistically and clinically significant relative indicator, thereby objectively and quantitatively determining their growth level within the population and providing accurate input for subsequent health status classification and diagnosis.

[0055] Step 102: Input the standard growth indicators into the classification diagnostic machine learning model to assess the growth status type and generate children's health profile data.

[0056] Specifically, the classification diagnostic machine learning model is a pre-trained computer algorithm model using a large amount of labeled children's growth data. Its function is to analyze and classify the input data. Growth status type assessment is the task this model performs; that is, based on the input standard growth indicators, it determines which predefined growth status category a child belongs to. Children's health profile data is a comprehensive, labeled description of a child's current health status, generated based on the model's assessment results. Examples include normal growth, growth retardation, overweight risk, underweight, or obesity.

[0057] The terminal uses standard growth indicators as input features and passes them to a pre-trained classification and diagnostic machine learning model. The model then performs a comprehensive analysis of these indicators based on the learned patterns and outputs one or more classification labels, transforming the quantitative growth indicators into qualitative health conclusions. This provides a direct basis for developing personalized nutritional intervention plans.

[0058] Step 103: Based on the child health profile data and the child growth status data, calculate the daily nutritional target range; whereby the daily nutritional target range includes the daily calorie target range, the daily macronutrient target range, and the daily key nutrient target range.

[0059] Specifically, the daily nutritional target range is a set of quantitative target ranges for various nutrients that should be consumed daily for a specific child, rather than fixed values. These include: the daily calorie target range, which is the upper and lower limits of the total energy intake that should be consumed daily, measured in kilocalories; the daily macronutrient target range, which is the range of the weight or energy percentage of nutrients such as protein, fat, and carbohydrates that should be consumed daily; and the daily key nutrient target range, which is the target range of micronutrients that are essential for children's growth and development, such as calcium, iron, zinc, vitamin A, and vitamin D.

[0060] Based on children's growth data, the terminal calculates basic energy consumption and basic nutrient requirements. According to the nutritional adjustment principles corresponding to children's health profile data, it makes adaptive adjustments to the basic requirements, determines a reasonable and personalized target range, and transforms the qualitative assessment of health into specific and actionable nutritional intake targets, providing precise quantitative standards for recipe generation.

[0061] Step 104: Load the ingredient database and match ingredients that meet the daily nutritional target range in the ingredient database to generate a recipe draft.

[0062] The food database is a database storing detailed information about various food ingredients. Each record includes the name of the ingredient and the content of various nutrients it contains. The recipe draft is generated by using algorithms to initially screen and combine ingredients from the food database, forming one or more preliminary recipe plans that meet the daily nutritional target range. The draft mainly focuses on achieving macro-level nutrient targets and has not yet considered the interactions between ingredients or taste acceptability.

[0063] The terminal loads a built-in food database, uses the daily nutritional target range as a constraint, and uses the food in the food database as a candidate set. Through a specific optimization algorithm, it calculates the food combinations and quantities for multiple meals a day, so that the total nutrients of all meals are as close as possible to the target range, generating a preliminary recipe plan that basically meets the nutritional requirements, which greatly improves the efficiency of recipe design.

[0064] Step 105: Based on the nutrient interaction map, identify food combinations with antagonistic effects from the draft recipe, generate nutrient conflict result information, and optimize and adjust the draft recipe based on the nutrient conflict result information to obtain a nutritionally balanced optimized children's recipe.

[0065] The nutrient interaction map is a specialized knowledge graph, using nutrients and ingredients as nodes and the relationships between these nodes as edges, to visually represent complex nutritional relationships. Nutrient antagonism refers to the effect where, when two or more nutrients or ingredients are ingested simultaneously, one component hinders the absorption or utilization of another by the body. For example, oxalic acid binds to calcium, affecting calcium absorption. The nutrient conflict outcome information specifically identifies which ingredient combinations in the draft recipe exhibit nutrient antagonism and describes the type and degree of conflict. The nutritionally balanced optimized children's recipe is a revised version of the draft recipe, optimized to address nutrient conflicts. It meets macro-nutritional goals and also considers micro-nutrient interactions, achieving a deeper level of nutritional balance.

[0066] The terminal breaks down all ingredients planned for use in the same meal in the recipe draft into specific nutrients and anti-nutrients. It then iterates through the nutrient interaction map to query the relationships between the components. Once an antagonistic relationship is identified, a nutrient conflict result is generated, clearly indicating the conflicting food pair, the type of conflict, and the possible severity. Based on the conflict type described in the nutrient conflict result, a corresponding solution is matched from a pre-set nutrient antagonism adjustment strategy library. Optionally, for the conflict between oxalic acid and calcium, the strategy might be time-series adjustment, such as arranging spinach and tofu to be eaten in different meals or food substitution, such as replacing spinach with vegetables with low oxalic acid content, such as lettuce, or adding promoters, increasing foods rich in vitamin C to promote iron absorption, indirectly optimizing the overall nutritional balance. The matched solution is applied to the recipe draft and modified. After each adjustment, it is re-verified whether the adjusted recipe still meets the daily nutritional target range and whether new nutrient conflicts have been introduced. This process is repeated multiple times until an optimal or near-optimal solution is found that both meets the macro-nutritional goals and eliminates known severe nutrient antagonistic effects as much as possible.

[0067] This embodiment provides a method for optimizing the nutritional balance of children's diets. The method involves acquiring children's growth status data and comparing it with standard growth curve data to obtain standard growth indicators. These indicators are then input into a classification diagnostic machine learning model to assess growth status types and generate children's health profile data. Based on the children's health profile data and growth status data, daily nutritional target ranges are calculated, including daily calorie target ranges, daily macronutrient target ranges, and daily key nutrient target ranges. A food database is loaded, and ingredients that meet the daily nutritional target ranges are matched to generate a draft diet plan. Based on a nutrient interaction map, combinations of ingredients exhibiting antagonistic effects are identified from the draft diet plan, generating nutrient conflict result information. Based on this information, the draft diet plan is optimized to obtain a nutritionally balanced optimized children's diet. Through the above methods, discrete nutritional knowledge, biochemical reaction principles, and group dietary effect data can be integrated into a structured cause-and-effect network. This network can not only identify simple combinations of ingredients but also quantitatively simulate the synergistic and antagonistic effects of key nutrients in the digestion and absorption process, thereby predicting the comprehensive bioeffective nutritional output of a meal, rather than a rough estimate of the total intake. It enables dynamic and personalized mapping of nutritional needs, transforming children's continuous growth and development indicators, real-time physiological states, and dietary preferences into a series of dynamic parameters. This allows for real-time calibration of the general knowledge network, generating a unique personal nutritional metabolic context model that closely matches the recommended focus and intervention intensity with the child's current actual needs.

[0068] In one embodiment, the child's growth status data includes parameters of the child's absorption capacity, and the nutrient interaction map is constructed using the following method:

[0069] Step 201: Based on the nutritional relationship data table, perform entity recognition to identify nutritional entities and extract the nutritional relationships between nutritional entities from the nutritional relationship data table; wherein, nutritional entities include food entities, nutrient entities and anti-nutrient entities, and nutritional relationships include biochemical relationships, physical relationships and conditional relationships.

[0070] The nutritional relationship data table serves as the original data source for constructing the knowledge graph. It is a structured table storing known scientific facts about the relationships between ingredients, nutrients, and other substances. Entity recognition is the process of automatically identifying and classifying key information objects from the original data table. Nutritional entities refer to identified objects that have independent significance in nutrition. Optionally, these include: food entities (specific foods); nutrient entities (components in food that have physiological benefits); and antinutrient entities (naturally occurring compounds in food that may interfere with the absorption or utilization of other nutrients). Nutritional relationships refer to the interaction patterns between different nutritional entities extracted from the data table. Optionally, these include: biochemical relationships (interactions at the biochemical level, such as oxalic acid combining with calcium to form insoluble precipitates); physical relationships (effects between entities through physical properties, such as dietary fiber adsorbing minerals and hindering their contact with the intestinal mucosa); and conditional relationships (relationships where their occurrence depends on specific conditions, such as vitamin D promoting calcium absorption, but this promotion requires sunlight exposure or sufficient vitamin D levels).

[0071] The terminal reads the nutrient relationship data table, cleans and standardizes the text, traverses each row of the data table, and identifies which type of nutrient entity each cell mentions through keyword matching, dictionary comparison, or simple natural language processing techniques. For example, it identifies spinach as a food entity, oxalic acid as an antinutrient entity, and calcium as a nutrient entity. Based on the identification of entities, it extracts relationship information from the fields describing relationships in the table, and stores the identified entities and extracted relationships in a structured format to prepare for the construction of the graph.

[0072] Step 202: Using nutrient entities as graph nodes and nutrient relationships as edges, a preliminary knowledge graph is constructed, and a logical consistency check is performed on the preliminary knowledge graph to obtain a general knowledge graph.

[0073] Specifically, in graph theory and knowledge graphs, nodes represent entities or concepts. In this embodiment, each identified nutritional entity will become an independent node. Edges represent relationships between nodes. In this embodiment, each extracted nutritional relationship will become an edge connecting two or more corresponding entity nodes. The preliminary knowledge graph is an unverified knowledge graph generated from a set of entities and relationships, and may contain redundant, erroneous, or contradictory information. Logical consistency checking refers to performing a series of rule verifications on the preliminary knowledge graph to ensure that its internal logic is consistent. The general knowledge graph is a relatively complete and accurate knowledge graph obtained after logical checks and quality control. It contains universal nutritional knowledge applicable to the general population and has not yet been personalized for individuals.

[0074] The terminal creates a unique graph node for each nutrient entity. Based on the extracted relationships, it establishes directed or undirected edges between the corresponding entity nodes, and labels the relationship type and possible attributes on the edges. It then runs inspection rules. For example, it checks whether there are both promoting and inhibiting relationships for the same group of entities without reasonable explanation, which is considered a logical contradiction. It also checks whether there are cycles such as A promoting B, B promoting C, and C promoting A, and evaluates their rationality. Finally, it checks whether the attributes of key edges or nodes are missing. For problems found during the inspection, it automatically corrects them according to preset rules. After processing all the problems, a logically consistent and reliable general knowledge graph is obtained.

[0075] Step 203: Based on the child's absorption capacity parameters, a physiological parameter vector is constructed, and based on the physiological parameter vector, the general knowledge graph is personalized to obtain a nutrient interaction graph.

[0076] Specifically, child absorption capacity parameters refer to personalized physiological indicators reflecting a child's ability to absorb and metabolize nutrients. These parameters may be derived from medical testing or inferred through models. A physiological parameter vector is a mathematical representation that combines multiple absorption capacity parameters into an ordered list of numbers. Personalized adjustment refers to modifying a general knowledge graph using an individual's physiological parameter vector to better reflect that individual's true physiological condition. A nutrient interaction map is a personalized knowledge graph specifically tailored to a particular child, emphasizing the most important nutritional relationships and potential problems for that child.

[0077] The terminal acquires the child's absorption capacity parameters, quantifies and standardizes these parameters, and combines them into a physiological parameter vector. Based on the specific values ​​in the physiological parameter vector, it searches for matching rules in a predefined nutritional pathway regulation rule table. According to the matched rules, the basic weights of relevant edges in the general knowledge graph are adjusted. For example, for a child with iron deficiency, the weight of relationships that inhibit iron absorption will be increased because iron is more important to this child; the weight of relationships that promote iron absorption will also be increased because they are positive factors that need to be utilized. Iterates through all the adjusted weights and temporarily removes edges with too low a weight. To ensure that important information is not mistakenly deleted, the necessity of the removed edges is verified to obtain a core set of relationships. A subgraph composed of the core set of relationships is extracted from the general knowledge graph. Combined with the child's current daily nutritional target range, relevant nutrient nodes are defined as cores, organized, and a nutritional interaction graph customized for the child is generated.

[0078] This embodiment transforms a universal knowledge graph into a highly personalized nutrient interaction graph, enabling more precise identification and resolution of the most impactful nutrient antagonism issues for the child during subsequent diet optimization, thus greatly improving the accuracy and effectiveness of personalized nutritional intervention.

[0079] In one embodiment, a personalized adjustment of a general knowledge graph is performed based on a physiological parameter vector to obtain a nutrient interaction graph, including:

[0080] Step 301: Based on the physiological parameter vector, match the regulation rules from the nutrient pathway regulation rule table to obtain the nutrient pathway regulation instructions.

[0081] The physiological parameter vector is a mathematical representation that combines multiple absorption capacity parameters of a specific child into an ordered list of numbers to quantitatively describe the child's physiological state. The nutritional pathway regulation rule table is a predefined data table storing multiple condition-action rules. Each rule specifies which relationships in the knowledge graph should be adjusted and what adjustment actions should be taken when the child's physiological parameters meet certain conditions. A regulation rule refers to a single rule in the rule table. Nutritional pathway regulation instructions are one or more specific, executable adjustment commands triggered after matching the child's physiological parameter vector with the nutritional pathway regulation rule table, specifying which types of edges in the knowledge graph should have their weights adjusted.

[0082] The terminal iterates through each rule in the nutritional pathway regulation rule table, compares the condition part of the rule with the input physiological parameter vector, and once the condition of a rule is met, the corresponding action part of the rule will be activated and transformed into a clear nutritional pathway regulation instruction. A physiological parameter vector may satisfy multiple rules at the same time, thereby generating a set of regulation instructions. The multiple instructions generated by the trigger are integrated or prioritized to avoid conflicts and form a complete set of personalized regulation instructions for the child.

[0083] Step 302: Based on the nutrient pathway regulation instructions, adjust the basic weights of the edges in the general knowledge graph to obtain personalized weights.

[0084] Specifically, a general knowledge graph refers to a knowledge graph containing universal nutritional knowledge that has undergone logical consistency checks. Each edge initially has a base weight, representing the importance or strength of that relationship in the general population. The base weight of an edge refers to the initial weight value of each edge in the knowledge graph before personalized adjustments; the weight is set based on general nutritional knowledge. Personalized weights are new weight values ​​obtained by modifying the base weights of relevant edges in the general knowledge graph according to nutritional pathway regulation instructions, reflecting the importance of that relationship to a specific child.

[0085] The terminal reads each nutrient pathway regulation instruction, which clearly indicates the type of relationship that needs to be adjusted. Based on this, it locates all target edges that meet the conditions in the general knowledge graph. According to the operation specified in the instruction, it performs mathematical calculations on the basic weight of each target edge. For example, the relationship of oxalate inhibiting calcium absorption has a basic weight of 0.6. For a child with normal calcium absorption, the instruction may not make any adjustment; but for a child with calcium deficiency, the instruction may increase its weight to 0.9 to highlight the importance of this antagonistic relationship. It traverses all edges, and the weights of edges covered by the instruction are updated, while the weights of uncovered edges are retained. Each edge in the graph receives a personalized weight.

[0086] Step 303: Traverse the personalized weights, remove edges whose personalized weights are less than the weight threshold, obtain highly relevant edges, and perform necessity verification on the removed edges based on the highly relevant edges to obtain the core relationship set.

[0087] Specifically, the weight threshold is a numerical threshold used to filter which edges are important; edges with weights below this threshold are considered less relevant to the current child. Highly relevant edges refer to those with personalized weights greater than or equal to the weight threshold, and are considered to require priority when constructing the child's personalized atlas. Necessity verification is a safety check step designed to prevent the accidental deletion of crucial, albeit low-weight, relationships during the initial removal of low-weight edges. The core relationship set is the set of all relationships deemed necessary to construct the child's nutritional interaction map after screening and verification.

[0088] The terminal traverses all edges in the knowledge graph, checks their personalized weights, and temporarily removes edges with weights below a preset weight threshold from the current core considerations. The remaining edges form a set of highly relevant edges, constituting the backbone of the personalized graph. The edges removed in the previous step are then reviewed. The verification rule is: check whether a removed edge is a necessary prerequisite or basis for a retained highly relevant edge. For example, the edge that vitamin D promotes calcium absorption is a highly relevant edge and is retained. Then, the edge that sunlight promotes vitamin D synthesis, even if it has a lower weight, is verified as needing to be retained because it is a basic condition of the former. The highly relevant edges are merged with the edges that have passed the necessity check to obtain the core relation set.

[0089] Step 304: Extract the original knowledge subgraph from the general knowledge graph based on the core relationship set.

[0090] The original knowledge subgraph is a smaller and more concise knowledge graph that is cut from the complete general knowledge graph based on the core relation set. The subgraph contains all the edges in the core relation set, as well as all the nodes connected by these edges.

[0091] The terminal uses the core relationship set as an index to find all edges and nodes in the complete general knowledge graph, and recombines the nodes and edges into a new, independent knowledge graph. This generates a dedicated knowledge graph that focuses on specific core nutritional issues for children without redundancy, thus preparing a structured data foundation for generating intuitive and usable nutritional interaction graphs.

[0092] Step 305: Based on the daily nutritional target range, define the nodes in the original knowledge subgraph of the corresponding key nutrients as the organization core to obtain the nutritional interaction map.

[0093] The organizational core refers to the node that serves as the layout and display center when constructing the final knowledge graph. By setting the organizational core, the structure of the knowledge graph can be arranged around a specific theme, making it easier to understand and apply. The Nutritional Interaction Graph is a knowledge graph that is completely customized for specific children and can be directly used for recipe optimization analysis, obtained by reorganizing the original knowledge subgraphs by setting organizational cores.

[0094] The terminal identifies key nutrients that require special attention from the daily nutritional target range calculated for the child. For example, if the target range indicates that calcium and iron need to be supplemented, then these two nutrients are identified as core nutrients. In the original knowledge subgraph, the nodes representing these key nutrients are marked as organizational cores. When the graph is presented graphically, the organizational core nodes are placed at the visual center, and nodes that are directly related to the core nodes are arranged around them, gradually expanding the secondary relationships. The resulting graph, centered on the key nutrients and containing all core relationships, is the nutritional interaction graph.

[0095] This embodiment, through personalization, allows the generated nutrient interaction map to directly focus on the most pressing nutritional issues for children. When using this map to review a draft recipe, it can immediately focus on food combinations that affect the absorption of core nutrients, making the recipe optimization process more efficient and targeted.

[0096] In one embodiment, ingredients that meet the daily nutritional target range are matched in a food database to generate a recipe draft, including:

[0097] Step 401: Allocate the intake proportion of the daily nutritional target range for each meal to obtain the daily nutrient supply target plan.

[0098] Meals refer to all meals eaten throughout the day, typically including breakfast, lunch, and dinner, and may also include snacks in the morning and afternoon. Intake ratios refer to the proportion of the daily nutritional target allocated to each meal; these ratios can be preset based on general nutritional recommendations or the user's dietary habits. A daily nutrient supply target plan breaks down the daily target according to preset intake ratios, resulting in specific nutritional goals to be achieved at each meal.

[0099] According to the preset meal nutrition allocation plan, the terminal allocates each item in the daily nutrition target range, namely total calories, protein, fat, carbohydrates and various key nutrients, to each meal in proportion. Each meal generates an independent nutrition target list. The collection of all meal targets constitutes the daily nutrient supply target plan. This plan provides a precise and quantitative basis for selecting ingredients for each meal.

[0100] Step 402: Based on the food database, calculate the objective function value with the goal of making the nutritional composition of the food in each meal combination as close as possible to the daily nutrient supply target plan.

[0101] The formula for the objective function value is:

[0102]

[0103] in, Let be the objective function, i be the nutrient index, j be the ingredient index, N be the nutrient set, and J be the ingredient set. Let be the weighting coefficient of the i-th nutrient. Let j be the amount of food ingredient of type j. Let be the nutrient content coefficient of the i-th nutrient in the j-th food group. A daily nutrient supply target plan for nutrient type i.

[0104] Specifically, the objective function is a mathematical expression used to quantify the quality of a solution. In this embodiment, the objective function is specifically used to measure the overall gap between the nutritional value of a recipe composed of ingredients and the target plan. The goal is to find the ingredient combination that minimizes the objective function value, i.e., the optimal solution. Ingredient indexes are unique identifiers for each ingredient in the ingredient database. Nutrient indexes are unique identifiers for the considered nutrients. The nutrient set is the set of all considered nutrients, typically including energy, macronutrients, and key micronutrients. The ingredient set is the set of candidate ingredients considered in the current optimization calculation. The weight coefficient is the weight assigned to the i-th nutrient, reflecting its importance; the more important the nutrient, the higher its weight, and the greater its impact on the result during optimization. Ingredient usage is a decision variable to be solved in the optimization calculation, representing the amount of the j-th ingredient used in the recipe. Nutrient content refers to the amount of the i-th nutrient per unit weight of the j-th ingredient; this data comes from the ingredient database. The daily nutrient supply target plan refers to the target value for the i-th nutrient for the meal that is currently being optimized.

[0105] The overall goal of the terminal is to find a combination of ingredients that makes the total amount of various nutrients provided by all ingredients as close as possible to the target amount of the meal. This involves calculating the total amount of the i-th nutrient provided by all ingredients in the recipe, calculating the relative deviation of the i-th nutrient (the difference between the actual amount and the target amount), and then dividing by the target amount. Using relative deviation instead of absolute deviation eliminates differences in units and magnitudes of different nutrients, allowing the objective function to fairly handle gram-level protein and milligram-level iron. The relative deviation is squared to convert all deviations into positive numbers, preventing positive and negative deviations from canceling each other out and increasing the penalty for larger deviations, making the optimization result more favorable to the target amount. The goal is to ensure that all nutrients are close to the target, rather than having some nutrients severely excessive while others are severely deficient. The relative deviation after square is multiplied by the weight of that nutrient. Nutrients with higher importance have a larger weight and a higher proportion in the overall objective function. The optimization process will prioritize ensuring that nutrients with higher importance meet the target. The weighted square deviations of all nutrients are summed to obtain a single quantitative value of the overall nutritional deviation of the entire recipe, which is the objective function value. The goal of optimization is to minimize this overall deviation value. During the optimization process, different combinations of ingredient amounts are tried. For each tried combination, an objective function value is calculated according to the above formula to evaluate the merits of the combination.

[0106] Step 403: Inject constraints into the objective function to obtain the optimization proposition, and solve the optimization proposition to obtain the recipe draft.

[0107] Specifically, constraints are limitations or rules that must be satisfied during the optimization process. They define the feasibility space of the solution. Having only an objective function may lead to a mathematically optimal but practically infeasible solution. An optimization proposition is a complete description of an optimization problem, consisting of an objective function and constraints. Its standard form is to find the value of the decision variable that minimizes the objective function while satisfying the constraints. Solving the problem refers to using mathematical optimization algorithms to find the combination of ingredient quantities that satisfies all constraints and minimizes the objective function. A draft recipe is a solution to the optimization proposition, i.e., a specific, quantified set of ingredient combinations that lists the recommended quantities of each selected ingredient.

[0108] The terminal adds a series of constraints to the objective function to form a complete optimization proposition. For example, the constraints include: non-negativity constraint (the amount of all ingredients used is greater than 0); total weight constraint (the total weight of all ingredients in a meal is within a certain range); ingredient diversity constraint (limiting the frequency or total amount of the same type of ingredient); meal structure constraint (breakfast must include grains and tubers, high-quality protein, etc.); and personal preference constraint (excluding ingredients that children are allergic to or dislike). The terminal then calls the built-in optimization solver to mathematically solve the constructed optimization proposition. The solver automatically searches among all possible solutions that satisfy the constraints for the optimal or near-optimal solution that minimizes the objective function value. The solution found by the solver, i.e., a set of optimal ingredient amounts, is extracted, formatted, and output as a recipe draft.

[0109] This embodiment, through a mathematically optimal and practically feasible preliminary recipe scheme, strictly ensures that the nutritional components of the recipe are as close as possible to the nutritional goals of the meal, while adhering to basic dietary common sense and the user's personal restrictions, thus achieving the automation and scientification of recipe generation.

[0110] In one embodiment, based on nutritional conflict information, the draft recipe is optimized and adjusted to obtain a nutritionally balanced optimized children's recipe, including:

[0111] Step 501: Based on the nutritional conflict results, match solutions with corresponding antagonistic relationships from the nutritional antagonism adjustment strategy library to obtain nutritional antagonism adjustment schemes.

[0112] The nutritional conflict outcome information refers to the description of specific nutritional antagonism issues identified from the recipe draft, including the conflicting food pairs, the specific nutrients involved, and the type and severity of the antagonistic effect. The nutritional antagonism adjustment strategy library is a predefined knowledge base that stores standard solutions or adjustment strategies for various known nutritional antagonistic relationships. A solution is a single handling method for a specific antagonistic relationship in the strategy library. A nutritional antagonism adjustment plan is a set of one or more specific, executable adjustment instructions matched from the strategy library based on the specific nutritional conflict outcome information.

[0113] The terminal reads the nutritional conflict results and uses them as keywords to search and match in the nutritional antagonism adjustment strategy library. The strategy library has multiple pre-set solutions. For example, for oxalate-calcium antagonism, possible solutions include suggesting blanching the oxalate-rich ingredients to remove some of the oxalate; replacing one conflicting ingredient with a nutritionally similar ingredient that will not cause antagonism; and introducing an ingredient that can promote absorption. Based on the context of the conflict and the priority of the strategies, the terminal selects the most suitable solution or combination of solutions to form a nutritional antagonism adjustment plan for the current recipe draft.

[0114] Step 502: Based on the nutritional antagonism adjustment plan, the draft recipe is adjusted to obtain the optimized recipe.

[0115] Specifically, the optimized recipe is a modified version obtained by applying the nutrient antagonism adjustment scheme to the recipe draft, eliminating known significant nutrient antagonism effects.

[0116] The terminal strictly follows the instructions in the nutritional antagonism adjustment plan to modify the draft recipe. If the plan is a time-series adjustment, conflicting ingredients are moved to different meals. If the plan is an ingredient replacement, nutritionally similar and non-conflicting ingredients are searched from the ingredient database for replacement. After adjustment, the nutritional composition of the optimized recipe is recalculated to ensure that it still meets the daily nutrient supply target plan. Since the adjustment may change the nutritional composition, a micro-increase process is required to re-achieve the target. After adjustment and verification, the optimized recipe is output.

[0117] Step 503: Input the optimized recipe into the acceptance prediction model to predict the acceptance level; the acceptance prediction model is trained using dietary preference data.

[0118] Specifically, the acceptance prediction model is a machine learning-trained model that predicts a child's liking for a particular dish or recipe. Dietary preference data, typically derived from historical dietary records, is used to train the model and clearly indicates the child's likes and dislikes of certain ingredients, flavors, and cooking methods. The predicted acceptance rate is a quantitative prediction by the model of the likelihood that the child will accept each dish or the entire optimized recipe.

[0119] The terminal takes the detailed information of the optimized recipe as input features and passes it to the acceptance prediction model that has been trained with dietary preference data. The model calculates based on the learned preference pattern of the child and outputs a set of predicted acceptance scores.

[0120] Step 504: Based on the predicted acceptance, identify the dishes in the optimized recipes whose predicted acceptance is lower than the acceptance threshold, and obtain the dishes to be adjusted.

[0121] The acceptance threshold is a preset acceptance score threshold used to determine whether the predicted acceptance is too low. The dishes or meals to be adjusted are specific dishes or meals in the optimized recipe that are identified as having a predicted acceptance below the acceptance threshold.

[0122] The terminal sets a reasonable acceptance threshold and compares the predicted acceptance of each dish in the recipe with this threshold. All dishes with a predicted acceptance below the threshold are marked and classified as dishes to be adjusted. This accurately identifies the shortcomings in the acceptability of the optimized recipe and points the way for targeted fine-tuning.

[0123] Step 505: Based on dietary preference data, fine-tune the meals to be adjusted to obtain a nutritionally balanced and optimized children's menu.

[0124] Among them, the nutritionally balanced optimized children's recipe is a recipe obtained by further fine-tuning the taste of the meals to be adjusted based on the optimized recipe, which simultaneously meets the three major goals of nutritional balance, optimized absorption and acceptable taste.

[0125] For each meal to be adjusted, the terminal makes fine-tuning based on the child's dietary preference data. Optionally, the fine-tuning strategies include: flavor fine-tuning, adjusting seasonings with minimal changes to nutrition; form fine-tuning, changing the shape of ingredients or cooking methods; and similar substitution, replacing disliked ingredients with those that the child likes and that are nutritionally similar. After each fine-tuning, the terminal quickly verifies whether the adjusted meal still meets the nutritional goals, ensuring that the fine-tuning does not introduce new nutritional imbalances or antagonisms. Once all meals to be adjusted have been fine-tuned and passed nutritional verification, the terminal outputs a nutritionally balanced and optimized children's menu.

[0126] This embodiment ensures that the generated recipes are not only scientifically sound in theory but also highly feasible in actual feeding practices, greatly improving the practicality and personalization of the recipes, which is key to achieving successful long-term nutritional interventions.

[0127] In one embodiment, based on children's health profile data and children's growth status data, the daily nutritional target range is calculated, including:

[0128] Step 601: Based on the children's health profile data, retrieve the corresponding energy consumption and nutritional adjustment parameters for each nutrient from the nutritional adjustment parameter comparison table.

[0129] The child health profile data refers to qualitative classifications of a child's growth status derived from machine learning models, such as normal growth, overweight, or growth retardation. The nutrition adjustment parameter reference table is a predefined, structured data table indexed by different health profile types. It specifies guiding parameters for personalized adjustments to a child's total energy expenditure and nutrient intake when the child belongs to a particular type. Energy expenditure adjustment parameters are adjustment coefficients or directions for a child's daily total energy requirements. Nutritional adjustment parameters are adjustment coefficients or principles for various nutrients; for example, for overweight children, the parameters might require increasing the proportion of energy from protein and decreasing the proportion from fat.

[0130] The terminal uses the child's health profile data as keywords to search the nutrition adjustment parameter reference table. It reads all energy consumption adjustment parameters and various nutrition adjustment parameters corresponding to the health profile from the table. The parameters may be specific numerical coefficients or descriptive rules, and generate a complete set of personalized adjustment parameter instructions for the child.

[0131] Step 602: Based on the child's growth status data, calculate the basic energy consumption, daily total energy requirement, nutrient baseline and trace element values ​​to obtain a list of basic nutrient values; among which, the child's growth status data includes basic information and daily activity level.

[0132] Specifically, children's growth status data includes basic information such as age, sex, height, weight, and daily activity level. Basal energy expenditure (BAE) refers to the minimum energy a child needs to maintain life at rest, calculated based on weight, height, etc. Total daily energy requirement (TOE) is the child's total daily energy expenditure multiplied by an activity coefficient determined based on their daily activity level. Nutrient baselines are the recommended daily intakes of various macronutrients calculated based on the child's age, sex, weight, etc., according to the Dietary Reference Intakes (DRIs). Micronutrient values ​​are the recommended daily intakes of various vitamins and minerals calculated based on the child's age, sex, etc., according to the DRIs. The basal nutrient value list is a list of all basal nutrient requirements calculated using standard formulas and authoritative data, without individual adjustments, serving as a scientific starting point for setting goals.

[0133] The terminal calculates the daily total energy requirement based on the child's height, weight, age, and gender using a recognized formula. It selects the corresponding activity coefficient based on the child's daily activity level and multiplies the basic energy consumption by the coefficient. Based on the child's age, gender, and other key information, it queries the built-in authoritative nutrient reference intake database to directly obtain or interpolate the baseline nutrient and micronutrient values. All the calculated and queried baseline values ​​are then compiled into a structured list of basic nutrient values.

[0134] Step 603: Based on the nutritional adjustment parameters, the basic nutritional value list is adaptively adjusted to obtain a draft of personalized nutritional goals.

[0135] Specifically, the draft of personalized nutrition goals is a preliminary personalized nutrition goal obtained by applying nutrition adjustment parameters to a basic nutrient value list and then performing mathematical calculations.

[0136] The terminal processes the values ​​in the basic nutrient value list item by item. For each item, it checks whether a corresponding nutrient adjustment parameter is provided. If an adjustment parameter exists, it performs the calculation according to the parameter instructions. For example, if the basic protein requirement is 50 grams and the adjustment parameter is a coefficient of 1.1, the adjusted target is 55 grams. If the adjustment parameter is a principle, it recalculates the target range of each macronutrient to make it conform to the new ratio requirements, while the total calories meet the adjusted energy requirements. After all values ​​are adjusted, a draft of personalized nutrient goals is output. The draft reflects the intervention intention for the child's specific health status.

[0137] Step 604: Perform a physiological range test on the draft personalized nutrition goals, obtain the physiological test results, and adjust the draft personalized nutrition goals based on the physiological test results to obtain the daily nutrition goal range.

[0138] Physiological range verification refers to the process of comparing the values ​​in the draft personalized nutrition goals with known, safe, and physiologically acceptable ranges for human health. These ranges typically have upper and lower limits to prevent nutritional deficiencies or excesses. The results of the physiological verification are used to identify which values ​​in the draft exceed the safe physiological range, and to what extent. The daily nutritional target range is a range of daily nutritional intake targets that, after physiological safety verification, meet both the needs of personalized intervention and ensure safety.

[0139] The terminal compares each nutrient target value in the draft with the preset physiological safety range of that nutrient, records all items that exceed the safety range, and generates a physiological test result report. For items that exceed the safety range, it adjusts them to the closest safety boundary value. After adjustment, it outputs the final daily nutrient target range, which is represented as an interval, providing some flexibility for recipe generation.

[0140] Based on children's basic physiological characteristics and universal scientific standards, this embodiment establishes an objective and fair baseline for nutritional needs, ensuring that personalized nutritional goals do not deviate from a safe track due to being too aggressive or negligent. The generated daily nutritional goal range is personalized, effective, safe, and reliable.

[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 system for optimizing the nutritional balance of children's recipes, which implements the above-mentioned method for optimizing the nutritional balance of children's recipes. The solution provided by this system is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the system for optimizing the nutritional balance of children's recipes provided below can be found in the limitations of the method for optimizing the nutritional balance of children's recipes described above, and will not be repeated here.

[0143] In one exemplary embodiment, such as Figure 2 As shown, a children's recipe nutritional balance optimization system 700 is provided, including:

[0144] The growth status module 701 is used to acquire children's growth status data and compare the children's growth status data with standard growth curve data to obtain standard growth indicators.

[0145] The health profile module 702 is used to input standard growth indicators into the classification and diagnostic machine learning model to assess the growth status type and generate children's health profile data.

[0146] Nutritional target module 703 is used to calculate the daily nutritional target range based on children's health profile data and children's growth status data; the daily nutritional target range includes the daily calorie target range, the daily macronutrient target range, and the daily key nutrient target range.

[0147] The recipe draft module 704 is used to load the ingredient database, match ingredients in the ingredient database that meet the daily nutritional target range, and generate a recipe draft.

[0148] The optimization and adjustment module 705 is used to identify food combinations with antagonistic effects from the recipe draft based on the nutrient interaction map, generate nutrient conflict result information, and optimize and adjust the recipe draft based on the nutrient conflict result information to obtain a nutritionally balanced optimized children's recipe.

[0149] Furthermore, the child's growth status data includes parameters of the child's absorption capacity, and the system also includes a mapping module for:

[0150] Based on the nutritional relationship data table, entity recognition is performed to identify nutritional entities, and the nutritional relationships between nutritional entities are extracted from the nutritional relationship data table. Among them, nutritional entities include food entities, nutrient entities and anti-nutrient entities, and nutritional relationships include biochemical relationships, physical relationships and conditional relationships.

[0151] Using nutrient entities as graph nodes and nutrient relationships as edges, a preliminary knowledge graph is constructed, and a logical consistency check is performed on the preliminary knowledge graph to obtain a general knowledge graph.

[0152] Physiological parameter vectors are constructed based on children's absorption capacity parameters, and then the general knowledge graph is personalized based on the physiological parameter vectors to obtain a nutrient interaction graph.

[0153] Furthermore, the atlas module is also used for:

[0154] Based on the physiological parameter vector, the regulatory rules are matched from the nutritional pathway regulation rule table to obtain the nutritional pathway regulation instructions;

[0155] Based on nutrient pathway regulation instructions, the basic weights of edges in a general knowledge graph are adjusted to obtain personalized weights.

[0156] Traverse the personalized weights, remove edges whose personalized weights are less than the weight threshold, obtain highly relevant edges, and perform necessity verification on the removed edges based on the highly relevant edges to obtain the core relationship set.

[0157] Based on the core relationship set, the original knowledge subgraph is extracted from the general knowledge graph;

[0158] Based on the daily nutritional target range, nodes in the original knowledge subgraph corresponding to key nutrients are defined as the organization core, resulting in a nutrient interaction map.

[0159] Furthermore, recipe draft module 704 is also used for:

[0160] Allocate the proportion of the daily nutrient target range to each meal to obtain the daily nutrient supply target plan;

[0161] Based on a food database, the objective function value is calculated with the goal of making the nutritional composition of the food in each meal combination as close as possible to the daily nutrient supply target plan.

[0162] The formula for the objective function value is:

[0163]

[0164] in, Let be the objective function, i be the nutrient index, j be the ingredient index, N be the nutrient set, and J be the ingredient set. Let be the weighting coefficient of the i-th nutrient. Let j be the amount of food ingredient of type j. Let be the nutrient content coefficient of the i-th nutrient in the j-th food group. A daily nutrient supply target plan for nutrient type i;

[0165] By injecting constraints into the objective function, an optimization proposition is obtained. Solving the optimization proposition yields a draft recipe.

[0166] Furthermore, the optimization and adjustment module 705 is also used for:

[0167] Based on the nutritional conflict results, solutions with corresponding antagonistic relationships are matched from the nutritional antagonism adjustment strategy library to obtain nutritional antagonism adjustment schemes;

[0168] Based on the nutritional antagonism adjustment plan, the draft recipe was adjusted to obtain the optimized recipe;

[0169] The optimized recipe is input into the acceptance prediction model to predict acceptance and obtain the predicted acceptance. The acceptance prediction model is trained using dietary preference data.

[0170] Based on the predicted acceptance, identify the dishes in the optimized recipe whose predicted acceptance is lower than the acceptance threshold, and obtain the dishes to be adjusted.

[0171] Based on dietary preference data, the meals to be adjusted were fine-tuned to obtain a nutritionally balanced and optimized children's menu.

[0172] Furthermore, the nutritional target module 703 is also used for:

[0173] Based on children's health profile data, the corresponding energy consumption and nutritional adjustment parameters for each nutrient can be obtained by querying the nutritional adjustment parameter comparison table.

[0174] Based on children's growth status data, basal energy expenditure, daily total energy requirement, nutrient baseline and micronutrient values ​​are calculated to obtain a list of basic nutrient values; among which, children's growth status data includes basic information and daily activity level;

[0175] Based on the nutritional adjustment parameters, the basic nutritional value list is adaptively adjusted to obtain a draft of personalized nutritional goals.

[0176] The draft personalized nutrition goals were subjected to a physiological range test to obtain the physiological test results. Based on the physiological test results, the draft personalized nutrition goals were adjusted to obtain the daily nutrition goal range.

[0177] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of the previously described method for optimizing the nutritional balance of a children's diet.

[0178] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0179] 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.

[0180] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A method for optimizing the nutritional balance of a child's diet, characterized in that, The method includes: Acquire children's growth status data and compare the children's growth status data with standard growth curve data to obtain standard growth indicators; The standard growth indicators are input into a classification diagnostic machine learning model to assess growth status and generate children's health profile data. Based on the child health profile data and the child growth status data, the daily nutritional target range is calculated; wherein, the daily nutritional target range includes the daily calorie target range, the daily macronutrient target range, and the daily key nutrient target range; Load the ingredient database, and match ingredients that meet the daily nutritional target range in the ingredient database to generate a recipe draft; Based on the nutrient interaction map, food combinations with antagonistic effects are identified from the draft recipe, nutrient conflict result information is generated, and the draft recipe is optimized and adjusted based on the nutrient conflict result information to obtain a nutritionally balanced optimized children's recipe.

2. The method of claim 1, wherein, The child growth status data includes parameters of the child's absorption capacity, and the nutrient interaction map is constructed using the following methods: Based on the nutritional relationship data table, entity recognition is performed to identify nutritional entities, and the nutritional relationships between the nutritional entities are extracted from the nutritional relationship data table; wherein, the nutritional entities include food entities, nutrient entities and anti-nutrient entities, and the nutritional relationships include biochemical relationships, physical relationships and condition-dependent relationships. Using the nutrient entities as graph nodes and the nutrient relationships as edges, a preliminary knowledge graph is constructed, and a logical consistency check is performed on the preliminary knowledge graph to obtain a general knowledge graph. A physiological parameter vector is constructed based on the child's absorption capacity parameters, and the general knowledge graph is personalized based on the physiological parameter vector to obtain the nutrient interaction graph.

3. The method of claim 2, wherein, The process of personalizing the general knowledge graph based on the physiological parameter vector to obtain the nutrient interaction graph includes: Based on the physiological parameter vector, regulation rules are matched from the nutrient pathway regulation rule table to obtain nutrient pathway regulation instructions; Based on the nutrient pathway regulation instructions, the basic weights of the edges in the general knowledge graph are adjusted to obtain personalized weights; Traverse the personalized weights, remove the edges whose personalized weights are less than the weight threshold to obtain highly relevant edges, and perform necessity verification on the removed edges based on the highly relevant edges to obtain the core relationship set. Based on the core relationship set, the original knowledge subgraph is extracted from the general knowledge graph; Based on the daily nutritional target range, the nodes in the original knowledge subgraph corresponding to the key nutrients are defined as the organization core, thus obtaining the nutritional interaction map.

4. The method of claim 1, wherein, The step of matching ingredients that meet the daily nutritional target range in the ingredient database and generating a recipe draft includes: The intake proportion of the daily nutrient target range is allocated to each meal to obtain the daily nutrient supply target plan; Based on the food database, with the goal of making the nutritional composition of the food in each meal combination as close as possible to the daily nutrient supply target plan, the objective function value is calculated. The formula for the objective function value is as follows: wherein, is the objective function, i is the nutrient index, j is the food material index, N is the nutrient set, J is the food material set, is the weight coefficient of the i-th nutrient, is the amount of use of the j-th food material, is the nutrient content coefficient of the i-th nutrient of the j-th food material, is the daily nutrient supply target plan of the i-th nutrient; By injecting constraints into the objective function, an optimization proposition is obtained, and by solving the optimization proposition, the draft recipe is obtained.

5. The method of claim 1, wherein, Based on the nutritional conflict results, the draft recipe is optimized and adjusted to obtain a nutritionally balanced optimized children's recipe, including: Based on the nutritional conflict results, solutions with corresponding antagonistic relationships are matched from the nutritional antagonism adjustment strategy library to obtain nutritional antagonism adjustment schemes; Based on the aforementioned nutrient antagonism adjustment scheme, the draft recipe was adjusted to obtain an optimized recipe; The optimized recipe is input into the acceptance prediction model to predict the acceptance level; the acceptance prediction model is trained using dietary preference data. Based on the predicted acceptance rate, identify the dishes in the optimized recipe whose predicted acceptance rate is lower than the acceptance threshold, and obtain the dishes to be adjusted. Based on the dietary preference data, the meals to be adjusted are fine-tuned to obtain the nutritionally balanced optimized children's diet.

6. The method of claim 1, wherein, The daily nutritional target range is calculated based on the child's health profile data and the child's growth status data, including: Based on the children's health profile data, the corresponding energy consumption and nutritional adjustment parameters for each nutrient can be obtained by querying the nutritional adjustment parameter comparison table. Based on the child's growth status data, the basic energy consumption, daily total energy requirement, nutrient baseline and trace element values ​​are calculated to obtain a list of basic nutrient values; wherein, the child's growth status data includes basic information and daily activity level; Based on the aforementioned nutritional adjustment parameters, the basic nutritional value list is adaptively adjusted to obtain a draft of personalized nutritional goals. The draft personalized nutrition goals are subjected to a physiological range test to obtain the physiological test results. Based on the physiological test results, the draft personalized nutrition goals are adjusted to obtain the range of daily nutrition goals.

7. A system for optimizing nutritional balance of children's recipes, characterized in that, The system includes: The growth status module is used to acquire children's growth status data and compare the children's growth status data with standard growth curve data to obtain standard growth indicators; The health profile module is used to input the standard growth indicators into the classification and diagnostic machine learning model to assess the growth status type and generate children's health profile data. The nutrition target module is used to calculate the daily nutrition target range based on the child health profile data and the child growth status data; wherein, the daily nutrition target range includes the daily calorie target range, the daily macronutrient target range, and the daily key nutrient target range; The recipe draft module is used to load the ingredient database, match ingredients in the ingredient database that meet the daily nutritional target range, and generate a recipe draft. The optimization and adjustment module is used to identify food combinations with antagonistic effects from the draft recipe based on the nutrient interaction map, generate nutrient conflict result information, and optimize and adjust the draft recipe based on the nutrient conflict result information to obtain a nutritionally balanced optimized children's recipe.

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

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

Images