Diet health assessment method and system, and computer equipment

Abstract

The invention provides a diet health assessment method and system, and computer equipment. The method comprises the following steps: quantifying diet assessment indexes into a first type of constraints and a second type of constraints; according to the first type of constraints and the second type of constraints, establishing a multi-objective optimization model; running the multi-objective optimization model on a predefined crowd cluster parameter set to obtain a crowd daily energy reference value and a crowd single meal food reference value; adjusting a second type of constraints based on the individual feature data to form individual constraint conditions; operating the multi-target optimization model under individual constraint conditions to obtain an individual daily energy target value and an individual single-meal food recommendation amount; performing quantitative comparison on the actual daily diet data, the individual daily energy target value, the individual single-meal food recommendation amount and the first type of constraints to generate a diet deviation result; and forming a diet adjustment suggestion according to the diet deviation result and outputting the diet adjustment suggestion to the user terminal. According to the method, diet adjustment suggestions are obtained by quantifying different types of constraints, so that precise evaluation of hyperlipidemia crowds in different scales is realized.

Description

Technical Field

[0001] This application relates to the field of dietary health analysis technology, and in particular to a dietary health assessment method and system, as well as computer equipment. Background Technology

[0002] With the high incidence of chronic diseases, the number of people with hyperlipidemia in my country is gradually increasing. Image- and text-based dietary assessment systems are widely used to record users' diets and provide nutritional suggestions. Existing solutions are mostly designed for healthy individuals and lack specific assessment capabilities for people with hyperlipidemia. They generally adopt a linear process of "taking a photo → recognition → comparison with fixed dietary guidelines → outputting qualitative suggestions." However, existing solutions do not set energy and nutrient boundaries for people with hyperlipidemia and are difficult to iteratively optimize based on real-time weight, blood lipids, and adherence feedback. This leads to a disconnect between reference values ​​and individual metabolism, and the assessment results remain vague suggestions such as "increasing or decreasing certain types of food." Users cannot quickly determine whether each meal is meeting their nutritional goals, resulting in insufficient personalization and efficiency. Summary of the Invention

[0003] This application provides a dietary health assessment method and system, as well as computer equipment. Through multi-objective optimization and reinforcement learning loops at two different scales—population clusters and individuals—it automatically generates daily energy and single-meal gram-level reference values ​​suitable for high blood lipids, achieving quantitative assessment and precise recommendations for each meal.

[0004] In a first aspect, embodiments of this application provide a dietary health assessment method, the method comprising: quantifying dietary assessment indicators into a first type of constraint and a second type of constraint, the first type of constraint including energy, nutrient energy ratio, and nutrient intake, the second type of constraint including the energy ratio of three meals, the intake of various foods, and food categories; establishing a multi-objective optimization model based on the first type of constraint and the second type of constraint, the multi-objective optimization model being used to optimize daily energy negative balance, dietary fiber intake, and the proportion of high micronutrient density foods; running the multi-objective optimization model on a predefined set of population cluster parameters to obtain a daily energy reference value and a single meal food reference value for the population, the population cluster parameter set including at least gender, age, height, weight, and B... The system uses MI (Minimum Intake), activity coefficient, and lipid metabolism-related indicators. Based on individual characteristic data collected from the user terminal, the second type of constraints are adjusted to form individual constraints. The individual characteristic data includes at least real-time weight, energy deficit target, food preferences, allergens, and lipid metabolism-related indicators. Under the individual constraints, the multi-objective optimization model is run to adjust the population's daily energy reference value and single-meal food reference value to obtain the individual's daily energy target value and recommended single-meal food quantity. The actual daily diet data uploaded by the user terminal is quantitatively compared with the individual's daily energy target value, recommended single-meal food quantity, and the first type of constraints to generate dietary deviation results. Based on the dietary deviation results, dietary adjustment suggestions are generated and output to the user terminal.

[0005] Secondly, embodiments of this application provide a dietary health assessment system, which includes a constraint quantification module, a model building module, a population solution module, an individual constraint module, an individual solution module, a deviation comparison module, and a suggestion output module. The constraint quantification module is used to quantify dietary assessment indicators into a first type of constraint and a second type of constraint. The first type of constraint includes energy, nutrient energy ratio, and nutrient intake. The second type of constraint includes the energy ratio of three meals, the intake of various foods, and food categories. The model building module is used to establish a multi-objective optimization model based on the first type of constraint and the second type of constraint. The multi-objective optimization model is used to optimize the daily energy negative balance, dietary fiber intake, and the proportion of high micronutrient density foods. The population solution module is used to run the multi-objective optimization model on a predefined population cluster parameter set to obtain the population's daily energy reference value and the population's single meal food reference value. The dataset includes at least gender, age, height, weight, BMI, activity level, and lipid metabolism-related indicators. The individual constraint module adjusts the second type of constraints to form individual constraint conditions based on individual characteristic data collected from the user terminal. The individual characteristic data includes at least real-time weight, energy deficit target, food preferences, allergens, and lipid metabolism-related indicators. The individual solution module runs the multi-objective optimization model under the individual constraint conditions to adjust the population's daily energy reference value and single-meal food reference value to obtain the individual's daily energy target value and recommended single-meal food quantity. The deviation comparison module quantitatively compares the actual daily dietary data uploaded by the user terminal with the individual's daily energy target value, recommended single-meal food quantity, and the first type of constraints to generate dietary deviation results. The suggestion output module generates dietary adjustment suggestions based on the dietary deviation results and outputs them to the user terminal.

[0006] Thirdly, embodiments of this application provide a computer device, the computer device including a memory and a processor, the memory being used to store a computer program; the processor being used to execute the computer program to implement the above-described dietary health assessment method.

[0007] The aforementioned dietary health assessment method and system, along with computer equipment, establishes a multi-objective optimization model by quantifying different types of constraints. It then sequentially runs the model to solve for the daily energy reference value and the single-meal food reference value of the population cluster. Individual constraints are adjusted using individual characteristic data, and reinforcement learning is used to iteratively scale the boundary to generate the individual's daily energy target value and recommended single-meal food quantity. Finally, the actual daily dietary data is quantified and compared to output the dietary deviation results, forming dietary adjustment suggestions. This enables accurate assessment of hyperlipidemia at both the population and individual scales. Attached Figure Description

[0008] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0009] Figure 1 A first flowchart of the dietary health assessment method provided in the embodiments of this application.

[0010] Figure 2 A flowchart of step S103 provided in the embodiments of this application.

[0011] Figure 3 A flowchart of step S1032 provided in the embodiments of this application.

[0012] Figure 4 A flowchart of sub-step S1033 provided for embodiments of this application.

[0013] Figure 5 A flowchart of step S105 provided in the embodiments of this application.

[0014] Figure 6 A flowchart of step S106 provided in the embodiments of this application.

[0015] Figure 7 The second flowchart is provided for the dietary health assessment method in the embodiments of this application.

[0016] Figure 8 This is a schematic diagram illustrating different types of constraints provided in the embodiments of this application.

[0017] Figure 9 This is a schematic diagram illustrating various foods and their nutrient content as provided in the embodiments of this application.

[0018] Figure 10 This is a schematic diagram illustrating the calculation of nutrient energy ratio and corresponding weight for embodiments of this application.

[0019] Figure 11 This is a schematic diagram illustrating the reference values ​​for single-meal food intake for a population and the constraint fulfillment status provided in the embodiments of this application.

[0020] Figure 12 A schematic diagram of individual characteristic data provided in an embodiment of this application.

[0021] Figure 13 This is a schematic diagram illustrating the recommended amount and nutritional components of an individual lunch provided in an embodiment of this application.

[0022] Figure 14 This is a schematic diagram illustrating the recommended amount of individual lunch food and the assessment of the first type of constraint provided in an embodiment of this application.

[0023] Figure 15 This is a schematic diagram illustrating the dietary deviation results and dietary adjustment suggestions provided in the embodiments of this application.

[0024] Figure 16 This is a structural block diagram of the dietary health assessment system provided in the embodiments of this application.

[0025] Figure 17 This is a schematic diagram of the internal structure of a computer device for applying a dietary health assessment method, as provided in an embodiment of this application.

[0026] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0027] 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. All other embodiments obtained by those skilled in the art based on the embodiments in this application without inventive effort are within the scope of protection of this application.

[0028] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar planned objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data are interchangeable where appropriate; in other words, the described embodiments are implemented according to a sequence other than that illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, may also include other content; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0029] It should be noted that the use of terms such as "first" and "second" in this application is for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. Furthermore, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of a person skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed in this application.

[0030] Please refer to Figure 1 This is a first flowchart of the dietary health assessment method provided in this application embodiment. This application provides a dietary health assessment method. This method is applicable to individuals with high blood lipids. Through multi-objective optimization and reinforcement learning loops at two different levels—population clusters and individuals—it automatically generates single-meal reference values ​​and outputs quantitative deviations and personalized adjustment suggestions based on actual intake data. The dietary health assessment method includes steps S101-S107.

[0031] Step S101: Quantify the dietary assessment indicators into first-type constraints and second-type constraints.

[0032] In step S101, the first type of constraint is a rigid constraint that must be strictly met, including energy, nutrient energy ratio, and nutrient intake. The second type of constraint is a flexible constraint that can be adjusted according to individual characteristics, including the energy ratio of three meals, the intake of various foods, and food categories. The dietary assessment indicators in this application can be derived from guidelines such as the "Dietary Guidelines for Adult Hyperlipidemia 2023," the "Nutrition and Exercise Guidelines for Hyperlipidemia 2024," and the "Chinese Dietary Guidelines 2022," and are integrated with clinical nutritional epidemiological evidence to ensure that the boundary values ​​corresponding to each constraint are both authoritative and applicable to the population. The meaning of each constraint in the different types of constraints will be briefly introduced below.

[0033] In the first type of constraint, energy is based on the total daily energy expenditure (TDEE) calculated as "Basal Metabolic Rate (BMR) × Activity Level (PAL)," with corresponding constraints for different population groups. For example, an upper limit of 300-500 kcal daily energy deficit is set for overweight / obese individuals, with men not less than 1200 kcal and women not less than 1000 kcal. The nutrient energy ratio directly determines the proportion of each nutrient's contribution to total energy, used to quickly determine whether the dietary structure deviates from the target for hyperlipidemia control. For example, the AMDR for carbohydrates is "≥50%E", the AMDR for fat is "20-25%E", and the AMDR for protein is "10-30%E". Among them, AMDR (Acceptable Macronutrient Distribution Range) represents the acceptable distribution range of macronutrients, referring to the recommended range of the proportion of a certain macronutrient in total energy intake, expressed as a percentage of energy (%E), used to guide dietary structure without limiting absolute weight. Accordingly, a carbohydrate AMDR of "≥50%E" indicates that carbohydrates should provide 50% or more of the total daily energy intake. Nutrient intake is the absolute value in grams calculated based on the nutrient energy ratio using an energy coefficient, used to establish weighable and verifiable boundary ranges for each constraint.

[0034] In the second type of constraint, the energy ratio of the three meals can be used to correct postprandial blood lipid fluctuations caused by uneven meal distribution. For example, the energy ratio of the three meals can be "30-35% for breakfast; 30-40% for lunch; and 30-35% for dinner." All food intake is expressed as net edible weight, ensuring users can directly weigh and execute the calculation. For example, the intake of various foods can include, but is not limited to, "300-500g of vegetables," "120-200g of meat," "200-350g of fruit," "25-35g of soybeans and nuts," and "≥25g of soy products." Food categories are used to ensure micronutrient density and gut microbiota diversity, while also preventing nutrient deficiency or excess caused by single ingredients at different times by setting minimum intake levels. For example, food categories can require ≥12 different types of food per day and ≥25 different types per week.

[0035] By quantifying the above indicators, a set of calculable, iterative, and comparable constraints is provided for subsequent multi-objective optimization models, realizing the transformation from qualitative dietary guidelines to quantitative boundary inputs.

[0036] Step S102: Establish a multi-objective optimization model based on the first type of constraints and the second type of constraints.

[0037] In step S102, the multi-objective optimization model is trained by feeding the sample dataset, first-type constraints, and second-type constraints into the initial training model. Specifically, the initial training model can be constructed using the NSGA-II framework, with a decision variable vector containing the weight of food at each meal and the daily energy value as the solution vector for the initial training model's optimization search. The first-type constraint is used as a hard penalty (i.e., violation results in elimination), and the second-type constraint is used as a soft penalty (i.e., weighted deduction). After feeding in the sample dataset, a preset number of non-dominated sorts are performed to generate corresponding frontier solutions, forming a frontier solution set. Subsequently, gradient descent local search with different step lengths is applied to each frontier solution. When the rate of change of the objective function in consecutive iterations is less than a preset threshold, convergence is determined and the search is terminated, thus obtaining the final optimization parameters. The final optimization parameters are output in the form of a weight matrix for subsequent population cluster or individual solutions.

[0038] A multi-objective optimization model is used to optimize daily negative energy balance, dietary fiber intake, and the proportion of high micronutrient density foods. The sample dataset includes historical dietary records of different populations and their corresponding lipid metabolism-related indicators, energy expenditure values, and nutrient intake values. The historical dietary records can be retrospective data of different populations at different time periods (e.g., within one day, within one week), including but not limited to food photos, weights, and meal times. This retrospective data can be pre-cleaned through image recognition and manual verification. Lipid metabolism-related indicators include, but are not limited to, TG (triglycerides), TC (total cholesterol), LDL-C (low-density lipoprotein cholesterol), and HDL-C (high-density lipoprotein cholesterol) for the same population. Energy expenditure value is the total daily energy expenditure. Nutrient intake values ​​can be the absolute amounts of macronutrients, dietary fiber, cholesterol, and micronutrients aligned with different time periods in the historical dietary records. The daily negative energy balance is used in the multi-objective optimization model to maximize the difference between total daily energy expenditure and daily energy intake, and this difference is not lower than the boundary values ​​of each constraint in the first type of constraint, adapting to weight loss requirements. Dietary fiber intake is maximized within the dietary fiber boundary given by the first type of constraint to promote lipid metabolism and increase satiety. The proportion of high micronutrient density foods aims to maximize the weighted value of the "micronutrient score to energy value" ratio, prioritizing dark green vegetables, whole grains, and nuts, reducing empty energy foods, and increasing dietary micronutrient density. This application uses a multi-objective optimization model that learns the mapping patterns between lipid metabolism-related indicators and diet from historical data while preserving the constraint boundaries, providing an interpretable and iterative optimization model for subsequent population cluster solutions and individual scaling.

[0039] Step S103: Run a multi-objective optimization model on the predefined population cluster parameter set to obtain the population's daily energy reference value and the population's single meal food reference value. The population cluster parameter set includes at least gender, age, height, weight, BMI, activity coefficient, and lipid metabolism-related indicators.

[0040] In step S103, the population cluster parameter set may also include blood pressure classification, liver and kidney function, smoking status, etc. The activity coefficient can be divided into corresponding parameters according to the daily activity intensity. For example, when working in the office, screen time ≥8h, or steps <5000, the activity coefficient is determined to be sedentary (1.20); when walking 5000-7499 steps per day, or equivalent moderate-intensity activity <150min / week, the activity coefficient is determined to be mild (1.375)... and so on, different activity coefficients are further divided.

[0041] Please refer to Figure 2 The flowchart below shows step S103, which is provided in the application embodiment. Steps S1031-S1034 include running a multi-objective optimization model on a predefined set of population cluster parameters to obtain the population's daily energy reference value and the population's single meal food reference value.

[0042] Step S1031: Calculate the basal metabolic rate and total daily energy consumption based on the population cluster parameter set.

[0043] In step S1031, the basal metabolic rate (BMR) can be calculated using the Mifflin-St Jeor formula. Specifically, the formula for men is: BMR = 10 × weight (kg) + 6.25 × height (cm) - 5 × age (years) + 5; the formula for women is: BMR = 10 × weight (kg) + 6.25 × height (cm) 5 × Age (years) - 161. The formula for Total Daily Energy Expenditure (TDEE) is: BMR × Activity Level (PAL).

[0044] Furthermore, if the population cluster parameter set indicates overweight / obese and the target is weight loss, then a pre-decrease of a preset energy value is used as the initial energy gap in the TDEE, but it is guaranteed not to be lower than the minimum energy boundary for gender under the first type of constraint (male ≥ 1200 kcal, female ≥ 1000 kcal). For example, the preset energy value can be 300 kcal.

[0045] Step S1032: Based on the basal metabolic rate and daily total energy consumption, a preset multi-objective genetic algorithm is used to perform a global search on the multi-objective optimization model to generate an initial non-dominated solution set.

[0046] In step S1032, the preset multi-objective genetic algorithm can be a non-dominated sorting genetic algorithm, such as the NSGA-II algorithm. The non-dominated solution set is the Pareto front solution set.

[0047] Please refer to Figure 3 This is a flowchart of step S1032, a sub-step provided in the embodiments of this application. Based on the basal metabolic rate and total daily energy consumption, a preset multi-objective genetic algorithm is used to perform a global search on the multi-objective optimization model to generate an initial non-dominated solution set, including steps S10321-S10322.

[0048] Step S10321: Taking maximizing dietary fiber intake as the first optimization objective, perform a first non-dominated sorting on the multi-objective optimization model to obtain the first non-dominated solution set.

[0049] In step S10321, the first non-dominated ranking of the multi-objective optimization model can be performed according to different meals. Taking lunch as an example, the weight of each food item in the lunch menu is used as the decision variable, and maximizing the total grams of dietary fiber is set as the first optimization objective. At the same time, the first type of constraint (energy, fat, cholesterol, etc.) is used as a hard penalty term. Through rapid non-dominated ranking, the corresponding frontier solutions are selected, that is, the individuals with the highest dietary fiber that are not dominated by other solutions, forming the first non-dominated solution set.

[0050] Step S10322: Within the first non-dominated solution set, with maximizing the proportion of foods with high micronutrient density as the second optimization objective, a second non-dominated sort is performed to form an initial non-dominated solution set.

[0051] In step S10322, within the first non-dominated solution set, the second optimization objective remains maximizing the proportion of foods with high micronutrient density. A non-dominated ranking is performed again on the first non-dominated solution set to comprehensively consider various micronutrients related to lipid metabolism (such as vitamins A, C, and E, as well as elements like calcium, magnesium, iron, zinc, and selenium), and after standardization, weighted by energy. Similarly, only the corresponding frontier solutions are retained, maximizing the micronutrients provided per unit of energy without reducing dietary fiber.

[0052] Step S1033: Perform gradient descent algorithm on the initial non-dominated solution set until convergence, and obtain the optimal solution that satisfies the first type of constraint. The optimal solution includes the daily energy value and the reference weights corresponding to various types of food.

[0053] Please refer to Figure 4 The flowchart below shows step S1033, which is provided in the embodiment of this application. Steps S10331-S10333 involve performing a gradient descent algorithm on the initial non-dominated solution set until convergence, to obtain the optimal solution that satisfies the first type of constraint.

[0054] Step S10331: Set the corresponding gradient descent step size according to the food category.

[0055] In step S10331, the gradient descent step size can be set according to the sensitivity of different food categories. The gradient descent step size for different food categories can be the same or different. For example, for highly sensitive foods (such as nuts, oils, etc.), the gradient descent step size can be set to 1 g; for moderately sensitive foods (such as meat, beans, staple foods, etc.), the gradient descent step size can be set to 3 g; and for low-sensitive foods (such as vegetables, fruits, etc.), the gradient descent step size can be set to 5 g, so as to avoid exceeding the limit with slight adjustments while ensuring sufficient search efficiency.

[0056] Step S10332: Iteratively update the weight of various foods along the preset gradient direction with the gradient descent step size to obtain a converged solution until the change in the converged solution of the multi-objective optimization model during the iterative update is less than a preset threshold, then the gradient descent algorithm is determined to have converged.

[0057] In step S10332, the weights of various foods are updated along the preset gradient direction with the composite objective function of maximizing dietary fiber, maximizing micronutrient density, and minimizing energy overrun penalty. After each iteration, the violation degree of the first type of constraint and the composite objective function value are recalculated.

[0058] Step S10333: Output the current converged solution as the optimal solution.

[0059] In step S10333, when the rate of change of the composite objective function value is less than the preset frequency in a series of preset number of iterations and all first-type constraints are satisfied, the search is terminated and the current solution is output as the optimal solution.

[0060] Step S1034: Set the daily energy value as the reference value for the population's daily energy, and set the reference weights corresponding to various foods as the reference values ​​for the population's single meal foods.

[0061] In step S1034, the daily energy value in the optimal solution is used as the daily energy reference value for the population. For various types of food, based on the energy ratio constraints of the three meals (breakfast 30-35%, lunch 30-40%, dinner 30-35%), and combined with the energy density of each type of food, the reference value for a single meal is calculated and output. It is understandable that if the weight of a certain type of food is lower than the corresponding gradient descent step size (e.g., nuts <1g), then the weight can be rounded down according to the lower limit of weighing, and the corresponding gradient descent step size can be determined as the minimum weighing unit to ensure the executability of the user terminal.

[0062] Taking a group of people with high blood lipids (female, 40 years old, height 165cm, BMI 30, activity coefficient 1.375) as an example, their basal metabolic rate (BMR) is 1498kcal, and their daily total energy expenditure (TDEE) is 2061kcal. Setting an energy deficit target of 500kcal, the daily energy reference value for this group, obtained through step S103, is 1561kcal (2061-500=1561). The corresponding recommended single-meal intake for this group is as follows: spinach 220g, oats 38g, chicken breast 90g, milk 260g, apple 140g, almonds 7g, and firm tofu 45g. This group's recommended single-meal intake satisfies all the first type of constraints, and the dietary fiber intake reaches 8.2g, making it the optimal solution under the current constraints.

[0063] Step S104: Based on the individual characteristic data collected by the user terminal, adjust the second type of constraints to form individual constraint conditions. The individual characteristic data includes at least real-time weight, energy deficit target, food preference, allergens and lipid metabolism-related indicators.

[0064] In step S104, the user terminal can be a smartphone, smartwatch, smart speaker with camera, smart plate with weighing function, etc. The user terminal can complete food image recognition and weight estimation through camera and LLM visual model, and can synchronize in real time with kitchen scale, wearable body fat bracelet, etc. through Bluetooth. It can also collect the user's real-time satiety, taste preferences, and allergy prompts through voice interaction. This data is parsed into individual characteristic data such as real-time weight, energy deficit target, food preferences, allergens, and lipid metabolism-related indicators, and then sent back to the multi-objective optimization model. Individual characteristic data may also include gut microbiota typing, blood glucose monitoring data, economic price range, work and rest schedule, etc., in order to form individual constraints that can be maintained in the long term.

[0065] Step S105: Run a multi-objective optimization model under individual constraints to adjust the daily energy reference value and the single meal food reference value of the population to obtain the individual daily energy target value and the individual single meal food recommendation.

[0066] Please refer to Figure 5 The flowchart below shows step S105, which is provided in the embodiment of this application. Running a multi-objective optimization model under individual constraints to adjust the population's daily energy reference value and the population's single-meal food reference value to obtain the individual's daily energy target value and individual single-meal food recommendation amount includes steps S1051-S1054.

[0067] Step S1051: Map individualized constraints to boundary parameters of a multi-objective optimization model.

[0068] In step S1051, the individual feature data is quantized into programmable boundary parameters. For example, piecewise linear scaling can be applied to the intake of various foods in the second type of constraint. When the economic price range is "low," the constraint boundary corresponding to nuts is linearly compressed from the upper limit of 35g to 25g. An enumeration mapping can be used for the energy ratio of the three meals. If the user's work schedule is consistently night shift, the constraint boundary corresponding to the dinner ratio is relaxed from the upper limit of 35% to 40%, and correspondingly, the constraint boundary corresponding to the breakfast ratio is reduced from 30% to 25%. A blacklist hard truncation can be used for food categories. If the allergen field contains "crustaceans," the "fish, shrimp, and crab" dimension is directly removed from the decision variable vector, and the boundary parameters of the corresponding constraint boundary are set to 0.

[0069] Step S1052: A preset multi-objective reinforcement learning algorithm is used to iteratively scale the boundary parameters using a composite reward function consisting of user compliance reward and lipid metabolism risk penalty.

[0070] In step S1052, a preset multi-objective reinforcement learning algorithm is used to dynamically scale the boundary parameters. Specifically, the constraint boundary corresponding to the current second type of constraint is used as the state space of the reinforcement learning. Each executable action is to apply a fine-tuning coefficient (e.g., ±5%) to any dimension of the constraint boundary, and to calculate the composite reward function value based on the candidate solutions generated after execution. The policy network is then updated through online exploration. To balance the two objectives of "users' willingness to persist" and "minimizing the risk of high cholesterol," the composite reward function is designed as follows: R_t = 0.6 × Compliance_Reward 0.4 × Metabolic_Risk Among them, Compliance_Reward is taken from the average overlap between the user's actual intake and the recommended intake over a historical time period (e.g., 7 days), ranging from 0 to 1, with higher overlap resulting in greater rewards; Metabolic_Risk represents the predicted probability of exceeding blood lipid limits in a future time period (e.g., 30 days), ranging from 0 to 1, with higher probability resulting in greater penalties. The preset multi-objective reinforcement learning algorithm uses online exploration to automatically find the optimal balance between "high compliance" and "low risk" for the boundary parameters, achieving adaptive iterative scaling of individual constraints.

[0071] Step S1053: After each iteration of scaling, run the multi-objective optimization model to generate corresponding candidate solutions and obtain a candidate solution set. Each candidate solution includes candidate values ​​for daily energy and candidate values ​​for food weight.

[0072] In step S1053, whenever the boundary parameters complete one reinforcement learning scaling, the NSGA-II algorithm is used to set the corresponding population size and number of iterations. The edible weight of each food category on the day is used as the decision variable to quickly output each candidate solution of the approximate frontier solution. Then, the daily energy candidate values ​​are discretized at preset intervals, and the food weight candidate values ​​retain the accuracy of the smallest weighing unit to form a candidate solution set.

[0073] Step S1054: Select candidate solutions from the candidate solution set that simultaneously satisfy the first type of constraint and have the largest composite reward function, and set the corresponding daily energy candidate value and food weight candidate value as the individual daily energy target value and the individual single meal food recommendation amount, respectively.

[0074] In step S1054, hard filtering is first performed on the candidate solution set to remove any solutions that violate the first type of constraint; the remaining solutions are sorted in descending order of the composite reward function, and the candidate solution with the largest composite reward function is selected.

[0075] For example, based on individual characteristic data collected from the user's terminal (the user's energy deficit target was adjusted from 500kcal at the population level to 300kcal), after optimization in step S105, the individual's daily energy target value was obtained as 1761kcal, corresponding to the following individual single-meal target dietary intake: spinach 210g, oatmeal 58g, chicken breast 85g, milk 250g, apple 170g, almond 5g, and firm tofu 53g. Compared with the population single-meal reference dietary intake, the individual optimization result was dynamically adjusted according to the user's personalized needs. Specifically, the intake of oatmeal and apple was increased to supplement carbohydrates and dietary fiber, while the intake of almonds was decreased to control fat, reflecting the fine-tuning process from population reference to different individual targets.

[0076] Furthermore, if multiple candidate solutions with the largest composite reward function appear, the dietary fiber content is further compared, and the solution with higher dietary fiber content is output first to enhance the intervention effect on blood lipid metabolism.

[0077] Step S106: The actual daily diet data uploaded by the user terminal is quantitatively compared with the individual's daily energy target value, the individual's recommended food quantity per meal, and the first type of constraint to generate a diet deviation result.

[0078] In step S106, the actual daily dietary data includes the actual total daily energy, the actual nutrient intake, and the weight of various foods in a single meal.

[0079] Please refer to Figure 6 The flowchart below shows step S106, which is a sub-step provided in the embodiment of this application. Steps S1061-S1064 involve quantitatively comparing the actual daily diet data uploaded by the user terminal with the individual's daily energy target value, the individual's recommended food quantity per meal, and the first type of constraint to generate a diet deviation result.

[0080] Step S1061: Calculate the energy deviation between the actual daily total energy and the individual daily energy target value.

[0081] In step S1061, the relative deviation between the actual total daily energy and the individual's daily energy target value is the core. Extremely high or low values ​​are first symmetrically processed, and then the exercise consumption equivalent actively reported by the user is automatically removed to ensure that the evaluation results only reflect whether the diet itself is excessive or insufficient, and to avoid misjudging as exceeding the standard after exercise.

[0082] Step S1062: Calculate the nutrient deviation between the actual nutrient intake and the first type of constraint.

[0083] In step S1062, the actual intake of each nutrient is standardized and aligned with the corresponding first-type constraint. A one-way penalty is applied to the one-way boundary indicator, and a two-way balance is applied to the two-way interval indicator. Finally, the comprehensive nutritional imbalance is summarized to obtain the nutrient deviation, so that any significant deviation of a single nutrient can be amplified and highlighted in the total deviation in real time.

[0084] Step S1063: Calculate the weight deviation between the weight of each type of food in a single meal and the recommended amount of food for the individual meal.

[0085] In step S1063, weights are automatically assigned based on the differences in the sensitivity of food categories to the impact on blood lipids, and a penalty coefficient is added to the averse ingredients marked in the user's individual characteristic data. This further considers feasibility beyond numerical errors and reduces the risk of the plan failing due to taste resistance.

[0086] Step S1064: Assign preset weights to the energy deviation, nutrient deviation and weight deviation respectively, and sum them up to obtain the dietary deviation result.

[0087] In step S1064, each type of deviation corresponds to a preset weight. The three types of deviations—energy, nutrients, and weight—are merged into a single score. The preset weights adaptively drift according to the user's recent compliance performance to output a gradient evaluation, which is then presented to the user terminal in real time and used for subsequent algorithm iterations.

[0088] Step S107: Based on the dietary deviation results, dietary adjustment suggestions are generated and output to the user terminal.

[0089] In step S107, the dietary adjustment suggestion is based on the dietary deviation results, and outputs specific operational instructions for the current day or the next meal to the user in the form of weight values, a list of food substitutes, and meal allocation prompts, so as to quickly guide the actual intake to the individual. The following will describe how to implement the dietary health assessment method with specific embodiments and accompanying drawings.

[0090] like Figure 8 As shown, Figure 8 The first and second types of constraints are displayed hierarchically, with each constraint's boundary shown separately (e.g., breakfast energy percentage 30-35%). This makes it easy to identify which indicators must be met (e.g., energy, fat, cholesterol) and which are recommended (e.g., meal ratio, food type). During subsequent deviation calculations, the user terminal can provide different visualizations for different types of constraints, enabling a visual conversion between qualitative and quantitative analysis.

[0091] like Figure 9 As shown, the elemental content of each food category in a 100g edible portion is first established: energy, carbohydrates, protein, fat, dietary fiber, and cholesterol, forming a visualized nutrition database. Then, users can upload photos of each food and input their weight values ​​through their user terminals. The corresponding food IDs are then called from the visual model, and the elemental content is converted into actual intake values ​​using "weight value ÷ 100 g" as a linear scaling factor, completing the real-time conversion from "image to nutrients" and providing standardized input for subsequent multi-objective optimization models.

[0092] When performing gradient descent on the initial non-dominated solution set until convergence, and obtaining the optimal solution satisfying the first type of constraint, it can be done according to... Figure 10 This illustrates the conversion of different nutrients into their corresponding weights. Taking fat as an example, when fat intake reaches 20% energy (E), the corresponding energy is 1561 × 20% = 312.2 (kcal), and the corresponding weight is 312.2 ÷ 9 = 34.7 (g). The energy and corresponding weight of different nutrients are updated in real time according to changes in different population groups or individuals, allowing users to directly perceive the correlation between percentages, energy, and corresponding weight values, thus increasing their willingness to participate.

[0093] Take lunch as an example. Figure 11 The left side lists the recommended weights for seven food categories, and the right side indicates whether each food category meets the first type of constraint. Specifically, the energy, fat, and cholesterol rows use a "hard pass" logic, meaning that if any one of them is not met, the entire category is not met. The dietary fiber and micronutrient density rows use a "soft bonus" logic, meaning that rewards can be increased in subsequent reinforcement learning. This is understandable when the user terminal displays... Figure 11 When the reference values ​​and constraints for a single meal are shown in the diagram, each food name represents only one type of food and its corresponding elemental content for the current food category. This diagram is displayed on the terminal as a left-right sliding card, allowing users to replace any currently displayed food with another food from the same category. For example, if a user slides the food corresponding to "Meat," the previously recommended 90g chicken breast instantly switches to "90g boneless duck breast." The system backend immediately reads the elemental content of the duck breast and re-verifies the energy, fat, and cholesterol rows, while simultaneously updating the reinforcement learning reward value.

[0094] During steps S104-S105, the collected individual characteristic data can be... Figure 12 The information is categorized into the following forms: basic information, lifestyle information, health goal information, dietary preference information, and disease-related information. For each category, at least the following data should be compiled: real-time weight, energy deficit target, food preferences, allergens, and lipid metabolism-related indicators.

[0095] After obtaining the individual's daily energy target value and the recommended food intake per meal, a system can be formed as follows: Figure 13 The individual lunch food recommendations and nutritional components are shown. Specifically, Figure 13 By using a dual-column comparison of population reference weight and individual recommended weight, users can easily see the fine-tuning of different foods and perceive the differences between individuals.

[0096] When quantitatively comparing the actual daily dietary data uploaded by the user terminal with the individual's daily energy target value, the individual's recommended food quantity per meal, and the first type of constraint, the following methods can be used: Figure 14 The chart shows a comparison of whether different constraints such as energy, fat, and cholesterol are met. It's understandable that when the gradient descent step size for carbohydrates is 3g and used as the smallest weighing unit, since 89.3-88=1.3<3, the carbohydrate column in the calculation of individual meal food recommendations can be considered "compliant".

[0097] like Figure 15 As shown, when the actual dietary data for lunch is: 100g spinach, 50g oatmeal, 50g chicken breast, 180g milk, 50g apple, 15g almonds, and 80g firm tofu, the calculated individual recommended weights are: "210g spinach, 58g oatmeal, 85g chicken breast, 250g milk, 170g apple, 5g almonds, and 53g firm tofu". First, the deviations of different foods are visualized (e.g., too little requires increasing "↑" or too much requires decreasing "↓"). Then, the following suggestions are output using the LLM language model: 1. Dietary fiber supplementation recommendations: Increase the proportion of high-fiber foods in your diet. Adjust the actual intake of spinach from 100g to 210g per meal and apples from 50g to 170g. By increasing the proportion of vegetables and fruits in your diet, you can make up for the dietary fiber deficiency and strengthen its intervention effect on blood lipids. 2. Nutritional Balance Recommendations: Moderately increase the intake of chicken breast (50g → 85g) and milk (180g → 250g) to ensure a supply of high-quality protein while controlling fat and carbohydrate intake; reduce the intake of firm tofu (80g → 53g) to achieve a precise and balanced ratio of various nutrients and improve the matching degree between the diet plan and the high blood lipid intervention target; at the same time, reduce the intake of almonds (15g → 5g) to control fat intake to meet the requirements.

[0098] Please refer to Figure 7 This is a second flowchart of the dietary health assessment method provided in the embodiments of this application. After generating dietary adjustment suggestions and outputting them to the user terminal, the dietary health assessment method further includes steps S201-S202.

[0099] Step S201: Collect user confirmation information on the implementation of dietary adjustment suggestions through the user terminal.

[0100] Step S202: The execution confirmation information is sent back to the multi-objective optimization model to update the individual constraints.

[0101] Please refer to Figure 16 This is a structural block diagram of the dietary health assessment system provided in the embodiments of this application. This application also provides a dietary health assessment system 11. The dietary health assessment system 11 includes a constraint quantification module 110, a model building module 111, a population solution module 112, an individual constraint module 113, an individual solution module 114, a deviation comparison module 115, and a suggestion output module 116.

[0102] The constraint quantification module 110 is used to quantify dietary assessment indicators into first-class constraints and second-class constraints. The first-class constraints include energy, nutrient energy ratio and nutrient intake, while the second-class constraints include the energy ratio of three meals, the intake of various foods and food categories.

[0103] The model building module 111 is used to build a multi-objective optimization model based on the first type of constraints and the second type of constraints. The multi-objective optimization model is used to optimize the daily negative energy balance, dietary fiber intake and the proportion of high micronutrient density foods.

[0104] The population solution module 112 is used to run a multi-objective optimization model on a predefined set of population cluster parameters to obtain the daily energy reference value and the single meal food reference value of the population. The population cluster parameter set includes at least gender, age, height, weight, BMI, activity coefficient and lipid metabolism-related indicators.

[0105] The individual constraint module 113 is used to adjust the second type of constraint to form individual constraint conditions based on the individual characteristic data collected by the user terminal. The individual characteristic data includes at least real-time weight, energy deficit target, food preference, allergens and lipid metabolism-related indicators.

[0106] The individual solver module 114 is used to run a multi-objective optimization model under individual constraints to adjust the daily energy reference value and the single meal food reference value of the population to obtain the individual daily energy target value and the individual single meal food recommendation.

[0107] The deviation comparison module 115 is used to quantitatively compare the actual daily diet data uploaded by the user terminal with the individual's daily energy target value, the individual's recommended food amount per meal, and the first type of constraint to generate diet deviation results.

[0108] The suggested output module 116 is used to generate dietary adjustment suggestions based on the dietary deviation results and output them to the user terminal.

[0109] Please refer to Figure 17 This is a schematic diagram of the internal structure of a computer device for applying a dietary health assessment method, as provided in the embodiments of this application.

[0110] like Figure 17 As shown, the computer device 100 includes a memory 901 and a processor 902. The processor 902 is used to run computer program instructions stored in the memory 901 to implement a dietary health assessment method.

[0111] The memory 901 includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 901 can be an internal storage unit of a computer device, such as a hard disk. In other embodiments, the memory 901 can be an external storage device of a computer device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., configured in the computer device. Furthermore, the memory 901 can include both internal and external storage units of a computer device. The memory 901 can be used not only to store application software and various types of data installed on the computer device, such as code for dietary health assessment methods, but also to temporarily store data that has been output or will be output.

[0112] Furthermore, the computer device 100 also includes a bus 903. The bus 903 can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. This bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 17 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0113] Furthermore, the computer device 100 may also include a display component 904. The display component 904 may be an LED display, a liquid crystal display, a touch-screen liquid crystal display, or an organic light-emitting diode (OLED) touchscreen, etc. The display component 904 may also be appropriately referred to as a display device or display unit, used to display information processed in the computer device 100 and to display a visual user interface.

[0114] Furthermore, the computer device 100 may also include a communication component 905. The communication component 905 may optionally include a wired communication component and / or a wireless communication component (such as a Wi-Fi communication component, a Bluetooth communication component, etc.), which is typically used to establish a communication connection between the computer device 100 and other computer devices.

[0115] Figure 17 Only a partial computer device 100 with components and implementing dietary health assessment methods is shown; those skilled in the art will understand that... Figure 17 The structure shown does not constitute a limitation on the computer device 100 and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0116] In the above embodiments, the implementation can be achieved, in whole or in part, through software, hardware, or any combination thereof. When implemented using software, it can be implemented, in whole or in part, in the form of a computer program product.

[0117] This computer program product includes one or more computer instructions. When these computer program instructions are loaded and executed on a computer, all or part of the flow or function according to embodiments of the present invention is generated. The computer device may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a server or data center that integrates one or more available media. The available medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., a solid-state disk (SSD)).

[0118] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes: USB flash drives, portable hard disks, read-only storage media (ROM), random access storage media (RAM), magnetic disks, optical disks, and other media capable of storing program code.

[0119] In the above embodiments, a multi-objective optimization model is established by quantifying different types of constraints. The model is then run sequentially to solve for the daily energy reference value and the single-meal food reference value of the population. Individual constraints are adjusted using individual characteristic data, and reinforcement learning is used to iteratively scale the boundary to generate the individual's daily energy target value and recommended single-meal food quantity. The actual daily diet data is then quantified and compared to output the diet deviation results, forming diet adjustment suggestions. This achieves accurate assessment of hyperlipidemia at both the population and individual scales.

[0120] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

[0121] It should be understood that although the steps in the flowcharts of the accompanying figures are shown sequentially as indicated by 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 accompanying figures may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0122] The above-listed embodiments are merely preferred embodiments of this application and should not be construed as limiting the scope of this application. Therefore, any equivalent variations made in accordance with the claims of this application shall still fall within the scope of this application.

Claims

1. A method for assessing dietary health, characterized in that, The dietary health assessment methods include: Dietary assessment indicators are quantified into first-type constraints and second-type constraints. The first-type constraints include energy, nutrient energy ratio, and nutrient intake, while the second-type constraints include the energy ratio of three meals, the intake of various foods, and food categories. Based on the first type of constraint and the second type of constraint, a multi-objective optimization model is established. The multi-objective optimization model is used to optimize the daily energy negative balance, dietary fiber intake and the proportion of high micronutrient density foods. The multi-objective optimization model is run on a predefined set of population cluster parameters to obtain the daily energy reference value and the single meal food reference value of the population. The set of population cluster parameters includes at least gender, age, height, weight, BMI, activity coefficient and lipid metabolism-related indicators. Based on individual characteristic data collected from user terminals, the second type of constraints are adjusted to form individual constraint conditions. The individual characteristic data includes at least real-time weight, energy deficit target, food preferences, allergens, and lipid metabolism-related indicators. The multi-objective optimization model is run under the individual constraints to adjust the daily energy reference value and the single meal food reference value of the population to obtain the individual daily energy target value and the individual single meal food recommendation amount; The actual daily dietary data uploaded by the user terminal is quantitatively compared with the individual's daily energy target value, the individual's recommended food quantity per meal, and the first type of constraint to generate dietary deviation results. Based on the dietary deviation results, dietary adjustment suggestions are generated and output to the user terminal.

2. The dietary health assessment method as described in claim 1, characterized in that, The multi-objective optimization model is trained by feeding the sample dataset, the first type of constraints, and the second type of constraints into the initial training model. The sample dataset includes historical dietary records of the population and their corresponding lipid metabolism-related indicators, energy consumption values, and nutrient intake values.

3. The dietary health assessment method as described in claim 1, characterized in that, The multi-objective optimization model is run on a predefined set of population cluster parameters to obtain daily energy reference values ​​and single-meal food reference values ​​for the population, including: Based on the set of parameters of the population cluster, calculate the basal metabolic rate and total daily energy consumption; Based on the basal metabolic rate and total daily energy consumption, a preset multi-objective genetic algorithm is used to perform a global search on the multi-objective optimization model to generate an initial non-dominated solution set. The gradient descent algorithm is executed on the initial non-dominated solution set until convergence, and the optimal solution that satisfies the first type of constraint is obtained. The optimal solution includes the daily energy value and the reference weights corresponding to various types of food. The daily energy value is set as the daily energy reference value for the population, and the reference weights corresponding to various types of food are set as the single-meal food reference values ​​for the population.

4. The dietary health assessment method as described in claim 3, characterized in that, Based on the basal metabolic rate and daily total energy consumption, a preset multi-objective genetic algorithm is used to perform a global search on the multi-objective optimization model to generate an initial non-dominated solution set, including: With maximizing dietary fiber intake as the primary optimization objective, the multi-objective optimization model is subjected to a first non-dominated ranking to obtain a first non-dominated solution set. Within the first non-dominated solution set, a second non-dominated sorting is performed with the second optimization objective of maximizing the proportion of foods with high micronutrient density, thus forming the initial non-dominated solution set.

5. The dietary health assessment method as described in claim 3, characterized in that, Perform gradient descent on the initial non-dominated solution set until convergence to obtain the optimal solution satisfying the first type of constraint, including: Set the corresponding gradient descent step size according to the food category; The weight of various foods is iteratively updated along the preset gradient direction with the gradient descent step size to obtain a converged solution. The convergence algorithm is determined to be converged when the change in the converged solution of the multi-objective optimization model during the iterative update is less than a preset threshold. Output the current converged solution as the optimal solution.

6. The dietary health assessment method as described in claim 1, characterized in that, Running the multi-objective optimization model under the individual constraints to adjust the population's daily energy reference value and single-meal food reference value to obtain the individual's daily energy target value and recommended single-meal food quantity, including: The individualized constraints are mapped to the boundary parameters of the multi-objective optimization model; A pre-defined multi-objective reinforcement learning algorithm is used to iteratively scale the boundary parameters by employing a composite reward function consisting of user compliance rewards and lipid metabolism risk penalties. After each iteration of scaling, the multi-objective optimization model is run to generate corresponding candidate solutions and obtain a candidate solution set. Each candidate solution includes candidate values ​​for daily energy and candidate values ​​for food weight. Select the candidate solution that simultaneously satisfies the first type of constraint and has the largest composite reward function from the candidate solution set, and set the corresponding daily energy candidate value and food weight candidate value as the individual's daily energy target value and individual single meal food recommendation amount, respectively.

7. The dietary health assessment method as described in claim 1, characterized in that, The actual daily dietary data includes actual total daily energy, actual nutrient intake, and the weight of various foods at each meal; the actual daily dietary data uploaded by the user terminal is quantitatively compared with the individual's daily energy target value, the individual's recommended food intake per meal, and the first type of constraint to generate dietary deviation results, including: Calculate the energy deviation between the actual total daily energy and the individual's daily energy target value; Calculate the nutrient deviation between the actual nutrient intake and the first type of constraint; Calculate the weight deviation between the weight of each type of food in the single meal and the recommended amount of food for the individual single meal; The energy deviation, nutrient deviation and weight deviation are each assigned a preset weight and then summed to obtain the dietary deviation result.

8. The dietary health assessment method as described in claim 1, characterized in that, After generating dietary adjustment recommendations and outputting them to the user terminal, the dietary health assessment method further includes: The user terminal collects confirmation information regarding the user's acceptance of the dietary adjustment recommendations. The execution confirmation information is sent back to the multi-objective optimization model to update the individual constraints.

9. A dietary health assessment system, characterized in that, The dietary health assessment system includes: The constraint quantification module is used to quantify dietary assessment indicators into a first type of constraint and a second type of constraint. The first type of constraint includes energy, nutrient energy ratio and nutrient intake, while the second type of constraint includes the energy ratio of three meals, the intake of various foods and food categories. The model building module is used to build a multi-objective optimization model based on the first type of constraints and the second type of constraints. The multi-objective optimization model is used to optimize the daily energy negative balance, dietary fiber intake and the proportion of high micronutrient density foods. The population solution module is used to run the multi-objective optimization model on a predefined set of population cluster parameters to obtain the daily energy reference value and the single meal food reference value of the population. The set of population cluster parameters includes at least gender, age, height, weight, BMI, activity coefficient and lipid metabolism-related indicators. The individual constraint module is used to adjust the second type of constraints to form individual constraint conditions based on individual characteristic data collected by the user terminal. The individual characteristic data includes at least real-time weight, energy deficit target, food preference, allergens and lipid metabolism-related indicators. The individual solution module is used to run the multi-objective optimization model under the individual constraints to adjust the daily energy reference value and the single meal food reference value of the population to obtain the individual daily energy target value and the individual single meal food recommended amount; The deviation comparison module is used to quantitatively compare the actual daily diet data uploaded by the user terminal with the individual's daily energy target value, the individual's recommended food amount per meal, and the first type of constraint to generate a diet deviation result. The suggestion output module is used to generate dietary adjustment suggestions based on the dietary deviation results and output them to the user terminal.

10. A computer device, characterized in that, The computer device includes: Memory, used to store computer programs; and A processor for executing the computer program to implement the dietary health assessment method as described in any one of claims 1-8.

Images