Personalized nutrition intake recommendation method and device, equipment and storage medium

Abstract

The invention discloses a personalized nutrition intake recommendation method and device, equipment and a storage medium, and relates to the technical field of nutrition intake assessment, and the personalized nutrition intake recommendation method comprises the steps: obtaining a nutrition diet standard, initial user information and a nutrition recommendation intake range; the initial user information is preprocessed to obtain target sequence set data, and the target sequence set data comprises current user information; training and adjusting an initial nutrition intake evaluation model according to the nutrition dietary standard, the target sequence set data and the recommended nutrition intake range to obtain a target nutrition intake evaluation model; and inputting the current user information into the target nutrition intake evaluation model to obtain a nutrition intake suggestion. According to the method and the device, periodic nutrition intake evaluation can be carried out according to the actual condition of an individual, accurate and personalized nutrition guidance is provided for the user, and the user is helped to realize a health target and maintain nutrition balance.

Description

Technical Field

[0001] The present application relates to the technical field of nutritional intake assessment, and in particular to a personalized nutritional intake recommendation method, device, equipment and storage medium. Background Art

[0002] With the accelerated pace of life and the diversification of dietary structure, individuals' understanding of nutritional intake has gradually deepened. However, traditional nutritional intake assessment methods often rely on static and fixed nutritional standards, which are difficult to accurately reflect the nutritional needs of individuals at different stages and states of life. This has led to a lack of personalized nutritional plans, and many people find it difficult to achieve scientific and reasonable intake in terms of diet.

[0003] In current nutrition assessment practices, many systems still use a unified standard, ignoring each person's physiological characteristics, living habits, and dietary preferences. As a result, individual nutrition intake assessments are often not accurate enough to meet the actual needs of users. This limitation not only affects people's health management, but also greatly reduces the effectiveness of nutrition intervention measures.

[0004] In addition, existing nutrition recommendation systems usually lack a dynamic update mechanism and cannot monitor individual health status and dietary changes in real time. This means that users often cannot get timely and appropriate adjustments when receiving nutrition advice. In order to better adapt to the changes of individuals at different stages and states. Therefore, how to conduct phased nutrition intake assessment based on the actual situation of individuals has become an urgent problem to be solved.

[0005] The above contents are only used to assist in understanding the technical solution of the present application and do not constitute an admission that the above contents are prior art. Summary of the invention

[0006] The purpose of this application is to provide a personalized nutritional intake recommendation method, device, equipment and storage medium, aiming to solve the technical problem of how to conduct phased nutritional intake assessment based on the actual situation of the individual.

[0007] To achieve the above objectives, the present application proposes a personalized nutrition intake recommendation method, the method comprising:

[0008] Obtain nutritional dietary standards, initial user information, and recommended nutritional intake ranges;

[0009] Preprocessing the initial user information to obtain target sequence set data, wherein the target sequence set data includes current user information;

[0010] The initial nutrition intake assessment model is trained and adjusted according to the nutrition diet standard, the target sequence set data and the recommended nutrition intake range to obtain a target nutrition intake assessment model;

[0011] The current user information is input into the target nutrition intake assessment model to obtain nutrition intake recommendations.

[0012] In one embodiment, the target sequence set data includes current user information and historical user information, and the step of training and adjusting the initial nutrition intake assessment model according to the nutrition diet standard, the target sequence set data and the recommended nutrition intake range to obtain the target nutrition intake assessment model includes:

[0013] Construct an initial nutrition intake assessment model based on the input layer, long short-term memory network layer, deep feature layer, and fully connected layer;

[0014] Using the historical user information to train the initial nutrition intake assessment model to obtain a reference nutrition intake assessment model;

[0015] The reference nutrition intake assessment model is adjusted according to the nutrition diet standard, the current user information and the recommended nutrition intake range to obtain a target nutrition intake assessment model.

[0016] In one embodiment, the step of using the historical user information to train the initial nutrition intake assessment model to obtain a reference nutrition intake assessment model includes:

[0017] Obtaining a predicted nutritional intake value according to the historical user information and the initial nutritional intake assessment model;

[0018] Calculate the recommended nutritional intake value according to the historical user information and the recommended nutritional intake range;

[0019] A loss value is calculated according to the predicted nutritional intake value, the recommended nutritional intake value and a mean square error loss function;

[0020] The parameters of the initial nutrition intake assessment model are adjusted using an adaptive moment estimation optimizer until the loss value is less than a preset loss value, thereby obtaining a reference nutrition intake assessment model.

[0021] In one embodiment, the step of obtaining a predicted nutritional intake value according to the historical user information and the initial nutritional intake assessment model comprises:

[0022] The historical user information is received through the input layer, wherein the historical user information is obtained by processing non-sequential data including basic information and dietary taboo information and sequential data including dietary record information, exercise information and special service records;

[0023] Determine the time corresponding to each different type of sequence data in the historical user information, and perform sorting processing according to the time dependency to obtain corresponding sequence data;

[0024] Abstracting the sequence data and learning deep features to obtain high-dimensional features;

[0025] The high-dimensional features are mapped to obtain a nutritional intake prediction value, wherein the nutritional intake prediction value includes the intake of nutrients in calories, protein, fat and carbohydrates.

[0026] In one embodiment, the step of adjusting the reference nutrition intake assessment model according to the nutrition diet standard, the current user information and the recommended nutrition intake range to obtain a target nutrition intake assessment model comprises:

[0027] Inputting the current user information into the reference nutrition intake assessment model to obtain a nutrition intake assessment result;

[0028] When the nutrition intake assessment result is outside the recommended nutrition intake range, the nutrition intake assessment result is used as feedback information, and parameters of the reference nutrition intake assessment model are adjusted according to the feedback information to obtain a target nutrition intake assessment model;

[0029] When the nutrition intake assessment result is within the recommended nutrition intake range, the reference nutrition intake assessment model is used as the target nutrition intake assessment model.

[0030] In one embodiment, the initial user information includes basic information, activity mode, activity duration, dietary taboo information, and special service record information, and the step of obtaining the recommended nutritional intake range includes:

[0031] Calculate the body mass index according to the basic information;

[0032] Obtaining weight status according to the body mass index;

[0033] Calculate the energy consumption according to the activity mode and the activity duration;

[0034] Obtaining physical activity intensity according to the energy consumption and a preset activity intensity-energy consumption range;

[0035] Obtaining an energy coefficient according to the physical activity intensity, the body weight status and a preset energy coefficient table;

[0036] Calculate the normal daily calorie intake range according to the nutritional dietary standard, the body weight status, the physical activity intensity and the energy coefficient;

[0037] When the special service record information is a body fat reduction service, the normal daily calorie intake range is adjusted according to the basal metabolic rate, exercise coefficient, fat reduction coefficient and the dietary taboo information to obtain a recommended nutritional intake range;

[0038] When the special service record information is for other services, the normal daily calorie intake range is adjusted according to the dietary taboo information and the special service record information to obtain the recommended nutritional intake range.

[0039] In one embodiment, the initial user information includes basic information, dietary taboo information, dietary record information, exercise record information, and special service record information of the user, and the step of preprocessing the initial user information to obtain target sequence set data includes:

[0040] Deleting invalid data in the initial user information, and processing outlier data in the dietary taboo information and the dietary record information to obtain second historical user information;

[0041] Encoding the basic information and the dietary taboo information in the second historical user information and converting them into binary column data;

[0042] Filling the missing information in the diet record information, the exercise record information and the special service record information with the adjacent average value or weighted average value to obtain continuous time series data;

[0043] Cutting the continuous time series data into time window segment data of fixed length;

[0044] Performing logarithmic transformation on the diet record information, the exercise record information, and the special service record information, and performing feature combination on the special service record information to obtain combined feature data;

[0045] The continuous time series data, the time window segment data, and the diet record information, the exercise record information, and the special service record information in the combined feature data are combined into a three-dimensional array, and normalized to obtain initial sequence set data;

[0046] The binary column data is added to each time step in the initial sequence set data to obtain the target sequence set data.

[0047] In addition, to achieve the above objectives, the present application also proposes a personalized nutrition intake recommendation device, the device comprising:

[0048] The data acquisition module is used to obtain nutritional dietary standards, initial user information, and recommended nutritional intake ranges;

[0049] A data processing module, used for preprocessing the initial user information to obtain target sequence set data, wherein the target sequence set data includes current user information;

[0050] A model training module, used to train and adjust the initial nutrition intake assessment model according to the nutrition diet standard, the target sequence set data and the recommended nutrition intake range to obtain a target nutrition intake assessment model;

[0051] The nutrition intake recommendation module is used to input the current user information into the target nutrition intake assessment model to obtain nutrition intake recommendations.

[0052] In addition, to achieve the above-mentioned purpose, the present application also proposes a personalized nutritional intake recommendation device, which includes: a memory, a processor, and a computer program stored on the memory and executable on the processor, and the computer program is configured to implement the steps of the personalized nutritional intake recommendation method described above.

[0053] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the personalized nutrition intake recommendation method as described above are implemented.

[0054] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the steps of the personalized nutrition intake recommendation method as described above.

[0055] One or more technical solutions proposed in this application have at least the following technical effects:

[0056] Obtain nutritional dietary standards, initial user information, and recommended nutritional intake ranges; preprocess the initial user information to obtain target sequence set data, which includes current user information; train and adjust the initial nutritional intake assessment model according to the nutritional dietary standards, the target sequence set data, and the recommended nutritional intake range to obtain a target nutritional intake assessment model; input the current user information into the target nutritional intake assessment model to obtain nutritional intake recommendations. The system first obtains nutritional dietary standards, initial user information, and recommended nutritional intake ranges, wherein the nutritional dietary standards provide a scientific basis, the initial user information covers the user's basic information, dietary taboos, exercise records, etc., and the recommended nutritional intake range determines the daily nutritional needs based on the user's personalized data. This ensures that the system can make a preliminary assessment based on scientific standards and user-specific needs. Next, preprocess the initial user information, clean invalid data, fill in missing values, and convert it into time series data. Through this step, the system improves data quality and ensures that the model can accurately capture the user's behavioral change trend. On this basis, the system uses nutritional dietary standards, preprocessed target sequence set data, and recommended ranges to train and optimize the initial nutritional intake assessment model, and finally generates a personalized target model. After training, the system can recognize the user's historical behavior patterns and improve the model's prediction accuracy for individual needs. Finally, the current user information is input into the target nutrition intake assessment model, and the system generates personalized nutrition intake recommendations, including calories and the ratio of the three major nutrients. This step ensures that the user's recommendations are dynamically adjusted and closely fit their health status and goals. Through this process, the system can conduct phased nutrition intake assessments based on the individual's actual situation, providing users with accurate and personalized nutrition guidance to help them achieve their health goals and maintain nutritional balance. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0058] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0059] Figure 1 A schematic diagram of a process flow provided for Example 1 of the personalized nutrition intake recommendation method of this application;

[0060] Figure 2 A schematic diagram of a process flow provided for Example 2 of the personalized nutrition intake recommendation method of this application;

[0061] Figure 3 This is a schematic diagram of the module structure of the personalized nutrition intake recommendation device according to an embodiment of the present application;

[0062] Figure 4 This is a schematic diagram of the device structure of the hardware operating environment involved in the personalized nutrition intake recommendation method in the embodiment of the present application.

[0063] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0064] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0065] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0066] The accelerated pace of life and diversified diets have led to a deeper understanding of people's nutritional needs, but traditional assessment methods have failed to be personalized and ignored individual differences, resulting in inaccurate nutrition plans, affecting health and intervention effects. The existing system lacks dynamic updates, cannot adjust recommendations in real time, and is not adaptable to individual changes.

[0067] The main solution of the embodiment of the present application is to establish a preliminary assessment by obtaining nutrition standards, user information and recommended intake ranges, pre-process the user information to improve data quality, and then train the optimization model to generate a personalized nutrition intake assessment. Finally, the system dynamically generates nutrition recommendations based on the user's latest information to ensure that the recommendations are in line with the user's current health status and goals.

[0068] It should be noted that the execution subject of the embodiment of the present application can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device capable of realizing the above functions, a personalized nutrition intake recommendation system, etc. The following takes the personalized nutrition intake recommendation system as an example to illustrate this embodiment and the following embodiments.

[0069] Based on this, the present application embodiment provides a personalized nutrition intake recommendation method, referring to Figure 1 , Figure 1 This is a schematic diagram of the process of the first embodiment of the personalized nutrition intake recommendation method of the present application.

[0070] In this embodiment, the personalized nutrition intake recommendation method includes steps S10 to S40:

[0071] Step S10, obtaining nutritional dietary standards, initial user information, and recommended nutritional intake range;

[0072] It should be noted that the nutritional dietary standards refer to the dietary guidelines and recommendations formulated by the country or region to ensure the health and nutrition of residents. These standards are usually based on a large number of nutritional studies and are intended to help individuals and groups obtain the required nutrients and prevent nutritional deficiencies and related diseases. In this embodiment, the nutritional dietary standards include the recommended intake and proportion of various nutrients (such as protein, fat, carbohydrates, vitamins and minerals) required daily to guide users to choose appropriate food and eating habits. The initial user information refers to the data related to the user's health status and lifestyle collected at the beginning of the nutrition recommendation system. This information usually includes the user's age, gender, weight, height, activity level, health status (such as whether there are chronic diseases), eating habits, etc. This information is crucial for formulating personalized nutritional advice, because the nutritional needs of different users may vary depending on these factors. In this embodiment, the initial user information will help the system evaluate the user's basal metabolic rate and the total energy required per day in order to provide more accurate nutritional advice. The recommended nutritional intake range (Ys[min, max]) refers to the range of recommended intakes of various nutrients provided to users based on the nutritional dietary standards and the initial user information. This is usually expressed in the form of minimum values ​​(Ys[min]) and maximum values ​​(Ys[max]) to ensure that the user does not consume these nutrients below the required minimum standard nor exceed the upper limit that may cause health problems. In this embodiment, the recommended nutrient intake range will be set according to the individual differences and nutritional standards of the user, thereby ensuring that the user's diet can meet their health needs.

[0073] Understandably, it is necessary to obtain nutritional dietary standards first. This process ensures that the nutritional intake based on the system is scientific and authoritative by consulting the guidelines issued by various authoritative health institutions (such as the World Health Organization, the National Nutrition Association, etc.). The importance of this step is that it provides a reliable reference framework for the development of personalized nutrition plans, which helps to ensure that the user's nutritional intake meets health needs, thereby preventing the risk of malnutrition or overnutrition. Next, collect initial user information. This step involves a detailed questionnaire or interview, including the user's basic information (such as age, gender, height, weight), lifestyle (such as daily activity level, exercise frequency), and eating habits and special health conditions (such as chronic diseases, allergy history, etc.). The purpose of this is to gain a deep understanding of the user's specific needs and background, to ensure that subsequent nutritional recommendations are targeted and personalized, so as to more effectively meet the user's health goals. Finally, the system sets the recommended nutritional intake range (Ys[min,max]) based on the above information. This step combines user information with nutritional dietary standards to clarify the minimum and maximum daily intake of specific nutrients to ensure that users get enough nutrition while maintaining health. This approach can effectively prevent health problems caused by unbalanced nutrition, improve the user's quality of life, and lay the foundation for them to develop a sustainable diet plan.

[0074] As an example, the initial user information includes basic information, activity mode, activity duration, dietary taboo information and special service record information, and the step of obtaining the recommended nutritional intake range includes: calculating a body mass index based on the basic information; obtaining a weight status based on the body mass index; calculating an energy consumption based on the activity mode and the activity duration; obtaining physical activity intensity based on the energy consumption and a preset activity intensity-energy consumption range; obtaining an energy coefficient based on the physical activity intensity, the weight status and a preset energy coefficient table; calculating a normal daily calorie intake range based on the nutritional diet standard, the weight status, the physical activity intensity and the energy coefficient; when the special service record information is a body fat reduction service, adjusting the normal daily calorie intake range based on the basal metabolic rate, the exercise coefficient, the fat reduction coefficient and the dietary taboo information to obtain a recommended nutritional intake range; when the special service record information is other services, adjusting the normal daily calorie intake range based on the dietary taboo information and the special service record information to obtain a recommended nutritional intake range.

[0075] Basic information refers to the user's basic personal information, including age, gender, height, weight, daily activity, etc. This information is the basis for calculating other health indicators (such as body mass index). Activity mode refers to the description of the user's daily activity type, such as sitting, light exercise (such as walking), moderate exercise (such as jogging) or high-intensity exercise (such as gym training). Different activity modes will affect the user's energy consumption. Activity duration refers to the length of time the user performs an activity, such as exercising for 30 minutes a day. This data is crucial for calculating energy consumption. Dietary taboo information refers to the user's taboos or preferences for certain foods, including steamed foods, dairy products, soy products, seafood and spicy foods, such as allergens (such as nuts, dairy products) or specific dietary requirements (such as vegetarian, gluten-free diet). This information helps personalize dietary recommendations. Special service record information refers to the specific nutrition or fitness service records received by the user, such as muscle building plans, blood pressure lowering services, blood sugar control services, blood lipid clearing services, uric acid lowering services and body fat reduction services. This information will affect the adjustment direction of nutrition recommendations. Among them, the blood pressure lowering service includes systolic blood pressure and diastolic blood pressure, the blood sugar control service includes fasting blood sugar and 2 hours after meal blood sugar, the blood lipid clearing service includes total cholesterol, triglycerides, low-density lipoprotein cholesterol and high-density lipoprotein cholesterol, the uric acid lowering service includes uric acid, the body fat reduction service includes exercise intensity, height and weight, and the nutritional information includes calories, protein, fat, and carbohydrates. The individual's basic information and dietary taboos are entered through the APP; the individual's dietary records are automatically recorded through the restaurant's smart devices; the individual's exercise records are automatically collected by wearing a bracelet; the individual's special service records are collected through OCR text recognition. Body Mass Index (BMI) refers to an indicator calculated by a formula, which is used to assess whether the weight is within a healthy range. The formula is: BMI = weight (kg) / height (m 2). For example, a person with a height of 1.75 meters and a weight of 70 kilograms has a BMI of 22.86, which is within the normal range. Weight status refers to the user's weight status based on the calculation results of the body mass index, which is usually divided into: thin, normal, overweight and obese. The assessment of weight status helps to understand the user's health level. Energy consumption refers to the energy consumed by the user due to activities in a certain period of time, usually calculated in calories (kcal). The calculation method involves basal metabolic rate and activity consumption. The more intense the activity, the more energy is consumed. Preset activity intensity-energy consumption range refers to presetting an energy consumption range for each activity based on different activity types and intensities. For example, slow walking consumes about 240-300 calories per hour, while high-intensity running may consume 600-800 calories / hour. Physical activity intensity refers to the exercise intensity assessed based on the user's activity mode and duration, which is usually divided into light, moderate and high intensity. For example, brisk walking for 30 minutes a day may be considered a moderate intensity activity. The preset energy coefficient table refers to the energy coefficients corresponding to different physical activity intensities. These coefficients are used to estimate the energy needs of users at specific activity levels. For example, the energy coefficient of light activity may be 1.2, moderate activity is 1.55, and high-intensity activity is 1.9. The energy coefficient refers to a value calculated based on factors such as the user's weight and activity intensity, which is used to evaluate the basis of daily energy needs. For example, if a user's energy coefficient is 1.5, it means that his daily energy needs are 1.5 times the basal metabolic rate. The daily normal calorie intake range refers to the daily calorie intake range calculated based on the user's basic information, weight status, physical activity intensity and energy coefficient, usually in calories. For example, the normal range may be 1500-2500 calories, depending on individual differences. Body fat reduction services refer to specific nutrition or fitness services designed to help users reduce body fat, such as weight loss diet plans, fitness plans, etc. Such services usually include nutritional adjustments and exercise guidance. Basal Metabolic Rate (BMR) refers to the minimum energy consumption required by the human body to maintain life in a resting state, which is usually determined by factors such as gender, age, weight and height. For example, the BMR of an adult male is usually between 1600-1800 calories. BMR (female) = 655 + (9.6 × weight kg) + (1.8 × height cm) - (4.7 × age years); BMR (male) = 66 + (13.7 × weight kg) + (5 × height cm) - (6.8 × age years). The exercise coefficient refers to a coefficient used to reflect the user's activity level, which is usually divided according to the intensity of the activity. For example, the exercise coefficient for light activity is 1.2, for moderate activity is 1.55, and for high-intensity activity is 1.9. The fat loss coefficient refers to the coefficient used in the fat loss service to adjust energy intake according to the user's goals.For example, if the fat loss coefficient is 0.8, it means that the user needs to reduce the daily calorie intake by 20% based on the normal daily calorie intake range to achieve fat loss. Other services refer to non-fat loss related services that users participate in, such as muscle building, sports nutrition guidance, etc. The adjustment plan of such services may be personalized based on dietary taboo information and user goals.

[0076] In the step of obtaining the recommended nutritional intake range, the user's initial information is first collected, including basic information, activity mode, activity duration, dietary taboo information, and special service record information. This is done to fully understand the user's personal situation and needs and ensure the accuracy of subsequent calculations. Next, the body mass index (BMI) is calculated based on the user's basic information. This step is to assess the user's weight status in order to formulate a suitable diet plan for them. Then, the energy consumption is calculated in combination with the user's activity mode and duration, which can accurately reflect the user's daily energy needs and provide a basis for nutritional intake. Using the preset activity intensity-energy consumption range, the user's physical activity intensity can be obtained, which helps to further adjust the nutritional intake to suit their activity level. Next, the energy coefficient is calculated by combining the physical activity intensity, weight status and the preset energy coefficient table to assess the user's energy needs at different activity levels and ensure the balance of energy intake and consumption. Then, the normal daily calorie intake range is calculated using the nutritional diet standard and the above parameters, which provides a healthy dietary goal for the user. When the special service record information shows that the user is taking the body fat reduction service, the normal daily calorie intake range will be adjusted based on the basal metabolic rate, exercise coefficient and fat reduction coefficient, aiming to help the user effectively reduce body fat and achieve fat reduction effects; and when the record information is other services, the calorie intake range will be adjusted accordingly based on the dietary taboo information and special service records to meet the user's special needs and health goals. Through this systematic step, it can ensure that the recommended nutritional intake range is both scientific and personalized, thereby effectively supporting users to achieve their health and nutrition goals.

[0077] Step S20, preprocessing the initial user information to obtain target sequence set data, wherein the target sequence set data includes current user information;

[0078] It should be noted that preprocessing refers to the steps of cleaning and organizing the initial user information, including removing duplicate information, filling missing values, standardizing data formats, etc. Preprocessing ensures the consistency and accuracy of the input data so that subsequent data analysis and processing can proceed smoothly, improving the effectiveness and reliability of the system. The target sequence set data is a data set formed after preprocessing, which contains all necessary information related to the user's health and nutrition. These data include not only basic information and the user's living habits, but also other features related to the user's goals, such as nutritional intake recommendations, activity levels, etc., so that the system can make accurate judgments based on this information when making personalized recommendations and analysis. Current user information refers to real-time data related to the user at a specific point in time, which may include the user's latest health status, weight changes, diet records, exercise status, etc. By tracking the current user information, the system can update and adjust nutrition recommendations in a timely manner to ensure that the user's health goals are always in an adaptive state, thereby achieving more personalized and effective nutrition management.

[0079] It is understandable that the user's basic information, lifestyle and eating habits are first collected and organized to ensure the integrity and consistency of the data. This process includes removing duplicate records, filling missing values, standardizing data formats and classification information, etc., so as to eliminate the interference of noise data on subsequent analysis. After preprocessing, the target sequence set data formed not only includes the user's basic health information, but also integrates real-time user dynamics, such as weight, eating habits and exercise. This target sequence set data is established to ensure that the system can analyze based on the latest and accurate information, provide personalized nutrition advice and health management plans, thereby improving user participation and satisfaction, and ultimately promoting the realization of their health goals.

[0080] As an example, the initial user information includes the user's basic information, dietary taboo information, dietary record information, exercise record information, and special service record information. The step of preprocessing the initial user information to obtain the target sequence set data includes: deleting invalid data in the initial user information, and processing outlier data in the dietary taboo information and the dietary record information to obtain second historical user information; encoding the basic information and the dietary taboo information in the second historical user information and converting them to obtain binary column data; filling in the missing information in the dietary record information, the exercise record information, and the special service record information by using the nearest average or weighted average. The method comprises the steps of: performing a three-dimensional array on the continuous time series data, performing a normalization process on the continuous time series data, and obtaining a continuous time series data; cutting the continuous time series data into time window segment data of a fixed length; performing a logarithmic transformation on the diet record information, the exercise record information, and the special service record information, and performing feature combination on the special service record information to obtain combined feature data; merging the continuous time series data, the time window segment data, and the diet record information, the exercise record information, and the special service record information in the combined feature data into a three-dimensional array, and performing a normalization process on the three-dimensional array to obtain an initial sequence set data; and adding the binary column data in each time step in the initial sequence set data to obtain a target sequence set data.

[0081] Dietary record information refers to the user's diet over a certain period of time, including detailed information such as the type, quantity and time of each meal, so as to evaluate their nutritional intake. Exercise record information refers to the user's exercise over a certain period of time, including the type, duration, intensity and frequency of exercise, so as to evaluate their physical activity and energy consumption. Invalid data refers to user information that cannot be used for analysis, such as data with incorrect format, missing key fields, repeated records or obviously illogical data. Outlier data refers to data that is obviously inconsistent with other data points, usually in the form of extreme values ​​or abnormal values, which may affect the accuracy of the analysis results. Second, historical user information refers to historical user information that is retained and cleaned after preprocessing, including basic information and dietary taboo information. Encoding refers to the process of converting the user's basic information and dietary taboo information into a specific format, usually converting categorical data into digital form for subsequent calculations. Binary column data refers to a data column in binary format generated by the encoded data, which is used to represent the user's basic information and dietary taboo information, and usually uses 0 and 1 to represent different states or categories. Missing information refers to data missing in the recorded information, such as certain values ​​or categories omitted in the diet or exercise records. The nearest average or weighted average refers to the method used to fill in missing information. The nearest average refers to the average of the adjacent data before and after the missing value, while the weighted average assigns different weights to different data points according to the importance of the data to calculate the average. Filling processing refers to the process of using a certain strategy (such as the nearest average or weighted average) to fill in missing data for missing information to ensure the integrity and continuity of the data. Continuous time series data refers to a data set of diet and exercise information recorded by users that is continuous in time, representing the dynamic situation that changes over time. Time window segment data refers to dividing continuous time series data into small segments of fixed length for local analysis and processing. Logarithmic transformation refers to the operation of logarithmic transformation of information on diet records, exercise records, and special service records, with the aim of reducing the skewed distribution of data, making the data closer to normal distribution, and improving the stability of the model. Feature combination refers to combining multiple related features to form new features to better describe the user's behavior or situation, and is usually used to improve the effect of analysis or prediction. Combined feature data refers to the data generated after feature combination, which contains information from multiple dimensions and helps to analyze the user's situation more comprehensively. A three-dimensional array refers to a data format that combines dietary record information, exercise record information, and special service record information into a three-dimensional structure for easy storage and calculation. Normalization processing refers to converting data to the same scale range (for example, between 0 and 1) to eliminate the dimensional differences between different features and make model training more effective. The initial sequence set data refers to the data set after normalization processing, which contains user information after multiple processing and conversion, ready for further analysis and modeling. Time step refers to each specific time point or time period in the time series data, representing the user's status information.

[0082] When preprocessing the initial user information to obtain the target sequence set data, firstly, invalid data is deleted to ensure the accuracy and reliability of the information to avoid interference with subsequent analysis. Then, the outlier data in the dietary taboo information and dietary record information is processed to eliminate the impact of outliers on the overall data quality, so as to obtain clearer second historical user information. Then, the basic information and dietary taboo information are encoded and converted into binary column data, which can simplify data processing and facilitate the input of subsequent models. For the missing data in the dietary record information, exercise record information and special service record information, the adjacent average or weighted average is used for filling processing, which can effectively complete the data and maintain the continuity of the time series, which can improve the prediction performance of the model. Next, the continuous time series data is cut into fixed-length time window fragment data, so that the data can be processed in segments, which helps to capture the time series characteristics. Logarithmic transformation of dietary record information, exercise record information and special service record information can reduce the skewness of the data, enhance the stability of the model, and improve the expression ability of information through feature combination. Finally, normalizing the three-dimensional array data helps eliminate the impact of different feature dimensions, improve the efficiency and accuracy of model training, and add binary column data at each time step to ensure that the target sequence set data contains rich information for subsequent analysis and model training.

[0083] Step S30, training and adjusting the initial nutrition intake assessment model according to the nutrition diet standard, the target sequence set data and the recommended nutrition intake range to obtain a target nutrition intake assessment model;

[0084] It should be noted that the initial nutrition intake assessment model refers to a preliminary model used to assess the user's nutrition intake. It is based on the provided nutrition diet standards, the user's target sequence set data, and the preliminary recommended nutrition intake range. It aims to analyze the relationship between the user's eating habits and the required nutrition, but it has not been fully trained and optimized, and there may be certain errors. The target nutrition intake assessment model is a trained and adjusted model, and its purpose is to provide a more accurate nutrition intake assessment. By combining nutrition diet standards, target sequence set data, and recommended nutrition intake ranges, the model can more effectively identify the user's actual nutrition needs and intake through algorithm optimization and parameter adjustment, thereby providing users with personalized nutrition recommendations. The establishment of this model can not only improve the accuracy of nutrition assessment, but also support subsequent diet adjustments and health management strategies, providing a scientific basis for users' dietary health.

[0085] It is understandable that the nutritional dietary standards need to be used as a reference framework to clarify the recommended intake of different nutrients and their standards, and then use the target sequence set data, which contains the user's dietary records and personal characteristics, as the input for model training to ensure that the model can capture the user's actual eating habits and physiological needs. Then, by comparing the recommended nutritional intake range with the user's actual intake, the model will identify potential insufficient or excessive intake of nutrients, and make dynamic adjustments. These data are repeatedly trained using machine learning algorithms or statistical analysis methods to improve the accuracy and predictive ability of the model, and finally form a target nutritional intake assessment model. This model can not only more accurately assess the user's nutritional intake, but also provide personalized dietary recommendations to meet the user's health needs and goals. Through this process, the scientificity and practicality of nutritional management can be effectively improved, helping users to better adjust their dietary structure.

[0086] Step S40, inputting the current user information into the target nutrition intake assessment model to obtain nutrition intake recommendations.

[0087] It should be noted that nutritional intake recommendations refer to personalized dietary recommendations generated after analyzing the current user information based on the target nutritional intake assessment model. These recommendations include specific nutrient intake, food types, dietary combinations, and meal arrangements, etc., aiming to help users achieve their health goals or improve their eating habits. Nutritional intake recommendations cover the following aspects: (1) Nutrient ratio: Recommended daily ratios of carbohydrates, protein, and fat to meet the user's nutritional needs. Protein: The system recommends daily protein intake based on the user's weight, exercise status, and health needs (such as muscle gain or fat loss), usually in grams, and gives a reasonable proportion of total daily calories (such as 15%-20%). Fat: Fat intake recommendations are adjusted based on the user's health status (such as blood lipid levels) and fat loss needs, giving the number of grams and proportion of daily fat intake. Carbohydrates: The system will give carbohydrate intake recommendations based on the user's physical activity level and health needs (such as blood sugar control), usually controlled at 50%-60% of total calories. (2) Food selection: Based on the user's dietary taboo information, the system recommends suitable food types, such as low-salt, low-sugar or foods rich in specific vitamins and minerals. (3) Meal schedule: Recommend the user's daily meal schedule, including the specific intake time and amount of breakfast, lunch, dinner and snacks. (4) Diet adjustment: Provide adjustment suggestions based on the user's health status or specific goals (such as fat loss, muscle gain), such as increasing the intake of certain nutrients or reducing the amount of high-calorie foods. (5) Micronutrient recommendations: The system may also provide recommendations for the intake of certain specific micronutrients (such as vitamins and minerals), especially for users with specific health needs (such as those with insufficient iron and calcium intake). (6) Food type and portion size recommendations: Based on the user's dietary preferences and taboo information, the system will recommend specific food types (such as eating more lean meat, beans, green vegetables, etc.) and appropriate food portions. For example, for patients with hypertension, it is recommended to reduce the intake of high-sodium foods; for patients with diabetes, it is recommended to reduce the intake of high-sugar foods. (7) Water intake recommendation: The system may also recommend the user's daily water intake. Especially for users who exercise a lot, the water demand will increase.

[0088] It is understandable that first, the system will extract the user's basic information (such as age, gender, weight and height), activity patterns and diet records to fully understand the user's health status and lifestyle. Then, this information will be passed to the trained target nutrient intake assessment model, which is based on the nutritional dietary standards and the user's historical eating habits for analysis. The model uses a machine learning algorithm to process the input data to calculate the specific intake of various nutrients required by the user. This process ensures the accuracy and scientificity of nutritional recommendations. Finally, the nutritional intake recommendations generated by the model not only include the recommended daily intake of various nutrients, but may also recommend specific food choices and meal arrangements to help users optimize their diet structure, thereby achieving health goals or improving quality of life. Through such personalized recommendations, users can more effectively follow healthy eating principles in their daily diet, thereby improving their overall health.

[0089] This embodiment provides a personalized nutrition intake recommendation method, which obtains nutrition diet standards, initial user information and nutrition recommended intake range; pre-processes the initial user information to obtain target sequence set data, which includes current user information; trains and adjusts the initial nutrition intake assessment model according to the nutrition diet standards, the target sequence set data and the nutrition recommended intake range to obtain a target nutrition intake assessment model; inputs the current user information into the target nutrition intake assessment model to obtain nutrition intake suggestions. The system first obtains nutrition diet standards, initial user information and nutrition recommended intake range, wherein the nutrition diet standards provide scientific basis, the initial user information covers the user's basic information, dietary taboos, exercise records, etc., and the nutrition recommended intake range determines the daily nutrition needs according to the user's personalized data. This ensures that the system can make a preliminary assessment based on scientific standards and user-specific needs. Next, the initial user information is pre-processed to clean invalid data, fill missing values, and convert it into time series data. Through this step, the system improves the data quality and ensures that the model can accurately capture the user's behavior change trend. On this basis, the system uses nutritional dietary standards, preprocessed target sequence set data and recommended ranges to train and optimize the initial nutritional intake assessment model, and finally generates a personalized target model. After training, the system can identify the user's historical behavioral patterns and improve the model's prediction accuracy for individual needs. Finally, the current user information is input into the target nutritional intake assessment model, and the system generates personalized nutritional intake recommendations, including calories and the ratio of the three major nutrients. This step ensures that the user's recommendations are dynamically adjusted and closely fit their health status and goals. Through this process, the system can conduct phased nutritional intake assessments based on the individual's actual situation, providing users with accurate and personalized nutritional guidance to help them achieve their health goals and maintain nutritional balance.

[0090] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above-mentioned embodiment 1 can refer to the above introduction, and will not be repeated later. Figure 2 , Figure 2 This is a flow chart of the second embodiment of the personalized nutrition intake recommendation method of the present application. The target sequence set data includes current user information and historical user information. Step S30 of the personalized nutrition intake recommendation method includes steps S31 to S33:

[0091] Step S31, constructing an initial nutrition intake assessment model according to the input layer, the long short-term memory network layer, the deep feature layer and the fully connected layer;

[0092] It should be noted that historical user information refers to the data of the user's diet, exercise, health status, etc. in the past period of time. These data can reflect the changes in user behavior and health indicators in the form of time series. Specifically, historical user information may include: the user's diet records in different time periods (such as the type of food consumed, calories, and nutrients); exercise records (such as exercise type, intensity, and duration); health indicators (such as weight, blood pressure, blood sugar, etc.). These historical data are used to analyze the user's long-term behavior patterns and help the model predict the user's future nutritional needs. The input layer is the first layer of the neural network model, which is used to receive and transmit input data. In this embodiment, the input layer is responsible for passing the user's historical information and current information to the model for processing. The input of the input layer is usually pre-processed data, including current user information and historical user information. The long short-term memory network layer (Long Short-Term Memory, LSTM) is a neural network layer for processing sequence data. LSTM can learn and capture the time dependency in the data, and is particularly suitable for processing the user's historical behavior data. In this embodiment, the LSTM layer remembers the user's long-term and short-term behavior patterns by analyzing the user's time series data (such as diet records and exercise records), and predicts the user's future nutritional needs based on these patterns. The deep feature layer refers to the hidden layer in the neural network, which is responsible for further extracting and processing the deep features in the user data. After being processed by the LSTM layer, the deep feature layer will further extract features from the data to find hidden complex relationships and patterns. Through this layer, the system can dig out more complex patterns in user behavior and help improve the model's predictive ability. The fully connected layer is the last layer of the neural network model, responsible for mapping the features extracted by the previous layer to the final output results. In this embodiment, the output of the fully connected layer is the user's nutritional intake recommendations, including specific intakes of calories, protein, fat, carbohydrates, etc. This layer generates personalized nutritional recommendations for users by comprehensively considering all input features.

[0093] It can be understood that, first of all, the target sequence set data consists of current user information and historical user information, where the current information reflects the user's latest status (such as diet and exercise on the day), while the historical information records the user's behavior and health changes over the past period of time (such as previous diet intake, exercise records, weight, etc.). These data are passed as input to the input layer of the model, which is responsible for receiving and processing this information. Next, the data enters the long short-term memory network layer (LSTM layer), which can effectively capture the time dependency in user behavior and health data, and remember the user's long-term and short-term behavior patterns, which is crucial for predicting future nutritional needs. Subsequently, the data passes through the deep feature layer, which further extracts complex features in the data and mines the potential relationship between user behavior and health status, thereby helping the model better understand the user's personalized needs. Finally, the data flows to the fully connected layer, which is responsible for summarizing the previously extracted features and finally generating the output of the model, that is, the preliminary nutritional intake assessment results, including nutritional intake recommendations such as daily calories, protein, fat and carbohydrates. Through this layer of processing, the model can provide personalized nutritional assessment based on the user's current status and historical behavior.

[0094] As an example, the historical user information is received through the input layer; the sequence features in the historical user information are extracted through the long short-term memory network layer; the sequence features are abstracted and processed through the deep feature layer to obtain high-dimensional features; and the high-dimensional features are mapped into nutritional intake prediction values ​​through the fully connected layer.

[0095] Sequence features are time-dependent and sequential features extracted from time series data in historical user information. These features can reflect the user's behavior patterns and changing trends over a period of time. For example, after the user's daily dietary intake, exercise intensity, weight fluctuations and other data over the past period of time are processed by the long short-term memory network layer (LSTM layer), the model can identify the time patterns of these behaviors (such as periodic eating habits, gradual weight loss, etc.). Sequence features can help the model understand the user's long-term health trends. Abstraction and processing refer to further processing the extracted sequence features in the deep feature layer to generate higher-level and more general features. Specifically, this process includes nonlinear transformation, combination or filtering of sequence features to mine complex and potential patterns. For example, the relationship between the user's weight change and dietary intake is combined to find the long-term impact of certain specific eating habits on weight. Through abstraction and processing, the system can better understand the deep relationships in the data, making the model have stronger predictive capabilities. High-dimensional features are feature sets containing a large amount of information generated after processing by the deep feature layer. These features are more complex than the original sequence features and can describe multiple dimensions of the user's behavior and health status. For example, processed high-dimensional features may include the user's comprehensive dietary pattern, exercise intensity change trend, the relationship between nutritional intake and weight change, etc. High-dimensional features enable the model to make judgments and predictions in multiple dimensions, thereby improving the accuracy of the prediction. The nutritional intake prediction value is the final result generated by the model, which usually includes daily nutritional recommendations, such as the calories, protein, fat, carbohydrates, etc. that the user should consume. These predictions are personalized nutritional recommendations derived from the model's processing based on the user's historical data and current status. Based on these predictions, the system provides users with specific dietary recommendations to help them optimize their nutritional intake to achieve health goals, such as fat loss, muscle gain, and sugar control.

[0096] First, the model receives the user's historical information through the input layer. This information includes the user's past diet records, exercise data, health indicators, etc., which constitute time series data. Then, the data enters the long short-term memory network layer (LSTM layer), which is specially used to process time series data and can extract sequence features in historical data, that is, the time-dependent patterns of user behavior and health changes (such as the periodicity of eating habits or the gradual change of weight). Then, the model passes these sequence features to the deep feature layer, where further abstraction and processing are performed, that is, complex nonlinear transformations and combinations of these features are performed to dig out deeper patterns and relationships, and generate richer high-dimensional features, such as the potential association between diet, exercise and health status. Finally, the model maps these high-dimensional features into specific nutritional intake predictions, that is, nutritional recommendations such as calories, protein, fat, carbohydrates, etc. required by users every day through the fully connected layer. This step ensures that the model generates accurate personalized nutritional guidance based on the user's historical behavior and current status.

[0097] As an example, the long short-term memory network layer includes an input gate, a forget gate and an output gate, and the step of extracting the sequence features in the historical user information through the long short-term memory network layer includes: writing the historical user information into a first unit state through the input gate; selectively discarding the historical user information in the first unit state through the forget gate to obtain a second unit state; and extracting the sequence features from the second unit state through the output gate.

[0098] The input gate is a control mechanism in the long short-term memory network (LSTM) that is responsible for deciding whether to allow new information (i.e., historical user information) to enter the LSTM unit. The role of the input gate is to control which information needs to be stored in the internal state of the LSTM (i.e., the first unit state) based on the characteristics of the current input data. In this step, historical user information (such as diet records, exercise data, etc.) is written into the memory unit of the LSTM network through the input gate. The forget gate is another control mechanism in the LSTM network that is used to decide which information already stored in the LSTM unit needs to be forgotten or retained. It screens the historical user information stored in the first unit state and selectively discards it based on current needs or information that is no longer relevant. This mechanism allows the LSTM network to retain important long-term information while discarding irrelevant or outdated content, thereby updating the second unit state. The output gate is the last control mechanism of the LSTM network, which is used to extract useful information from the current state of the LSTM unit as output. Specifically, the output gate determines which information needs to be output based on the second unit state, and the output information is the extracted sequence features. These sequence features are time-dependent patterns extracted by the LSTM layer from historical user information, which can help the model better predict the user's nutritional needs. The first unit state refers to the initial internal state stored by the LSTM unit after receiving historical user information. This state contains the initial representation of the historical user information in the model. The input gate controls which information can be stored in this unit state. The second unit state refers to the unit state after being filtered and updated by the forget gate, which represents the valid historical information stored by the LSTM unit. It retains the information that is considered useful for prediction after selection and discards irrelevant information. Finally, the model extracts sequence features from the second unit state as output.

[0099] First, the LSTM layer contains three main gates: input gate, forget gate, and output gate, which control the storage, discarding, and output of information. Specifically, through the input gate, historical user information (such as the user's diet record, exercise data, etc.) is written into the first unit state of the LSTM. The input gate determines which historical information is important and stores it in the memory unit of the model. Then, through the forget gate, the historical user information that has been stored in the first unit state is selectively discarded. The forget gate determines which information is no longer useful and needs to be discarded based on current needs and inputs, thereby updating the state and obtaining the second unit state. This step ensures that the model only retains long-term information that is helpful for current predictions and avoids interference from irrelevant information on the model. Finally, through the output gate, useful information, namely sequence features, are extracted from the second unit state. These features are important patterns or laws in historical information that have been screened and can provide a basis for subsequent predictions. This process enables the LSTM model to capture the time dependency in user behavior and accurately extract key long-term and short-term information.

[0100] As an example, the deep feature layer includes a bottom layer, a middle layer and a top layer. The step of abstracting and processing the sequence features through the deep feature layer to obtain high-dimensional features includes: extracting primary features from the sequence features through the bottom layer; combining the primary features through the middle layer to obtain intermediate features; and transforming the intermediate features through the top layer to obtain high-dimensional features.

[0101] The bottom layer is the most basic layer in the deep feature layer. It is responsible for extracting the most basic, low-level features from the input data. In this step, the bottom layer extracts primary features related to user behavior from the sequence features (time-dependent features extracted by the LSTM layer). These primary features are usually simple patterns or basic laws in the data. The middle layer is located between the bottom layer and the top layer. It is responsible for more complex processing of the primary features extracted by the bottom layer. Specifically, the middle layer performs feature combination, that is, multiple primary features are correlated and combined to generate more meaningful features. This step combines simple features to form intermediate features that can better explain user behavior. For example, the user's diet and exercise features are combined to obtain their mutual influence. The top layer is the highest level of the deep feature layer. It is responsible for further processing of the intermediate features generated by the middle layer, which is called feature transformation. Feature transformation is a nonlinear transformation, induction, or further abstraction of features so that they can express more complex patterns or laws, and finally obtain high-dimensional features. These high-dimensional features contain the complex relationship between user behavior and health status, and are important inputs for model prediction. Primary features are basic features extracted from sequence features. They are the most primitive patterns or regularities extracted from data by the bottom layer. For example, the most direct data points such as the user's dietary intake and exercise volume are primary features. Feature combination is the process of processing primary features by the middle layer. It generates higher-level features by correlating and combining different primary features. For example, the user's dietary intake and exercise intensity can be combined to reflect their interaction in the user's health changes. Intermediate features are the results of feature combination. Compared with primary features, they are more complex and can describe the relationship between multidimensional data. Intermediate features provide a more in-depth description of behavior and health patterns. Feature transformation is the further processing of intermediate features by the top layer. Through feature transformation, the model can perform nonlinear transformation, induction and compression on features to generate high-dimensional features. High-dimensional features can more accurately reflect the complex patterns in user behavior and help the model make more accurate predictions.

[0102] First, the bottom layer extracts the most basic primary features from the sequence features extracted by the LSTM layer. These primary features are usually simple and direct user behavior data patterns or basic rules, such as the user's daily dietary intake or exercise duration. Then, it enters the middle layer, which combines these primary features, that is, associates different primary features with each other to generate more complex intermediate features, such as combining the features of diet and exercise to describe the interaction between them. Finally, the data enters the top layer, and through feature transformation, the intermediate features are processed and abstracted nonlinearly to generate more complex high-dimensional features. These high-dimensional features can describe the complex relationship of user behavior more deeply and contain richer information, which is ultimately used to guide the model to make more accurate nutritional intake predictions. Through the step-by-step processing of the bottom layer, middle layer, and top layer, the system can extract deep, high-dimensional features from basic user behavior information and improve the model's prediction ability.

[0103] As an example, the step of mapping the high-dimensional features into nutritional intake prediction values ​​through the fully connected layer includes: linearly transforming the high-dimensional features through matrix multiplication to obtain linear combination features; performing weight transformation and biasing on the linear combination features to obtain output features; and converting the output features into a form through an activation function to map them into nutritional intake prediction values.

[0104] Matrix multiplication refers to the process of multiplying high-dimensional features with the weight matrix in the network in the fully connected layer. This is a basic operation of neural networks, which is used to combine input data (high-dimensional features) with model parameters (weight matrix) to generate new feature representations. Through matrix multiplication, the model can combine multiple input features to form linear combination features. Linear transformation is the process of processing high-dimensional features through matrix multiplication, with the purpose of mapping these features to a new space. This transformation is linear, which means that it can maintain the weighted linear relationship of the input features. Through linear transformation, the model can generate a set of new features, which are weighted linear combinations of high-dimensional features. Linear combination features are the result of linear transformation. They are new features derived from high-dimensional features through matrix multiplication and linear transformation. Each feature is a weighted combination of high-dimensional features. They combine different information of the input and embody specific linear patterns. Linear combination features provide the basis for subsequent model processing. Weight transformation refers to adjusting the importance of input features by multiplying with the weight matrix. Bias is a constant term added to the result of weight transformation. The weight determines the contribution of each input feature to the output result. Bias is used to adjust the flexibility of the model. By weight transformation and adding bias terms to linear combination features, the model can fit the data more flexibly and improve the prediction accuracy. Output features are the results of weight transformation and bias processing. They represent the preliminary processing results of the model on the input data. These features have not been processed nonlinearly, but have combined important information of the input features, indicating the linear response of the model. Activation function is a mathematical function used to introduce nonlinearity in neural networks. Through activation function, the model can handle more complex nonlinear relationships. In this step, the activation function processes the output features so that the model can capture more complex patterns and relationships between data. Form conversion refers to converting the output features into numerical values ​​that are more in line with the actual situation after nonlinear processing through activation function. After form conversion, the output data can be better mapped to the user's nutritional intake prediction value, such as the daily recommended intake of calories, protein, fat and carbohydrates.

[0105] First, through matrix multiplication, the high-dimensional features are multiplied by the weight matrix and linearly transformed to generate a new set of features, called linear combination features. These linear combination features are weighted combinations of high-dimensional features and reflect the linear relationship of the input data. Next, the model performs weight transformation and bias on these linear combination features, that is, adjusts the weights of the features to determine the contribution of each feature to the result, and increases the flexibility and accuracy of the model by adding bias terms. The result of this step is called output features, which are linear responses to the input data. Then, the output features are processed nonlinearly through activation functions, which is called form transformation. This step enables the model to capture more complex nonlinear relationships. Finally, after processing with the activation function, the output features are mapped to nutritional intake predictions, that is, personalized recommendations for nutrients such as calories, protein, fat and carbohydrates required by users every day. Through this series of steps, the model can effectively transform high-dimensional features into personalized nutritional intake recommendations.

[0106] Step S32, using the historical user information to train the initial nutrition intake assessment model to obtain a reference nutrition intake assessment model;

[0107] It should be noted that the reference nutrition intake assessment model refers to the model obtained by training the initial nutrition intake assessment model with historical user information. After training, this model can capture the user's nutrition intake pattern based on a large amount of historical user data and provide a reliable basis for future nutrition assessment. It combines the user's past diet, exercise, health status and other data to adjust the model's parameters so that it can more accurately predict the user's nutritional needs.

[0108] It is understandable that, first of all, historical user information includes the user's past diet records, exercise data, health indicators, etc., and the system uses this data to train the initial model. During the training process, the model gradually adjusts its internal parameters (such as weights and biases) by repeatedly inputting historical data to learn the user's long-term behavior patterns and nutritional needs. In this way, the model continuously optimizes its predictive ability, enabling it to better capture the user's personalized needs and changes in health status. After the training is completed, the model can provide a more accurate assessment of nutritional intake and become a reference nutritional intake assessment model, that is, a model with strong predictive ability, providing a basis for subsequent personalized nutritional recommendations. Through this training process, the model can more accurately reflect the user's historical behavior, thereby improving the accuracy and personalization of the overall assessment.

[0109] As an example, the steps of using the historical user information to train the initial nutrition intake assessment model to obtain a reference nutrition intake assessment model include: obtaining a nutrition intake prediction value based on the historical user information and the initial nutrition intake assessment model; calculating a nutrition intake recommendation value based on the historical user information and the recommended nutrition intake range; calculating a loss value based on the nutrition intake prediction value, the nutrition intake recommendation value and a mean square error loss function; and using an adaptive moment estimation optimizer to adjust parameters of the initial nutrition intake assessment model until the loss value is less than a preset loss value, thereby obtaining a reference nutrition intake assessment model.

[0110] The predicted nutritional intake value refers to the estimated result of the user's daily nutritional intake generated by the initial nutritional intake assessment model after receiving historical user information. This includes the model's predicted daily intake of nutrients (such as calories, protein, fat, carbohydrates, etc.) required by the user. This predicted value is the model's preliminary judgment based on historical data, reflecting the impact of the user's past behavior on future nutritional needs. The recommended nutritional intake value is the ideal daily nutritional intake calculated based on historical user information and nutritional dietary standards. It is based on factors such as the user's health status and activity level, and is determined by nutritional guidelines or standards (such as the "Dietary Guidelines for Chinese Residents"). These recommended values ​​represent the nutritional intake goals that users should achieve under ideal conditions and are the reference standards for model adjustment. The mean square error (MSE) loss function is a commonly used function to measure the gap between the model's prediction results and the true value (recommended value). Specifically, it measures the error size of the model by calculating the square of the difference between the predicted value and the recommended value and then taking the average. The loss value is the result calculated by the mean square error loss function, which indicates the average error between the model prediction value and the recommended value. The smaller the loss value, the higher the prediction accuracy of the model; the larger the loss value, the less accurate the model prediction is. The goal of model training is to continuously reduce the loss value so that the prediction value is closer to the recommended value. The adaptive moment estimation optimizer is a commonly used gradient optimization algorithm (Adam, i.e. Adaptive Moment Estimation). It dynamically adjusts the learning rate of model parameters by calculating the momentum and variance of the gradient to adapt to different parameter update requirements. Compared with traditional optimizers, the Adam optimizer converges faster and is more stable, so it is widely used in training models. Its working principle is to automatically adjust model parameters (such as weights and biases) according to the change of loss value to minimize the loss value and optimize model performance. The preset loss value is a target error threshold set during the model training process. It represents an acceptable error range. That is, when the model's loss value drops below the threshold, the training stops and the model's prediction result is considered close enough to the recommended value. For example, the preset loss value may be set to 0.01, which means that when the model's prediction error is less than 0.01, the training is completed and the final reference nutrition intake assessment model is obtained. Suppose a user's daily recommended calorie intake is 2000 kcal, which is the recommended value for nutrition intake. The model initially predicts a calorie intake of 2200 kcal, which is the predicted value for nutrition intake. The loss value calculated by the mean square error loss function is 100 (indicating that the predicted value is far from the recommended value). Using the adaptive moment estimation optimizer, the model continuously adjusts its parameters until the loss value drops below the preset loss value of 0.01, and finally generates a more accurate reference nutrition intake assessment model with a predicted calorie intake close to 2000 kcal.

[0111] First, the system inputs the user's historical data (such as diet records, exercise information, etc.) into the initial model to generate a predicted nutritional intake value, that is, the model estimates the daily nutritional intake required by the user. Next, the system calculates the ideal nutritional intake recommendation value based on historical user information and nutritional dietary standards, which is the amount of nutrition that the user should consume under ideal conditions. Then, the system compares the difference between the predicted value and the recommended value through the mean square error loss function and calculates the loss value, which reflects the accuracy of the model's prediction results. Next, the adaptive moment estimation optimizer (Adam optimizer) is used to dynamically adjust the parameters of the model, continuously optimize the model, and gradually reduce the loss value. The training process continues until the loss value is less than the preset loss value, indicating that the prediction accuracy of the model has met the requirements. At this time, the model optimization is completed and the reference nutritional intake assessment model is obtained. Through this training process, the model can more accurately predict the user's nutritional needs and gradually improve its personalized assessment capabilities.

[0112] As an example, the step of obtaining a predicted nutritional intake value based on the historical user information and the initial nutritional intake assessment model includes: receiving the historical user information through the input layer, wherein the historical user information is obtained by processing non-sequential data including basic information and dietary taboo information, and sequence data including dietary record information, exercise information, and special service records; determining the time corresponding to each different type of sequence data in the historical user information, sorting the data according to the temporal dependency, and obtaining corresponding sequence data; abstracting the sequence data and learning deep features to obtain high-dimensional features; mapping the high-dimensional features to obtain predicted nutritional intake values, wherein the predicted nutritional intake values ​​include the intake of nutrients in calories, protein, fat, and carbohydrates.

[0113] Non-sequential data refers to static information that is not time-dependent and does not change over time. Specifically, non-sequential data includes basic information about users (such as age, gender, height, weight, etc.) and dietary information (such as taboos or preferences for certain foods). These data do not rely on time order in the model, and they are static information that describes the basic situation and dietary restrictions of users. Sequential data refers to dynamic data that changes over time and has time dependence. In this solution, sequence data includes user dietary record information (such as the types and amounts of food consumed daily), exercise information (such as the type and intensity of daily exercise), and special service records (such as health conditioning services for users, such as lowering blood pressure and controlling blood sugar). These data are arranged in chronological order and record the user's behavior and health status at different time points. Abstracting and learning deep features means that the model identifies high-level patterns and relationships hidden in the data by performing complex processing and feature extraction on the input sequence data. Through this deep learning, the model can capture long-term trends and complex associations in user behavior and health status. For example, by analyzing the user's long-term diet and exercise data, the model identifies how these behaviors affect the user's nutritional needs. High-dimensional features are multi-dimensional information abstracted from sequence data. They contain more complex and rich user behavior patterns and health change characteristics. After deep feature learning of the model, the system can extract a large number of highly correlated features from the original time series data, such as the interaction between the user's eating habits and health status. High-dimensional features are the core information for the model to make nutritional intake predictions. The nutritional intake prediction value is an estimated value of the user's daily nutritional needs generated by the model based on high-dimensional features, specifically including the amount of calories, protein, fat and carbohydrates that the user should consume daily. These predictions reflect the model's assessment of the user's personalized needs and aim to provide users with scientific nutritional intake guidance.

[0114] First, the input layer receives historical user information, which includes two types of data: non-sequential data (such as basic information and dietary taboos of users) and sequential data (such as dietary records, exercise information, and special service records of users). Non-sequential data is static information, while sequential data is dynamic information with time dependency. Then, the system sorts and processes these sequential data according to the time order corresponding to them and the time dependency relationship to ensure that the model can understand the time sequence of the data. Subsequently, the model abstracts and learns deep features from these sorted sequential data, that is, it identifies complex patterns and trends from these time-dependent behavioral data and extracts high-dimensional features that can reflect the user's long-term behavior and health status. Finally, the system maps these high-dimensional features into nutritional intake predictions, including the daily intake of calories, protein, fat, and carbohydrates required by the user. Through this process, the model can generate personalized nutritional intake recommendations for users and accurately assess their daily nutritional needs.

[0115] Step S33, adjusting the reference nutrition intake assessment model according to the nutrition diet standard, the current user information and the recommended nutrition intake range to obtain a target nutrition intake assessment model.

[0116] It should be noted that the target nutrition intake assessment model refers to the final model after dynamic adjustment and optimization, which can generate the most accurate and personalized nutrition intake recommendations based on the user's current health status, nutrition needs and dietary standards. It is based on the reference nutrition intake assessment model, combined with the user's current information (such as recent diet, exercise and health status), nutrition dietary standards (such as scientific nutrition guidelines), and nutrition recommended intake range (personalized nutrition intake upper and lower limits set for users).

[0117] It can be understood that, first of all, historical user information includes data such as the user's dietary intake, exercise records, and health indicators in the past period of time. These data are used as input to train the initial model. Through training, the model can learn the user's nutritional intake pattern from the user's historical behavior and capture the laws of dietary and health changes over a long period of time. During the training process, the model will continuously adjust its internal parameters so that it can more accurately predict the user's nutritional needs. Finally, the trained model is called the reference nutritional intake assessment model. Compared with the initial model, the reference model has been optimized and has better predictive and personalized processing capabilities, which can provide a more accurate basis for subsequent nutritional intake assessments.

[0118] As an example, the reference nutrition intake assessment model is adjusted according to the nutrition diet standard, the current user information and the recommended nutrition intake range to obtain a target nutrition intake assessment model, including the steps of: inputting the current user information into the reference nutrition intake assessment model to obtain a nutrition intake assessment result; when the nutrition intake assessment result is outside the recommended nutrition intake range, using the nutrition intake assessment result as feedback information, and adjusting the parameters of the reference nutrition intake assessment model according to the feedback information to obtain a target nutrition intake assessment model; when the nutrition intake assessment result is within the recommended nutrition intake range, using the reference nutrition intake assessment model as the target nutrition intake assessment model.

[0119] The nutritional intake assessment result refers to the predicted value of the user's daily nutritional intake generated by the model after the current user information is input into the reference nutritional intake assessment model. This includes specific nutrient intake recommendations (such as calories, protein, fat, carbohydrates, etc.). The assessment result provides personalized nutritional advice based on the user's current health status, diet record, exercise information, etc. Feedback information refers to a set of information generated by the system when the nutritional intake assessment result exceeds the recommended nutritional intake range. This information indicates that there is a deviation between the current model output and the preset nutritional target. The feedback information includes the difference between the model prediction result and the recommended range, which is used to guide model adjustment. It helps the model identify which parameters cause prediction errors and suggests how to improve them. Parameter adjustment refers to the process of modifying the internal parameters (such as weights, biases, etc.) of the reference nutritional intake assessment model based on feedback information. This process aims to optimize the model so that its output results are closer to the user's recommended nutritional intake range. By adjusting the model parameters, the system can gradually correct the prediction errors to ensure that the generated nutritional intake assessment results are more in line with the user's actual needs, and ultimately obtain a more accurate target nutritional intake assessment model.

[0120] First, the system inputs the current user information (such as the user's latest diet, exercise, and health status) into the reference nutrition intake assessment model. The model generates nutrition intake assessment results based on the user's historical behavior and current data, that is, predicts the user's daily nutrition intake. If the assessment result exceeds the preset recommended nutrition intake range, the system will use the assessment result as feedback information, which indicates the deviation of the model output from the ideal range. Based on this feedback information, the system adjusts the parameters of the model (such as weights and biases) and corrects the prediction error of the model until the model's prediction results meet the recommended range, and finally obtains a more accurate target nutrition intake assessment model. If the model's evaluation results are within the recommended range, it means that the model has achieved the expected accuracy and no further adjustment is required. The reference model is directly used as the target model. This process ensures that the model is dynamically optimized according to the user's personalized needs and provides accurate nutrition recommendations.

[0121] This embodiment constructs an initial nutrition intake assessment model based on the input layer, the long short-term memory network layer, the deep feature layer and the fully connected layer; the initial nutrition intake assessment model is trained using the historical user information to obtain a reference nutrition intake assessment model; the reference nutrition intake assessment model is adjusted according to the nutritional diet standard, the current user information and the recommended nutritional intake range to obtain a target nutrition intake assessment model. First, the model receives the user's historical and current data through the input layer, processes the time series information through the long short-term memory network layer (LSTM layer), extracts the long-term and short-term patterns in the user's behavior, and then further processes these features through the deep feature layer to generate complex high-dimensional features, and finally maps these features to preliminary nutrition intake prediction values ​​through the fully connected layer. This step establishes a basic initial nutrition intake assessment model that can comprehensively analyze the user's health behavior. Next, the system uses the user's historical information to train the model. The model adjusts the internal parameters to learn the user's long-term nutrition intake rules and form a reference nutrition intake assessment model. This step enables the model to accurately predict the user's nutritional needs. Finally, the current user information is input into the reference model, and dynamically adjusted according to the nutritional dietary standards and the recommended nutritional intake range. If the predicted result exceeds the recommended range, the model parameters are adjusted through the feedback mechanism until the predicted value meets the standard, and the final target nutritional intake assessment model is obtained. This step ensures that the model can adapt to the user's current status in real time, provide more accurate and personalized nutritional intake recommendations, and help users achieve health goals. Through this series of steps, the model is gradually optimized from basic analysis to precise adjustment to maximize the accuracy and personalization of user nutrition management.

[0122] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the personalized nutrition intake recommendation method of the present application. More simple transformations based on this technical concept are all within the scope of protection of the present application.

[0123] This application also provides a personalized nutrition intake recommendation device, please refer to Figure 3 , the personalized nutrition intake recommendation device comprises:

[0124] The data acquisition module 10 is used to obtain nutritional dietary standards, initial user information and recommended nutritional intake range;

[0125] A data processing module 20 is used to pre-process the initial user information to obtain target sequence set data, wherein the target sequence set data includes current user information;

[0126] A model training module 30 is used to train and adjust the initial nutrition intake assessment model according to the nutrition diet standard, the target sequence set data and the recommended nutrition intake range to obtain a target nutrition intake assessment model;

[0127] The nutrition intake recommendation module 40 is used to input the current user information into the target nutrition intake assessment model to obtain nutrition intake recommendations.

[0128] In one embodiment, the model training module 30 is also used to construct an initial nutrition intake assessment model based on the input layer, the long short-term memory network layer, the deep feature layer and the fully connected layer; use the historical user information to train the initial nutrition intake assessment model to obtain a reference nutrition intake assessment model; adjust the reference nutrition intake assessment model according to the nutritional diet standard, the current user information and the recommended nutrition intake range to obtain a target nutrition intake assessment model.

[0129] In one embodiment, the model training module 30 is also used to obtain a predicted nutritional intake value based on the historical user information and the initial nutritional intake assessment model; calculate the recommended nutritional intake value based on the historical user information and the recommended nutritional intake range; calculate the loss value based on the predicted nutritional intake value, the recommended nutritional intake value and the mean square error loss function; use an adaptive moment estimation optimizer to adjust the parameters of the initial nutritional intake assessment model until the loss value is less than the preset loss value, so as to obtain a reference nutritional intake assessment model.

[0130] In one embodiment, the model training module 30 is also used to receive the historical user information through the input layer, wherein the historical user information is obtained by processing non-sequential data including basic information and dietary taboo information and sequence data including dietary record information, exercise information and special service records; determining the time corresponding to each different type of sequence data in the historical user information, sorting the data according to the temporal dependency, and obtaining corresponding sequence data; abstracting the sequence data and learning deep features to obtain high-dimensional features; mapping the high-dimensional features to obtain nutritional intake prediction values, wherein the nutritional intake prediction values ​​include the intake of nutrients in calories, protein, fat and carbohydrates.

[0131] In one embodiment, the model training module 30 is also used to input the current user information into the reference nutrition intake assessment model to obtain a nutrition intake assessment result; when the nutrition intake assessment result is outside the recommended nutrition intake range, the nutrition intake assessment result is used as feedback information, and the reference nutrition intake assessment model is adjusted according to the feedback information to obtain a target nutrition intake assessment model; when the nutrition intake assessment result is within the recommended nutrition intake range, the reference nutrition intake assessment model is used as the target nutrition intake assessment model.

[0132] In one embodiment, the data acquisition module 10 is further used to calculate a body mass index based on the basic information; to obtain a weight status based on the body mass index; to calculate an energy consumption based on the activity mode and the activity duration; to obtain physical activity intensity based on the energy consumption and a preset activity intensity-energy consumption range; to obtain an energy coefficient based on the physical activity intensity, the weight status and a preset energy coefficient table; to calculate a normal daily calorie intake range based on the nutritional diet standard, the weight status, the physical activity intensity and the energy coefficient; when the special service record information is a body fat reduction service, the normal daily calorie intake range is adjusted according to the basal metabolic rate, the exercise coefficient, the fat reduction coefficient and the dietary taboo information to obtain a recommended nutritional intake range; when the special service record information is other services, the normal daily calorie intake range is adjusted according to the dietary taboo information and the special service record information to obtain a recommended nutritional intake range.

[0133] In one embodiment, the data processing module 20 is also used to delete invalid data in the initial user information, and process outlier data in the dietary taboo information and the dietary record information to obtain second historical user information; encode the basic information and the dietary taboo information in the second historical user information and convert them to binary column data; fill the missing information in the dietary record information, the exercise record information and the special service record information by using the adjacent average or weighted average to obtain continuous time series data; cut the continuous time series data into time window fragment data of fixed length; logarithmically transform the dietary record information, the exercise record information and the special service record information, and perform feature combination on the special service record information to obtain combined feature data; merge the continuous time series data, the time window fragment data and the dietary record information, the exercise record information and the special service record information in the combined feature data into a three-dimensional array, and perform normalization processing to obtain initial sequence set data; add the binary column data to each time step in the initial sequence set data to obtain target sequence set data.

[0134] The personalized nutrition intake recommendation device provided by the present application adopts the personalized nutrition intake recommendation method in the above-mentioned embodiment, which can solve the technical problem of how to conduct a phased nutrition intake assessment according to the actual situation of the individual. Compared with the prior art, the beneficial effects of the personalized nutrition intake recommendation device provided by the present application are the same as the beneficial effects of the personalized nutrition intake recommendation method provided by the above-mentioned embodiment, and the other technical features in the personalized nutrition intake recommendation device are the same as the features disclosed in the above-mentioned embodiment method, which will not be repeated here.

[0135] The present application provides a personalized nutrition intake recommendation device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the personalized nutrition intake recommendation method in the above-mentioned embodiment one.

[0136] Reference below Figure 4, which shows a schematic diagram of the structure of a personalized nutrition intake recommendation device suitable for implementing the embodiment of the present application. The personalized nutrition intake recommendation device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 4 The personalized nutrition intake recommendation device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0137] like Figure 4 As shown, the personalized nutrition intake recommendation device may include a processing device 1001 (e.g., a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the personalized nutrition intake recommendation device are also stored. The processing device 1001, ROM1002, and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the personalized nutrition intake recommendation device to communicate wirelessly or wired with other devices to exchange data. Although the figure shows a personalized nutrition intake recommendation device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or provided alternatively.

[0138] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0139] The personalized nutrition intake recommendation device provided by the present application adopts the personalized nutrition intake recommendation method in the above-mentioned embodiment, which can solve the technical problem of how to perform phased nutrition intake assessment according to the actual situation of the individual, and the beneficial effects are the same as the beneficial effects of the personalized nutrition intake recommendation method provided by the above-mentioned embodiment, and other technical features are also the same, which will not be repeated here. It should be understood that the various parts disclosed in the present application can be implemented with hardware, software, firmware or a combination thereof. In the description of the above-mentioned embodiments, specific features, structures, materials or characteristics can be combined in a suitable manner in any one or more embodiments or examples.

[0140] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, wherein the computer-readable program instructions are used to execute the personalized nutrition intake recommendation method in the above-mentioned embodiment.

[0141] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination of the above.

[0142] The above-mentioned computer-readable storage medium may be included in the personalized nutrition intake recommendation device; or it may exist independently without being assembled into the personalized nutrition intake recommendation device. The above-mentioned computer-readable storage medium carries one or more programs. When the above-mentioned one or more programs are executed by the personalized nutrition intake recommendation device, the personalized nutrition intake recommendation device: obtains the nutrition diet standard, the initial user information and the nutrition recommended intake range; pre-processes the initial user information to obtain the target sequence set data, and the target sequence set data includes the current user information; trains and adjusts the initial nutrition intake assessment model according to the nutrition diet standard, the target sequence set data and the nutrition recommended intake range to obtain the target nutrition intake assessment model; inputs the current user information into the target nutrition intake assessment model to obtain nutrition intake recommendations.

[0143] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0144] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0145] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0146] The readable storage medium provided in the present application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned personalized nutritional intake recommendation method, and can solve the technical problem of how to conduct phased nutritional intake assessment according to the actual situation of the individual. The beneficial effects are the same as those of the personalized nutritional intake recommendation method provided in the above-mentioned embodiment, and will not be repeated here.

[0147] The present application also provides a computer program product, including a computer program, which implements the steps of the personalized nutrition intake recommendation method as described above when executed by a processor. The computer program product provided by the present application can solve the technical problem of how to perform a phased nutrition intake assessment according to the actual situation of the individual. Compared with the prior art, the beneficial effects of the computer program product provided by the present application are the same as the beneficial effects of the personalized nutrition intake recommendation method provided in the above embodiment, which will not be repeated here.

[0148] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A personalized nutrition intake recommendation method, characterized in that: The method comprises: Obtain nutritional dietary standards, initial user information, and recommended nutritional intake ranges; Preprocessing the initial user information to obtain target sequence set data, wherein the target sequence set data includes current user information; The initial nutrition intake assessment model is trained and adjusted according to the nutrition diet standard, the target sequence set data and the recommended nutrition intake range to obtain a target nutrition intake assessment model; The current user information is input into the target nutrition intake assessment model to obtain nutrition intake recommendations.

2. The method according to claim 1, characterized in that The target sequence set data includes current user information and historical user information, and the steps of training and adjusting the initial nutrition intake assessment model according to the nutrition diet standard, the target sequence set data and the recommended nutrition intake range to obtain the target nutrition intake assessment model include: Construct an initial nutrition intake assessment model based on the input layer, long short-term memory network layer, deep feature layer, and fully connected layer; Using the historical user information to train the initial nutrition intake assessment model to obtain a reference nutrition intake assessment model; The reference nutrition intake assessment model is adjusted according to the nutrition diet standard, the current user information and the recommended nutrition intake range to obtain a target nutrition intake assessment model.

3. The method according to claim 2, characterized in that The step of using the historical user information to train the initial nutrition intake assessment model to obtain a reference nutrition intake assessment model comprises: Obtaining a predicted nutritional intake value according to the historical user information and the initial nutritional intake assessment model; Calculate the recommended nutritional intake value according to the historical user information and the recommended nutritional intake range; A loss value is calculated according to the predicted nutritional intake value, the recommended nutritional intake value and a mean square error loss function; The parameters of the initial nutrition intake assessment model are adjusted using an adaptive moment estimation optimizer until the loss value is less than a preset loss value, thereby obtaining a reference nutrition intake assessment model.

4. The method according to claim 3, characterized in that The step of obtaining a predicted nutritional intake value according to the historical user information and the initial nutritional intake assessment model comprises: The historical user information is received through the input layer, wherein the historical user information is obtained by processing non-sequential data including basic information and dietary taboo information and sequential data including dietary record information, exercise information and special service records; Determine the time corresponding to each different type of sequence data in the historical user information, and perform sorting processing according to the time dependency to obtain corresponding sequence data; Abstracting the sequence data and learning deep features to obtain high-dimensional features; The high-dimensional features are mapped to obtain a nutritional intake prediction value, wherein the nutritional intake prediction value includes the intake of nutrients in calories, protein, fat and carbohydrates.

5. The method according to claim 2, characterized in that The step of adjusting the reference nutrition intake assessment model according to the nutrition diet standard, the current user information and the recommended nutrition intake range to obtain a target nutrition intake assessment model comprises: Inputting the current user information into the reference nutrition intake assessment model to obtain a nutrition intake assessment result; When the nutrition intake assessment result is outside the recommended nutrition intake range, the nutrition intake assessment result is used as feedback information, and parameters of the reference nutrition intake assessment model are adjusted according to the feedback information to obtain a target nutrition intake assessment model; When the nutrition intake assessment result is within the recommended nutrition intake range, the reference nutrition intake assessment model is used as the target nutrition intake assessment model.

6. The method according to claim 1, characterized in that The initial user information includes basic information, activity mode, activity duration, dietary taboo information and special service record information. The step of obtaining the recommended nutritional intake range includes: Calculate the body mass index according to the basic information; Obtaining weight status according to the body mass index; Calculate the energy consumption according to the activity mode and the activity duration; Obtaining physical activity intensity according to the energy consumption and a preset activity intensity-energy consumption range; Obtaining an energy coefficient according to the physical activity intensity, the body weight status and a preset energy coefficient table; Calculate the normal daily calorie intake range according to the nutritional dietary standard, the body weight status, the physical activity intensity and the energy coefficient; When the special service record information is a body fat reduction service, the normal daily calorie intake range is adjusted according to the basal metabolic rate, exercise coefficient, fat reduction coefficient and the dietary taboo information to obtain a recommended nutritional intake range; When the special service record information is for other services, the normal daily calorie intake range is adjusted according to the dietary taboo information and the special service record information to obtain the recommended nutritional intake range.

7. The method according to claim 1, characterized in that The initial user information includes basic information of the user, dietary taboo information, dietary record information, exercise record information and special service record information. The step of preprocessing the initial user information to obtain target sequence set data includes: Deleting invalid data in the initial user information, and processing outlier data in the dietary taboo information and the dietary record information to obtain second historical user information; Encoding the basic information and the dietary taboo information in the second historical user information and converting them into binary column data; Filling the missing information in the diet record information, the exercise record information and the special service record information with the adjacent average value or weighted average value to obtain continuous time series data; Cutting the continuous time series data into time window segment data of fixed length; Performing logarithmic transformation on the diet record information, the exercise record information, and the special service record information, and performing feature combination on the special service record information to obtain combined feature data; The continuous time series data, the time window segment data, and the diet record information, the exercise record information, and the special service record information in the combined feature data are combined into a three-dimensional array, and normalized to obtain initial sequence set data; The binary column data is added to each time step in the initial sequence set data to obtain the target sequence set data.

8. A personalized nutrition intake recommendation device, characterized in that: The device comprises: A data acquisition module is used to obtain nutritional dietary standards, initial user information, and recommended nutritional intake ranges; A data processing module, used for preprocessing the initial user information to obtain target sequence set data, wherein the target sequence set data includes current user information; A model training module, used to train and adjust the initial nutrition intake assessment model according to the nutrition diet standard, the target sequence set data and the recommended nutrition intake range to obtain a target nutrition intake assessment model; The nutrition intake recommendation module is used to input the current user information into the target nutrition intake assessment model to obtain nutrition intake recommendations.

9. A personalized nutrition intake recommendation device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the personalized nutrition intake recommendation method according to any one of claims 1 to 7.

10. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the personalized nutrition intake recommendation method as described in any one of claims 1 to 7 are implemented.

Images