A dietary information push method and system based on big data

Abstract

The present invention discloses a method and system for pushing dietary information based on big data, and relates to the technical field of dietary information pushing, including obtaining the user's dietary, exercise and sleep data through the user's portable smart device, and performing preliminary fusion to obtain a ternary data snapshot; performing multi-dimensional cleaning and standardization processing on the user's ternary data snapshot, and outputting standardized ternary data; based on the user's standardized ternary data, using a machine learning algorithm to predict the user's ternary features; combining the predicted ternary features and the standardized ternary data to calculate the user's comprehensive health score; generating a personalized dietary recommendation plan for the user based on the user's comprehensive health score. The present invention achieves effective prediction of future health status by predicting the user's future health characteristics using a machine learning algorithm.

Description

Technical Field

[0001] The present invention relates to the technical field of dietary information push, and in particular to a dietary information push method and system based on big data. Background Art

[0002] With the rapid development of information technology, especially the popularization of big data and artificial intelligence technology, the field of dietary health has ushered in unprecedented changes. Traditionally, dietary advice mainly relies on the professional knowledge and personal experience of nutritionists. Although this method is effective, it has problems such as lack of personalization and slow response.

[0003] In recent years, with the widespread use of smart devices (such as smartphones, smart bracelets, etc.), users can easily record their diet, exercise and sleep data, which provides the possibility of realizing personalized health management based on big data. At present, the technologies for health management and diet recommendations based on personal health data collected by smart devices mostly focus on single-dimensional data analysis, such as only considering diet or exercise data, and lack effective integration of multi-source heterogeneous data. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a dietary information push method based on big data to solve the problem of lack of effective integration of multi-source heterogeneous data.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a method for pushing dietary information based on big data, which comprises:

[0008] Through the user's portable smart device, the user's diet, exercise and sleep data are obtained and initially integrated to obtain a three-dimensional data snapshot;

[0009] Perform multi-dimensional cleaning and standardization on the user's triple data snapshots and output standardized triple data;

[0010] Based on the standardized ternary data of the user, the machine learning algorithm is used to predict the user's ternary features;

[0011] The predicted ternary features and the standardized ternary data are combined to calculate the user's comprehensive health score;

[0012] Generate personalized diet recommendations for users based on their comprehensive health scores.

[0013] As a preferred solution of the method for pushing dietary information based on big data of the present invention, the dietary, exercise and sleep data of the user are obtained through the user's portable smart device, and preliminary fusion is performed to obtain a triple data snapshot. The specific steps are as follows:

[0014] Collect users’ dietary data by taking photos of food, describing it verbally, and entering it manually using smart devices;

[0015] The dietary data of users, which are recorded through taking photos, verbal descriptions and manual input, are stored as structured records to form a single meal record;

[0016] Through the smart devices carried by users, the user's exercise data is collected and converted into a unified format to record the exercise type, time and energy consumption to form a single-day exercise record;

[0017] The sleep monitoring device collects the user's sleep data and calculates the user's sleep quality score based on the sleep data;

[0018] Store the user's sleep data and sleep quality score as a structured record to form a single-day sleep record

[0019] Based on a unified time series format, the daily meal records, daily exercise records, and daily sleep records are timestamped and aligned to obtain a triple metadata snapshot.

[0020] As a preferred solution of the method for pushing dietary information based on big data of the present invention, wherein: the user's ternary data snapshot is cleaned and standardized in multiple dimensions to output standardized ternary data, and the specific steps are as follows:

[0021] Perform integrity check and missing value completion on the diet data, exercise data, and sleep data in the triple data snapshot respectively;

[0022] Filter and correct the noise and outliers in the diet data, exercise data and sleep data in the ternary data;

[0023] The diet data, exercise data and sleep data in the denoised ternary data are standardized respectively and stored in a unified time series format to obtain standardized ternary data.

[0024] As a preferred solution of the method for pushing dietary information based on big data of the present invention, the method uses a machine learning algorithm to predict the ternary features of the user based on the standardized ternary data of the user, and the specific steps are as follows:

[0025] The nutrition and total energy intake values, exercise energy consumption values, and sleep quality scores in the historical standardized ternary data are used as the feature value sequence of the health graph node and stored as a node feature matrix;

[0026] constructing edges between health graph nodes according to their physiological dependencies;

[0027] The Pearson correlation coefficient is used to calculate the linear correlation between the nodes in the health graph, which is used as the weight of the edge between the nodes in the health graph and stored in the adjacency matrix. The expression is as follows:

[0028] ;

[0029] in, Health graph node and The weight of the edge between is the number of samples of the characteristic value sequence of the health graph nodes, and Health graph nodes and The corresponding eigenvalue sequence is and Health graph nodes and The mean of the eigenvalue sequence of and Health graph nodes and The standard deviation of the eigenvalue sequence of

[0030] Connect the daily health graphs through time series to obtain a dynamic health graph;

[0031] The dynamic health graph is input into the graph neural network, and the node feature matrix in the dynamic health graph is updated in the graph convolution layer. The expression is:

[0032] ;

[0033] in, For the The node feature matrix of the layer, is the activation function, is the adjacency matrix, For graph neural network The weight matrix of the layer, For the The node feature matrix of the layer is initially composed of the original nutrition and total energy intake value, exercise energy consumption value, and sleep quality score;

[0034] Through the temporal attention mechanism, the node feature matrix at each time point is linearly transformed and projected into query vector, key vector and value vector;

[0035] The importance weight is obtained by calculating the product of the query vector and the key vector. The expression is as follows:

[0036] ;

[0037] in, is the importance weight, For time point The query vector is Indicates time point The key vector of Power, is the number of time points, is the dimension of the key vector, Express Find the square root, Indicates normalizing the weights to a probability distribution;

[0038] Apply the importance weights to the value vector to obtain a time-weighted node feature matrix, and integrate the time-weighted node feature matrices of all time points to obtain a comprehensive node feature matrix;

[0039] The comprehensive node feature matrix is ​​input into the fully connected layer, and the output is the predicted user's daily nutrition and total energy intake value, exercise energy expenditure value and sleep quality score.

[0040] As a preferred solution of the method for pushing dietary information based on big data of the present invention, the following specific steps are used to calculate the comprehensive health score of the user by combining the predicted ternary features and the standardized ternary data:

[0041] Calculate the user's daily nutritional and total energy recommended intake values ​​based on the user's age, gender, weight and height, and with reference to existing nutritional intake guidelines;

[0042] By calculating the difference between the predicted daily nutrition and total energy intake values ​​of the user in the standardized ternary data and the recommended daily nutrition and total energy intake values, the health scores of all nutrients and total energy are obtained, and then summarized to obtain the dietary health score;

[0043] Calculate the user's daily recommended exercise energy consumption value based on the user's age, gender and weight and referring to the existing exercise guidelines;

[0044] By comparing the predicted daily exercise energy consumption value of the user with the daily recommended exercise energy consumption value of the user, the exercise health score is obtained;

[0045] The dietary health score, exercise health score and predicted user sleep quality score are aggregated to obtain a comprehensive health score.

[0046] As a preferred solution of the method for pushing dietary information based on big data of the present invention, the specific steps of generating a personalized dietary recommendation plan for a user according to the user's comprehensive health score are as follows:

[0047] Access to the user's health information and health restrictions;

[0048] Calculate the frequency of occurrence of each type of food in the historical (past seven days) standardized triple data;

[0049] Obtain the nutrient content and energy supply value of food in the historical standardized ternary data from the food database, and compare them with the user's daily nutrient and total energy recommended intake values ​​to obtain the food nutrition score;

[0050] According to the existing health guidelines' recommendations on nutrition and energy intake, physical activity, and sleep, health thresholds for dietary health scores, physical activity health scores, and sleep quality scores are set respectively;

[0051] Based on the comparison results of the dietary health score, exercise health score and sleep quality score with the corresponding health thresholds, dietary recommendations are made according to the frequency of occurrence of each type of food in the historical standardized ternary data and the food nutrition score from high to low;

[0052] When the user's geographical location changes, the external data interface of the smart device is used to obtain local specialty food recommendations;

[0053] Dynamically push diet plans based on the user's real-time situation.

[0054] As a preferred solution of the method for pushing dietary information based on big data of the present invention, the specific steps of dynamically pushing the dietary plan in combination with the real-time situation of the user are as follows:

[0055] Based on the user's historical standardized triple data of daily meal times and frequency, a complete diet recommendation plan is pushed before meals;

[0056] Based on the time and duration of daily exercise in the user's historical standardized ternary data, after exercise, a recommended diet plan for physical recovery is pushed in real time;

[0057] Based on the daily sleep time in the user's historical standardized ternary data, combined with the user's sleep health score and the corresponding health threshold, determine whether special adjustments are needed;

[0058] When the sleep quality score is lower than the corresponding health threshold, it is recommended, otherwise it is not recommended.

[0059] In a second aspect, the present invention provides a diet information push system based on big data, including a data collection module, a data standardization module, a feature prediction module, a health scoring module and a diet recommendation module;

[0060] The data acquisition module is used to obtain the user's diet, exercise and sleep data through the user's portable smart device, and perform preliminary fusion to obtain a three-dimensional data snapshot;

[0061] The data standardization module is used to perform multi-dimensional cleaning and standardization processing on the user's ternary data snapshot and output standardized ternary data;

[0062] The feature prediction module is used to predict the user's ternary features using a machine learning algorithm based on the user's standardized ternary data;

[0063] The health scoring module is used to calculate the user's comprehensive health score by combining the predicted ternary features and the standardized ternary data;

[0064] The diet recommendation module is used to generate a personalized diet recommendation plan for the user based on the user's comprehensive health score.

[0065] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the method for pushing dietary information based on big data as described in the first aspect of the present invention is implemented.

[0066] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the method for pushing dietary information based on big data as described in the first aspect of the present invention is implemented.

[0067] The beneficial effects of the present invention are as follows: the present invention collects the user's diet, exercise and sleep data through smart devices, and performs preliminary fusion to form a ternary data snapshot, and uses machine learning algorithms to predict the user's future health characteristics, thereby achieving effective prediction of future health status, and providing users with forward-looking health management plans. The comprehensive health score is calculated in combination with the prediction results, which not only allows users to clearly understand their own health status, but also provides a basis for formulating personalized diet recommendations. The personalized diet recommendation plan takes into account the user's health information and personal preferences, ensures the pertinence and practicality of the diet advice, helps to improve the user's health behavior, and dynamically pushes the diet plan according to the user's real-time situation, optimizes the daily diet decision-making process, and helps users make healthy diet choices at the right time, ultimately achieving the purpose of improving the quality of life and promoting a healthy lifestyle. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0069] Figure 1 This is a flow chart of the method for pushing dietary information based on big data in Example 1.

[0070] Figure 2 This is a module diagram of the diet information push system based on big data in Example 1. DETAILED DESCRIPTION

[0071] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0072] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0073] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0074] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, and provides a method for pushing dietary information based on big data, comprising the following steps:

[0075] S1. Obtain the user's diet, exercise and sleep data through the user's portable smart device, and perform preliminary fusion to obtain a three-dimensional data snapshot.

[0076] Users collect their dietary data by taking photos of food, describing it verbally, and entering it manually using smart devices;

[0077] Specifically, users upload food photos through smartphone cameras, and the smartphone’s built-in image recognition module identifies the type (e.g., apples, rice) and amount of food in the food photos;

[0078] The actual weight of the food is calculated by combining the size of the two-dimensional image of the food with the standard size in the reference database, and the nutritional category and corresponding content of the food are obtained through the food database of the smartphone;

[0079] Users describe the dishes they have eaten by voice (e.g., "I ate 100 grams of chicken and 200 grams of salad today"), which is then converted into text using the smartphone's voice recognition technology. Natural language processing technology is then used to extract key information, including the name, weight, and quantity of the food, so that the nutrient category and corresponding content of the food can be obtained through the smartphone's food database;

[0080] Users can manually input unrecognized special foods (such as homemade dishes) through the smartphone terminal, which will automatically match the nutritional composition data of similar foods in the food database and allow users to make fine adjustments;

[0081] The dietary data of users, which are recorded through taking photos, verbal descriptions and manual input, are stored as structured records to form a single meal record;

[0082] The user's exercise data is collected through the smart device carried by the user and converted into a unified format. The exercise type, time and energy consumption are recorded in minutes to form a single-day exercise record;

[0083] Specifically, the smart bracelet captures the user's exercise type and duration through the built-in accelerometer and gyroscope, and combines with the GPS module to record the exercise distance and path. The collected data includes exercise type, exercise duration (minutes) and exercise energy consumption value (kcal);

[0084] The sleep monitoring device collects the user's sleep data and calculates the user's sleep quality score based on the sleep data;

[0085] Specifically, a smart bracelet or mattress sensor is used to monitor the user's sleep data, including the duration of deep sleep (minutes), the duration of light sleep (minutes), the number of sleep interruptions, the time of going to bed, the time of falling asleep and the time of waking up (inferred through the built-in pressure sensor and heart rate detection module of the device);

[0086] Combined with the user's sleep data, the sleep quality score is calculated using a multi-factor weighted model. The expression is as follows:

[0087] ;

[0088] in, Rate your sleep quality. For the duration of deep sleep, For light sleep duration, is the number of sleep interruptions, The time span from getting ready for bed to falling asleep;

[0089] Store the user's sleep data and sleep quality score as a structured record to form a single-day sleep record

[0090] Based on a unified time series format, the daily meal records, exercise records, and sleep records are timestamped at the minute level to obtain a triple metadata snapshot.

[0091] S2. Perform multi-dimensional cleaning and standardization on the user's triple data snapshot and output standardized triple data.

[0092] Perform integrity check and missing value completion on the diet data, exercise data, and sleep data in the triple data snapshot respectively;

[0093] Specifically, check whether there is data without indicating the type of food or amount in the diet record entered by the user. For example, if the user only enters salad without indicating its weight or ingredients, match the common standard ingredients of salad (such as lettuce, cucumber, carrot) and the corresponding proportions through the food database, and complete it according to the default amount;

[0094] If the exercise type or duration is missing in the exercise record, the exercise type and duration can be inferred by combining the heart rate fluctuation data;

[0095] A heart rate during exercise that exceeds 80% of the heart rate before exercise is defined as high-intensity exercise (such as running, cycling, etc.), and a heart rate that does not exceed 80% is defined as low-intensity activity (such as walking, yoga, etc.);

[0096] The time span from the increase in heart rate to the return to the initial level was determined as the exercise duration;

[0097] Missing sleep onset times or sleep interruptions are filled in using linear interpolation. For example, if there are missing time intervals in the user's sleep interruption data, the average value of the previous and next time points is used to fill in the missing time intervals.

[0098] After completing the missing value filling, the noise and outliers of the diet data, exercise data, and sleep data in the ternary data are filtered and corrected;

[0099] Specifically, detect abnormal calorie intake values ​​in a single meal record (such as the calorie intake of a single meal exceeds twice the user's daily recommended calorie intake), and correct the detected abnormal calorie intake values ​​using the average intake correction method, that is, the daily recommended calorie intake is divided by the number of meals per day;

[0100] Detect abnormal energy consumption values ​​in exercise data (such as 1000 kcal consumed in 10 minutes of running), and correct the energy consumption of exercise through the exercise type and duration matching model. The expression is:

[0101] ;

[0102] in, is the energy consumption value of exercise (kcal), Metabolic equivalents are determined based on the type of exercise. is the user's weight, is the duration of exercise (hours);

[0103] The standard sleep quality score is calculated based on a sleep time of 480 minutes per sleep session (including 100 minutes of deep sleep and 380 minutes of light sleep), a sleep-onset speed of 10 minutes, and no sleep interruptions. An abnormal sleep quality score is defined as one that exceeds the standard sleep score by 30% or is lower than the standard sleep quality score by 50%. The abnormal sleep quality score is smoothed using a time series smoothing method (such as a moving average method).

[0104] The diet data, exercise data, and sleep data in the denoised ternary data are standardized and converted into a unified time series format. The data are recorded in minutes to ensure the consistency of data dimensions and stored in a unified time series format to obtain standardized ternary data.

[0105] Split the user's daily intake data into minute-level data, and record the average energy intake and main nutrients (such as protein, fat, and carbohydrates) per minute during the meal;

[0106] Split the user's daily exercise data into minute levels, and record the exercise type and energy consumption every minute during exercise;

[0107] The user's sleep data is split into minutes, recording the sleep state (deep sleep, light sleep or interrupted) and the corresponding sleep quality score every minute.

[0108] S3. Based on the user's standardized ternary data, a machine learning algorithm is used to predict the user's ternary features.

[0109] The nutrition and total energy intake values, exercise energy consumption values, and sleep quality scores in the historical (past seven days) standardized triple data are used as the characteristic value sequence of the health graph node (such as the mean and variance of the past seven days) and stored as a node feature matrix;

[0110] constructing edges between health graph nodes according to their physiological dependencies;

[0111] Specifically, nutrition and total energy intake values ​​directly determine whether the energy expenditure during exercise is sufficient. For example, consuming too much high-calorie food may lead to decreased exercise efficiency, or the need to supplement carbohydrates to restore energy after exercise;

[0112] High-intensity exercise can affect sleep quality. For example, scientific studies have shown that moderate exercise can promote deep sleep, but excessive exercise may cause excessive fatigue in the body, thus causing sleep disruption.

[0113] Sleep quality can affect appetite levels. For example, lack of sleep may lead to an increase in hunger hormones (such as ghrelin) and suppress satiety hormones (such as leptin), thereby triggering overeating and causing excessive nutrient and total energy intake.

[0114] The Pearson correlation coefficient is used to calculate the linear correlation between the health graph nodes as the weight of the edge between the health graph nodes and stored in the adjacency matrix to represent the strength of the relationship between each health graph node. The expression is as follows:

[0115] ;

[0116] in, Health graph node and The weight of the edge between them, i.e. the health graph node and The strength of the relationship, is the number of samples of the characteristic value sequence of the health graph nodes (standardized ternary data for 7 days), and Health graph nodes and The corresponding feature value sequence (diet calorie intake, exercise energy consumption and sleep quality score for the past 7 days), and Health graph nodes and The mean of the eigenvalue sequence of and Health graph nodes and The standard deviation of the eigenvalue sequence of

[0117] Connect the daily health graphs according to the time series to obtain a dynamic health graph, and update the adjacency matrix through the Pearson correlation coefficient to reflect the dynamic changes in the relationship strength between the health graph nodes in the time dimension;

[0118] The dynamic health graph is input into the graph neural network, and the feature sequence values ​​of the nodes in the dynamic health graph are updated in the graph convolution layer. The expression is:

[0119] ;

[0120] in, For the The node feature matrix of the layer, is the activation function (such as ReLU), is the adjacency matrix, For graph neural network The weight matrix of the layer, For the The node feature matrix of the layer is initially composed of the original nutrition and total energy intake value, exercise energy consumption value, and sleep quality score;

[0121] Through the temporal attention mechanism, the node feature matrix at each time point (every day) is linearly transformed and projected into query vector, key vector and value vector;

[0122] The importance weight is obtained by calculating the product of the query vector and the key vector. The expression is as follows:

[0123] ;

[0124] in, is the importance weight, For time point The query vector is Indicates time point The key vector of Power, is the number of time points, such as the historical (past seven days) standardized triple data, then , is the key vector The dimension of Express Find the square root, It means normalizing the weights to a probability distribution so that the weights at all time points add up to 1;

[0125] Apply the importance weights to the value vector to obtain the time-weighted node feature matrix, and integrate the time-weighted node feature matrices of all time points to obtain the comprehensive node feature matrix, which is expressed as follows:

[0126] ;

[0127] in, is the comprehensive node feature matrix, For time point A vector of values ​​of ;

[0128] The comprehensive node feature matrix is ​​input into the fully connected layer, and the predicted daily nutrition and total energy intake value, exercise energy consumption value and sleep quality score of the user are output and summarized into ternary features;

[0129] Specifically, based on the multi-task learning method, different output heads of the fully connected layer output the predicted user's daily nutrition and total energy intake, exercise energy consumption, and sleep quality score, respectively. The expressions are as follows:

[0130] ;

[0131] ;

[0132] ;

[0133] in, To predict the user's daily nutrition and total energy intake, is the predicted daily exercise energy consumption value of the user, To predict the user's sleep quality score, , , are the weight matrices for predicting the user's daily nutrition and total energy intake, exercise energy consumption, and sleep quality score output heads, respectively. , , They are bias items for predicting the user's daily nutrition and total energy intake, exercise energy expenditure, and sleep quality score output heads respectively.

[0134] S4. Combine the predicted ternary features and the standardized ternary data to calculate the user's comprehensive health score.

[0135] According to the user's age, gender, weight and height, and referring to the existing nutritional intake guidelines (such as WHO daily nutritional recommendations), the user's daily nutritional and total energy recommended intake values ​​are calculated as follows:

[0136] ;

[0137] in, Recommended daily nutritional and total energy intake values ​​for users. is the user's weight, is the user's height, is the user's age, is the gender constant, with +5 for males and -161 for females;

[0138] By calculating the difference between the predicted daily nutrition and total energy intake values ​​of the user in the ternary features and the recommended daily nutrition and total energy intake values, the health scores of all nutrients and total energy are obtained, and then summarized to obtain the dietary health score, which is expressed as follows:

[0139] ;

[0140] in, Score the healthiness of your diet. The number of nutrients and total energy, To predict the user’s Nutritional or total energy intake, For the Recommended intake values ​​for nutrients or total energy, It is the index of the number of nutrient types and total energy;

[0141] Calculate the user's recommended daily exercise energy expenditure based on the user's age, gender and weight, and refer to existing exercise guidelines (such as 150 minutes of moderate-intensity exercise per week recommended by the WHO);

[0142] ;

[0143] in, The recommended daily exercise energy expenditure value for users is expressed as the metabolic equivalent of exercise (the MET value for moderate-intensity exercise is usually 4-6). is the user's weight, The WHO recommends 2.5 to 5 hours of moderate-intensity exercise per week.

[0144] By comparing the predicted daily exercise energy consumption value of the user in the ternary features with the daily recommended exercise energy consumption value of the user, the exercise health score is obtained. The expression is as follows:

[0145] ;

[0146] in, Score your sports health. To predict the user's daily exercise energy consumption value, Recommend daily exercise energy consumption values ​​for users;

[0147] The dietary health score, exercise health score and sleep quality score predicted by the ternary features are summarized to obtain a comprehensive health score.

[0148] S5. Generate a personalized diet recommendation plan for the user based on the user's comprehensive health score.

[0149] Obtain the user's health information, including age, gender, weight, height, and health restrictions (such as chronic diseases, allergies, dietary preferences and taboos, etc.);

[0150] Calculate the frequency of occurrence of each type of food in the historical (past seven days) standardized triple data, that is, the ratio of the number of occurrences of each type of food to the total number of occurrences of all types of food;

[0151] The nutrient content and energy supply value of the food in the historical standardized triple data are obtained from the food database, and compared with the user's daily nutrient and total energy recommended intake values ​​to obtain the food nutrition score, which is expressed as:

[0152] ;

[0153] in, Give foods a nutritional score, The number of nutrients and total energy, For the Nutritional or total energy intake, For the Recommended intake values ​​for nutrients or total energy;

[0154] According to the existing health guidelines' recommendations on nutrition and energy intake, physical activity, and sleep, health thresholds for dietary health scores, physical activity health scores, and sleep quality scores are set respectively;

[0155] For example, 90% of the dietary health score, exercise health score, and sleep quality score obtained by the existing health guidelines is the health threshold;

[0156] Based on the comparison results of the dietary health score, exercise health score and sleep quality score with the corresponding health thresholds, dietary recommendations are made from high to low according to the frequency of occurrence of each type of food in the historical (past seven days) standardized ternary data and the food nutrition score;

[0157] Specifically, when the user's diet health score, exercise health score, and sleep quality score are all higher than the corresponding health thresholds, it means that the user is currently in good physical condition, and food recommendations can be made based on the food preference score or food nutrition score. When the user's diet health score, exercise health score, and sleep quality score are all lower than the corresponding health thresholds, it means that the user's historical eating habits are not good. At this time, diet recommendations are made for the user in accordance with existing health guidelines, while avoiding health restrictions on the user (such as allergens, dietary taboos due to illness, etc.);

[0158] For example, if the user has diabetes, low GI (glycemic index) foods are recommended first, or if the user has high blood pressure, reduce foods with high sodium content (such as pickled foods), etc.

[0159] For example, if the sleep quality score is lower than the healthy threshold, after avoiding restrictions such as the user's allergens, foods that help sleep, such as milk and nuts, can be recommended to the user;

[0160] When the user's geographical location changes, the external data interface of the smart device is used to obtain local specialties for recommendation while avoiding the user's health restrictions (such as allergens, dietary restrictions due to illness, etc.);

[0161] Dynamically push diet plans based on the user's real-time context (such as time, location, and device terminal);

[0162] Based on the user's historical standardized triple data of daily meal times and frequency, a complete diet recommendation plan is pushed before meals;

[0163] Based on the time and duration of daily exercise in the user's historical standardized ternary data, after exercise, a recommended diet plan for physical recovery is pushed in real time;

[0164] Based on the daily sleep time in the user's historical standardized ternary data, combined with the user's sleep health score and the corresponding health threshold, determine whether special adjustments are needed;

[0165] When the sleep quality score is lower than the corresponding health threshold, it is recommended, otherwise it is not recommended;

[0166] Specifically, one hour before a meal, a complete diet recommendation plan is pushed through the user's commonly used devices (such as mobile phones and smart watches), including recommended dishes, required ingredients, nutritional components and cooking methods, to help users plan their diet in advance;

[0167] Push high-protein food recommendations in real time after exercise to help users recover quickly. If the user is outdoors, a concise list of high-protein foods will be pushed through the smart watch. If the user is at home, specific dishes will be recommended through the mobile phone or smart speaker.

[0168] When the sleep quality score is lower than the corresponding health threshold, foods rich in sleep-inducing ingredients (such as tryptophan and magnesium) are recommended for dinner, and users are prompted to avoid consuming caffeinated beverages or high-sugar foods.

[0169] The present embodiment also provides a diet information push system based on big data, including: a data acquisition module, a data standardization module, a feature prediction module, a health scoring module and a diet recommendation module; the data acquisition module is used to obtain the user's diet, exercise and sleep data through the user's portable smart device, and perform preliminary fusion to obtain a ternary data snapshot; the data standardization module is used to perform multi-dimensional cleaning and standardization processing on the user's ternary data snapshot, and output standardized ternary data; the feature prediction module is used to predict the user's ternary features based on the user's standardized ternary data using a machine learning algorithm; the health scoring module is used to combine the predicted ternary features and the standardized ternary data to calculate the user's comprehensive health score; the diet recommendation module is used to generate a personalized diet recommendation plan for the user based on the user's comprehensive health score.

[0170] This embodiment also provides a computer device, which is suitable for the case of a diet information push method based on big data, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the diet information push method based on big data proposed in the above embodiment.

[0171] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0172] The present embodiment also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for pushing dietary information based on big data as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable red-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, disk or optical disk.

[0173] In summary, the present invention achieves effective prediction of future health status through: smart devices collect users' diet, exercise and sleep data, and perform preliminary fusion to form a ternary data snapshot, and uses machine learning algorithms to predict users' future health characteristics, thereby providing users with forward-looking health management plans, and calculating comprehensive health scores based on the prediction results, which not only allows users to clearly understand their own health conditions, but also provides a basis for formulating personalized diet recommendations. The personalized diet recommendation plan takes into account the user's health information and personal preferences, ensures the pertinence and practicality of the diet advice, helps to improve the user's health behavior, and dynamically pushes the diet plan according to the user's real-time situation, optimizes the daily diet decision-making process, and helps users make healthy diet choices at the right time, ultimately achieving the purpose of improving the quality of life and promoting a healthy lifestyle.

[0174] Example 2, referring to Table 1, is the second example of the present invention. To further verify the technical solution of the present invention, experimental simulation data of a dietary information push method based on big data is provided.

[0175] In order to compare the effects of the present invention with existing health management applications on the market (such as My Fitness Pal), all participants used My Fitness Pal to record their daily activities in the first two weeks, and switched to the method of the present invention in the next two weeks. During this period, the researchers regularly checked the data upload status and understood the user experience through questionnaires.

[0176] This experiment aims to verify the advantages of a diet information push method based on big data over existing technologies. This method obtains users' diet, exercise and sleep data through smart devices, and generates personalized diet recommendations after processing. The experiment selected 30 volunteers aged between 25 and 45 years old to participate. Each participant was equipped with a monitoring system consisting of a smartphone, a smart bracelet and a mattress sensor.

[0177] First, to ensure the accuracy of data collection, all participants received a week of basic training to learn how to use smart devices to take photos of food, voice descriptions, and manually input dietary information. At the same time, in order to evaluate the relationship between exercise type and energy consumption, personalized heart rate interval setting guidance was also provided. For sleep monitoring, participants were required to wear smart bracelets or use mattress sensors for at least 7 hours every night to ensure the integrity of the data.

[0178] During the data collection stage, every morning the system will automatically generate a snapshot of the previous day's three-dimensional data, and clean and standardize the data. Next, it will use machine learning algorithms to predict the user's health characteristics and calculate a comprehensive health score based on this. Finally, based on personal health status and preferences, the system will automatically generate a personalized dietary recommendation. During this process, special attention is paid to data differences between different individuals, such as the influence of factors such as gender, age, and weight.

[0179] The details are shown in Table 1 below:

[0180] Table 1 Experimental data comparison table

[0181]

[0182] Through the analysis of the above data, it can be seen that the dietary information push method of the present invention can more effectively control the user's total energy intake while maintaining nutritional balance. For example, after using the present invention, the total energy intake of user A and user B is 2100 kcal and 1900 kcal respectively, and the ratio of protein, fat and carbohydrates is more reasonable, which helps to maintain a healthy weight and good physical condition.

[0183] In contrast, although users C and D can also record dietary information when using the existing health management application MyFitnessPal, the lack of personalized dietary advice leads to a higher total energy intake (2300 kcal and 2200 kcal, respectively), and a high ratio of fat and carbohydrates, which may not be conducive to health maintenance in the long run.

[0184] Further analysis shows that the present invention not only focuses on the intake of individual nutrients, but also takes into account the interactions between them and their overall impact on human health. For example, by optimizing the ratio of protein, fat and carbohydrates, it can better meet the differentiated nutritional needs of individuals, promote metabolism and enhance immunity. In addition, the present invention can also dynamically adjust dietary recommendations based on the user's real-time situation, which is one of the functions that is difficult to achieve with existing technologies.

[0185] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for pushing dietary information based on big data, characterized in that: include, Through the user's portable smart device, the user's diet, exercise and sleep data are obtained and initially integrated to obtain a three-dimensional data snapshot; Perform multi-dimensional cleaning and standardization on the user's triple data snapshots and output standardized triple data; Based on the user's standardized ternary data, the machine learning algorithm is used to predict the user's ternary features. The specific steps are as follows: The nutrition and total energy intake values, exercise energy consumption values, and sleep quality scores in the historical standardized ternary data are used as the feature value sequence of the health graph node and stored as a node feature matrix; constructing edges between health graph nodes according to their physiological dependencies; The Pearson correlation coefficient is used to calculate the linear correlation between the nodes in the health graph, which is used as the weight of the edge between the nodes in the health graph and stored in the adjacency matrix; Connect the daily health graphs through time series to obtain a dynamic health graph; The dynamic health graph is input into the graph neural network, and the node feature matrix in the dynamic health graph is updated in the graph convolution layer; Through the temporal attention mechanism, the node feature matrix at each time point is linearly transformed and projected into query vector, key vector and value vector; The importance weight is obtained by calculating the product of the query vector and the key vector; Apply the importance weights to the value vector to obtain a time-weighted node feature matrix, and integrate the time-weighted node feature matrices of all time points to obtain a comprehensive node feature matrix; The comprehensive node feature matrix is ​​input into the fully connected layer, and the predicted daily nutrition and total energy intake value, exercise energy consumption value and sleep quality score of the user are output and summarized into ternary features; The predicted ternary features and the standardized ternary data are combined to calculate the user's comprehensive health score; Generate personalized diet recommendations for users based on their comprehensive health scores.

2. The method for pushing dietary information based on big data as claimed in claim 1, characterized in that: The user's diet, exercise and sleep data are obtained through the user's portable smart device, and preliminary fusion is performed to obtain a three-dimensional data snapshot. The specific steps are as follows: Collect users’ dietary data by taking photos of food, describing it verbally, and entering it manually using smart devices; The dietary data of users, which are recorded through taking photos, verbal descriptions and manual input, are stored as structured records to form a single meal record; Through the smart devices carried by users, the user's exercise data is collected and converted into a unified format to record the exercise type, time and energy consumption to form a single-day exercise record; The sleep monitoring device collects the user's sleep data and calculates the user's sleep quality score based on the sleep data; Store the user's sleep data and sleep quality score as a structured record to form a single-day sleep record Based on a unified time series format, the daily meal records, daily exercise records, and daily sleep records are timestamped and aligned to obtain a triple metadata snapshot.

3. The method for pushing dietary information based on big data as claimed in claim 2, characterized in that: The user's triple data snapshot is cleaned and standardized in multiple dimensions to output standardized triple data. The specific steps are as follows: Perform integrity check and missing value completion on the diet data, exercise data, and sleep data in the triple data snapshot respectively; Filter and correct the noise and outliers in the diet data, exercise data and sleep data in the ternary data; The diet data, exercise data and sleep data in the denoised ternary data are standardized respectively and stored in a unified time series format to obtain standardized ternary data.

4. The method for pushing dietary information based on big data as claimed in claim 1, characterized in that: Calculate the weight of the edge between nodes, the expression is as follows: ； in, Health graph node and The weight of the edge between is the number of samples of the characteristic value sequence of the health graph nodes, and Health graph nodes and The corresponding eigenvalue sequence is and Health graph nodes and The mean of the eigenvalue sequence of and Health graph nodes and The standard deviation of the eigenvalue sequence of In the graph convolution layer, the node feature matrix of the dynamic health graph is updated, and the expression is: ； in, For the The node feature matrix of the layer, is the activation function, is the adjacency matrix, For graph neural network The weight matrix of the layer, For the The node feature matrix of the layer; Calculate the importance weight, the expression is as follows: ； in, is the importance weight, For time point The query vector is Indicates time point The key vector of Power, is the number of time points, is the dimension of the key vector, Express Find the square root, It means normalizing the weights into probability distribution.

5. The method for pushing dietary information based on big data according to claim 4, characterized in that: The predicted ternary features and the standardized ternary data are combined to calculate the user's comprehensive health score. The specific steps are as follows: Calculate the user's daily nutritional and total energy recommended intake values ​​based on the user's age, gender, weight and height, and with reference to existing nutritional intake guidelines; By calculating the difference between the predicted daily nutrition and total energy intake values ​​of the user in the ternary features and the recommended daily nutrition and total energy intake values, the health scores of all nutrients and total energy are obtained, and then summarized to obtain the dietary health score; Calculate the user's daily recommended exercise energy consumption value based on the user's age, gender and weight and referring to the existing exercise guidelines; By comparing the matching degree between the user's daily exercise energy consumption value predicted by the ternary features and the user's daily recommended exercise energy consumption value, the exercise health score is obtained; The dietary health score, exercise health score and the user's sleep quality score predicted from the ternary features are aggregated to obtain a comprehensive health score.

6. The method for pushing dietary information based on big data as claimed in claim 5, characterized in that: The specific steps of generating a personalized diet recommendation plan for a user based on the user's comprehensive health score are as follows: Access to the user's health information and health restrictions; Calculate the frequency of occurrence of each type of food in the historical standardized triple data; Obtain the nutrient content and energy supply value of food in the historical standardized ternary data from the food database, and compare them with the user's daily nutrient and total energy recommended intake values ​​to obtain the food nutrition score; According to the existing health guidelines' recommendations on nutrition and energy intake, physical activity, and sleep, health thresholds for dietary health scores, physical activity health scores, and sleep quality scores are set respectively; Based on the comparison results of the dietary health score, exercise health score and sleep quality score with the corresponding health thresholds, dietary recommendations are made according to the frequency of occurrence of each type of food in the historical standardized ternary data and the food nutrition score from high to low; When the user's geographical location changes, the external data interface of the smart device is used to obtain local specialty food recommendations; Dynamically push diet plans based on the user's real-time situation.

7. The method for pushing dietary information based on big data according to claim 6, characterized in that: The specific steps of dynamically pushing a diet plan based on the user's real-time situation are as follows: Based on the user's historical standardized triple data of daily meal times and frequency, a complete diet recommendation plan is pushed before meals; Based on the time and duration of daily exercise in the user's historical standardized ternary data, after exercise, a recommended diet plan for physical recovery is pushed in real time; Based on the daily sleep time in the user's historical standardized ternary data, combined with the user's sleep health score and the corresponding health threshold, determine whether special adjustments are needed; When the sleep quality score is lower than the corresponding health threshold, it is recommended, otherwise it is not recommended.

8. A dietary information push system based on big data, based on the dietary information push method based on big data according to any one of claims 1 to 7, characterized in that: Including data collection module, data standardization module, feature prediction module, health scoring module and diet recommendation module; The data acquisition module is used to obtain the user's diet, exercise and sleep data through the user's portable smart device, and perform preliminary fusion to obtain a three-dimensional data snapshot; The data standardization module is used to perform multi-dimensional cleaning and standardization processing on the user's ternary data snapshot and output standardized ternary data; The feature prediction module is used to predict the user's ternary features using a machine learning algorithm based on the user's standardized ternary data; The health scoring module is used to calculate the user's comprehensive health score by combining the predicted ternary features and the standardized ternary data; The diet recommendation module is used to generate a personalized diet recommendation plan for the user based on the user's comprehensive health score.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the diet information push method based on big data described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for pushing dietary information based on big data described in any one of claims 1 to 7 are implemented.

Images