Nutrition management method and system, terminal and storage medium

Abstract

The invention relates to the field of health management and intelligent terminals, and discloses a nutrition management method and system, a terminal and a storage medium, and the nutrition management method comprises the following steps: an integrated voice recognition module is used for capturing a voice instruction of a user and analyzing a user demand; deploying an augmented reality engine, generating a virtual food model according to a user demand, and displaying a visual effect of the food combination in real time; and establishing a personalized diet database, comprehensively analyzing health data and living habits of the user, and generating a dynamic diet plan. According to the invention, a multi-modal fusion voice recognition and augmented reality scheme is adopted, and a personalized diet database and a sensor network are combined, so that real-time diet management and plan adjustment are realized. The diet health influence is dynamically displayed through an augmented reality simulation tool, flexible intervention is provided through an intelligent reminding system, and the problems that in the prior art, the feedback mode is single, and behavior guidance is not in place are solved.

Description

Technical Field

[0001] The present invention relates to the fields of health management and intelligent terminals, and particularly to a nutrition management method, system, terminal and storage medium. Background Art

[0002] With the increasing demand for health management among people, traditional diet management technologies have become increasingly limited. Existing health management platforms usually rely on text input or simple mobile application interfaces. This single interaction method lacks intuitiveness and interest, and users are prone to losing interest during long-term use. Especially when quick decisions are needed, static charts and texts are difficult to provide efficient support. In addition, most platform diet suggestions are generated based on fixed templates, relying on users to manually input historical data, with low update efficiency. For the real-time monitoring of intake data and physical changes, existing technical means are limited, resulting in suggestions that are difficult to match the dynamic changes of users' health status.

[0003] At the same time, existing platforms usually use static data charts for feedback. Users need to interpret the health significance behind the data by themselves. This one-way information transmission method lacks intuitiveness and depth, and is difficult to meet users' needs especially in the dynamic prediction of long-term health impacts. In addition, most reminder systems rely on fixed times or rules to trigger, with a single content form, lacking the ability to flexibly adjust according to users' actual intake behaviors, and unable to sense users' emotional states. The reminder methods are significantly mechanized and prone to causing users' aversion or even neglect.

[0004] In summary, existing technologies have deficiencies in interactivity, real-time data update, feedback intuitiveness, and pertinence of behavior intervention, and are difficult to effectively promote users to form scientific health management habits. There is an urgent need for a new technical solution to solve the above problems. Summary of the Invention

[0005] In view of the deficiencies of the prior art, the present invention provides a nutrition management method, system, terminal and storage medium, which solves the problems of single interaction method, insufficient real-time performance, unintuitive feedback and inflexible behavior intervention in existing diet management technologies.

[0006] To achieve the above objectives, the present invention is realized through the following technical solutions: A nutrition management method, including the following steps: Integrate a voice recognition module for capturing users' voice commands and parsing users' needs; Deploy an augmented reality engine to generate virtual food models according to users' needs and display the visual effects of food combinations in real time; Establish a personalized diet database, comprehensively analyze users' health data and living habits, and generate a dynamic diet plan; Demonstrate the impact of different food combinations on the user's health through an augmented reality simulation tool; Set up a sensor network to monitor the user's food intake in real time and calibrate the diet plan according to the monitoring results; Enable a reminder system to send reminders to guide the user to adjust their behavior when it detects that the user's behavior deviates from the diet plan.

[0007] Preferably, the nutrition management method further includes: adjusting the diet plan based on the user's geographical location and social activity schedule to adapt to the user's activity scenario and dietary needs.

[0008] Preferably, capture the user's emotional state through an emotion perception module and adjust the interface display and dietary advice content when the user is in a negative emotion.

[0009] A nutrition management system includes: A voice recognition module for capturing the user's voice commands and parsing the user's needs; An augmented reality engine for generating virtual food models and presenting the visual effects of food combinations; A personalized diet database for storing and analyzing the user's health data and living habits and generating a dynamic diet plan; An augmented reality simulation tool for demonstrating the impact of food combinations on health; A sensor network for monitoring the user's food intake and calibrating the diet plan; A reminder module for sending reminders when the user's behavior deviates from the diet plan.

[0010] Preferably, the nutrition management system further includes an emotion perception module for capturing changes in the user's facial expressions and voice tones and dynamically adjusting the user interaction interface and dietary advice based on the emotional state.

[0011] Preferably, the augmented reality engine further includes a multimedia display module for presenting the nutritional value of food, cooking suggestions, and the effects of food combinations.

[0012] A nutrition management terminal includes: A voice input module for capturing the user's voice commands; A data processing module that communicates with the personalized diet database for analyzing the user's health data and generating a diet plan; A display module for presenting virtual food models and diet plans through augmented reality technology; A monitoring module for communicating with the sensor network to collect the user's food intake data in real time; A prompt module for providing dietary adjustment suggestions based on the monitoring results.

[0013] Preferably, the data processing module further includes: a scenario matching algorithm module for analyzing the user's geographical location, social activity schedule, and seasonal food ingredient information, and generating a diet plan adapted to a specific scenario.

[0014] Preferably, a computer-readable storage medium stores a computer program which, when executed by a computer, implements the method as described above.

[0015] Preferably, the program further includes instructions for adjusting diet recommendations and the interaction interface according to the user's emotional state.

[0016] The present invention provides a nutrition management method, system, terminal, and storage medium. It has the following beneficial effects: 1. The present invention adopts a speech recognition and augmented reality solution based on multi-modal fusion technology, achieving the technical effect of realizing real-time diet interaction management through the combination of speech and vision. Compared with the prior art solutions that rely solely on text input or static chart display, it solves the deficiencies of boring user experience, unintuitive interaction, and difficulty in maintaining for a long time.

[0017] 2. The present invention realizes precise monitoring and plan adjustment of the user's real-time food intake by constructing a personalized diet database and combining it with a dynamic sensor network. Compared with the prior art solutions that only rely on manual input by users or pre-designed plans, it solves the problems of untimely data update and lack of pertinence in diet recommendations.

[0018] 3. The augmented reality simulation tool in the present invention is linked with the health prediction model, achieving the technical effect of intuitively and dynamically displaying the impact of diet on health. Compared with the prior art solutions that only provide static data feedback, it solves the deficiencies of users' unintuitive and in-depth understanding of the impact of diet on health, providing more efficient support for users' decision-making.

[0019] 2. The intelligent reminder system configured in the present invention can combine the degree of deviation of the user's behavior, time nodes, and emotional state, achieving a diverse and flexible health intervention reminder effect. Compared with the prior art solutions that only provide a fixed reminder mechanism, it solves the problems of users' boredom with reminder content and inadequate guidance for behavior adjustment. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a schematic flowchart of the steps of the method in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] Next, in combination with the accompanying drawings in the specification of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0022] Embodiment 1: Please refer to the attached Figure 1 , the present invention provides a nutrition management method. This method generates a personalized nutrition plan through the coordinated operation of modules such as a speech recognition module, an augmented reality engine, a personalized diet database, an augmented reality simulation tool, a sensor network, and a reminder system, dynamically monitors the user's intake, and enhances the user's usage stickiness and the persistence of healthy behaviors through multi-sensory interaction.

[0023] Specifically, it includes the following steps: The integrated speech recognition module captures the user's needs In the present invention, the purpose of this step is to capture the user's needs through the speech recognition module, complete the processing and parsing of natural language instructions, so as to provide a clear input data basis for subsequent modules such as the augmented reality engine and the personalized diet database. This module needs to maintain effective information interaction with the augmented reality engine and the personalized diet database, and at the same time ensure that it can fully identify the user's personalized needs, such as dietary preferences, health goals, etc. To achieve the smooth connection of such multiple modules, the present invention designs a speech recognition and natural language processing system based on deep learning algorithms and combines context information to achieve efficient user instruction parsing.

[0024] In some embodiments, the speech recognition module can be embedded in an intelligent terminal (such as a mobile phone, a tablet or a dedicated device), and captures the user's speech in a full-duplex interaction manner. Generally, this module is jointly composed of hardware (such as a microphone, a DSP chip) and software algorithms (such as a speech signal processing and parsing model). After capturing the speech signal through the hardware, the software module is used to preprocess the signal, extract features and perform semantic analysis, so as to lay a foundation for the dynamic parsing of the user's needs.

[0025] In this embodiment, the main functional implementation steps of the speech recognition module are as follows: First, the speech signal is collected by the microphone and converted into a digital audio signal. The audio signal is sampled as a discrete time series x(t), where t represents time, and the sampling frequency is f s . Specifically, the sampled signal is divided into several frames, and each frame of the signal is windowed to reduce the spectral leakage effect. In this embodiment, the window function can be selected as the Hamming window, and its expression is: Among them, w(n) represents the current window function value, which is the weighting coefficient corresponding to the nth sample point; n represents the sequence number of the sample point in the current window function, ranging from 0 to N - 1; N represents the total length of the window function, that is, the number of samples in each frame of the signal.

[0026] Then, perform Fourier transform (FFT) on the processed frame signal to obtain the short-time spectrum signal X(f): Among them, x(f): the frequency-domain signal, which is the complex value corresponding to the frequency f; x(n): the value of the nth sampling point of the time-domain signal; w(n): the window function value, which is the weighting coefficient corresponding to each sample point (such as the Hamming window); e -j2πfn / N : the basis function of the Fourier transform, which includes the real part and the imaginary part and is used to map the signal from the time domain to the frequency domain; j: the imaginary unit, satisfying j 2 = -1; f: the frequency index; N: the total number of sampling points of the signal frame.

[0027] The frequency-domain signal X(f) here is used as the basis for subsequent speech feature extraction.

[0028] In the feature extraction process, the present invention uses Mel Frequency Cepstral Coefficients (MFCC) as the speech feature representation form. Specifically, the short-time spectrum is processed by a Mel filter bank to simulate the auditory perception characteristics of the human ear. The center frequency f of the filter m is calculated by the following formula: Among them, f represents the true frequency (in Hz), which is the frequency-domain frequency value obtained by Fourier transform; f m represents the Mel frequency value, corresponding to the frequency scale perceived by the human ear; 2595 represents a constant coefficient used to convert the linear frequency scale to an approximate logarithmic perception frequency of the human ear; log 10 represents the logarithm operation with base 10; 700 represents a constant used to adjust the mapping relationship between the linear frequency and the Mel frequency.

[0029] After Mel filtering, take the logarithmic power value and generate the final cepstral features through discrete cosine transform (DCT).

[0030] In some possible implementations, to enhance the robustness of speech features, the speech recognition module can also apply technical means such as energy normalization, noise suppression, and speech rate adaptation. For example, energy normalization standardizes the feature vectors to ensure the consistency of the dynamic range of the speech signal, thereby adapting to speech inputs in different recording environments.

[0031] After the speech signal features are extracted, it enters the semantic analysis stage: In this embodiment, semantic analysis is based on natural language processing (NLP) technology and mainly adopts the Transformer architecture based on deep learning. After the input feature sequence z = {z 1 , z 2 ,..., z T} is mapped to a high-dimensional semantic vector space through the embedding layer, the correlation between features is calculated through the multi-head self-attention mechanism. Specifically, the self-attention mechanism completes the weighted calculation through the following formula: Among them, Attention(Q, K, V): the output result of the self-attention mechanism, representing the weighted correlation between different input features; Q: query matrix (Query), representing the degree of attention of the current input feature to other features; K: key matrix (Key), representing the representation of each feature in the feature space; V: value matrix (Value), representing the numerical information of each feature; QK T : matrix multiplication, representing the dot product operation between the query matrix and the key matrix, used to calculate the correlation between features; d k : the dimension of the key matrix, used to scale the correlation score; Normalization coefficient, preventing the dot product result from being too large; softmax: normalization function, used to convert the dot product result into a probability distribution.

[0032] In some embodiments, to further improve the accuracy of semantic analysis, the context data of the user can be modeled. Generally, the system will provide semantic supplements by combining the user's health records (such as diet preferences, allergy information) and real-time status (such as recent diet records). For example, when the user's voice command is "I want to eat low-sugar food", the system will search the database for the user's diet preference history, further clarify the specific definition of "low-sugar", and match recommended options.

[0033] Finally, the parsing of the speech recognition result and the connection with subsequent modules: After semantic analysis of the user instructions output by the speech recognition module, structured data is generated, such as: Dietary goal: low sugar Preference type: vegetarian Time range: today This data is transmitted to the augmented reality engine for generating virtual food models and stored in the personalized diet database for subsequent dynamic adjustment.

[0034] Some extensible designs: In one option, the speech recognition module can further support dialect recognition by training a specific language model to adapt to the speech inputs of users in different regions.

[0035] In another implementation, the system can utilize cloud computing resources to process complex speech recognition tasks and return the parsing results to the terminal in real time, reducing the local computing load.

[0036] Extended embodiments and design ideas As an extended approach, the speech recognition module can also incorporate an emotion perception function. By analyzing the emotional features in the user's speech (such as speech rate and intonation fluctuations), the user's intent can be further enriched. For example, when the user's speech shows signs of fatigue, the system can preferentially recommend simple and healthy meals to reduce the user's decision-making burden.

[0037] During the speech signal acquisition process, if the user's environmental noise is high, the system can suppress background noise through an adaptive noise reduction algorithm to improve the clarity of the speech signal. Common methods include adaptive filters or Wiener filters.

[0038] Through the above steps, the speech recognition module realizes a complete chain from speech signal acquisition to user demand parsing, providing data support for subsequent personalized diet recommendations. This module design combines the advantages of speech signal processing and natural language processing, can adapt to various scenario requirements, and ensures the efficient operation of the overall system function.

[0039] Deploying the augmented reality engine to build virtual food models In this step, based on the user requirements parsed by the speech recognition module, an augmented reality (AR) engine is deployed to generate virtual food models and display the user's customized diet suggestions in real time. The augmented reality engine needs to be able to dynamically construct food models according to the user's semantic requirements and generate a reasonable visual presentation in combination with the information in the personalized diet database to ensure the consistency of the virtual model with the user's health needs. This step forms a data flow connection with the step of integrating the speech recognition module to capture user requirements, and the requirements parsed by the speech recognition will be directly transmitted to the AR engine as the instruction input for subsequent model generation and display.

[0040] The augmented reality engine runs on a smart terminal, such as a smartphone or AR glasses. Generally, the engine obtains user environment data through the terminal camera and superimposes virtual models on the real scene, thereby realizing an immersive visual interaction experience. As an option, the virtual models not only support static display but also enable dynamic interaction through gestures or touch, such as zooming in, rotating, or adjusting food combinations.

[0041] In this embodiment, the specific implementation of the augmented reality engine includes the following: First, the augmented reality engine uses the camera of the user terminal to capture the real-time scene. The scene data is processed into a depth information map and an environmental lighting model to determine the superimposition relationship between the virtual model and the real scene. Specifically, the depth information is obtained through a depth camera or a vision-based SLAM (Simultaneous Localization and Mapping) algorithm, and the formula is as follows: where d(x, y) is the depth value of the pixel point in the scene; B is the baseline distance of the camera; f is the focal length of the camera; Z(x, y) is the parallax value.

[0042] The acquisition of depth information can be used for the spatial positioning of virtual food models. In some embodiments, by constructing a physical rendering layer, the occlusion relationship of the virtual food model is ensured to be consistent with the real environment. For example, when a virtual food is placed on the user's table, the lower part of the model is naturally occluded by the table, enhancing the visual authenticity.

[0043] Specifically, the augmented reality engine generates virtual models by calling the food model data in the personalized diet database. Each food model is stored as a standardized 3D file, such as in OBJ or FBX format. The model contains geometric information, texture maps, and material properties. In a possible implementation, the augmented reality engine will dynamically modify the model according to the user's health needs. For example, if the user specifies a "high-protein and low-fat diet", the engine will preferentially load food models that meet the requirements (such as chicken breast, fish, quinoa, etc.) and display the models in a reasonable proportion combination.

[0044] To enable the virtual model to intuitively reflect the nutritional information of the food, in this embodiment, the display of the model is also bound to the nutritional data. The nutritional data is provided by the personalized diet database and attached to the model surface in the form of labels. For example, each virtual food can display its calorie, protein content, fat content, etc., which are calculated through the following formula: Among them, nutritional value: the proportion of a certain nutrient component (unit: percentage); Component content i : The mass value of the i-th nutrient component in the food (unit: gram); Recommended intake i : The recommended intake of the i-th nutrient component (unit: gram); N: The total number of nutrient component types.

[0045] After calculating with this formula, the nutritional value is presented in percentage form and marked by color coding (for example, green indicates healthy and red indicates exceeding the standard).

[0046] To further enhance the user's interaction experience, the augmented reality engine supports multiple interaction methods. Generally, users can modify the food combination through gestures or voices. For example, when a user selects a salad, they can add chicken breast as an additional protein source through a voice command, and the augmented reality engine will dynamically adjust the combination and update the relevant nutritional information at the same time.

[0047] As an option, the augmented reality engine can also show the potential impact of the food combination on health in the form of animations. For example, when a user selects a high-sugar dessert, the system will display the simulation of the blood sugar curve fluctuation and prompt the possible health risks in an animated way. This dynamic demonstration can help users more intuitively understand the consequences of their choices.

[0048] To improve the rendering effect of augmented reality, the light estimation technology is adopted in this embodiment to make the light and shadow effect of the virtual model match the real scene. The light estimation is based on the direction and intensity of the ambient light, and the light and shadow mapping of the virtual scene is calculated through the light model: I v = I e + I r Among them, I v is the light intensity of the virtual model, I e is the ambient light intensity, and I r is the reflected light intensity. The calculation of the reflected light can adopt the Phong light model, and the formula is: I r = k s ·(R·V) n Among them, k s is the specular reflection coefficient, R is the light reflection direction, V is the viewing direction, and n is the glossiness parameter of the material.

[0049] To adapt to the computing power limitations of mobile devices, the augmented reality engine supports cloud-based 3D rendering. Specifically, the user's terminal sends a food combination request to the cloud server, where high-quality model generation and rendering are completed, and the rendering result is returned to the terminal in the form of a video stream.

[0050] In another possible implementation, the augmented reality engine can also be linked with social platforms. For example, after the user completes a personalized food combination, they can directly generate short videos or pictures with virtual models through the engine and share them on social media. This extended function not only increases the user's interest in using it but also makes the promotion of healthy eating more convenient.

[0051] Through the above steps, the augmented reality engine of the present invention realizes the full-process functions from scene capture, model generation to real-time display. This module not only provides users with intuitive health guidance but also improves the flexibility and applicability of the system through diverse interaction methods.

[0052] Establish a personalized diet database In the present invention, the establishment of the personalized diet database is an important central function that connects the user's health record, real-time data analysis, and subsequent dynamic recommendations. This database not only needs to receive the instruction parsing results from the speech recognition module but also integrate the virtual models and nutritional information generated by the augmented reality engine to provide users with accurate diet plans and subsequent health adjustment suggestions. Through the dynamic update of the personalized diet database, the user's diet plan can be adjusted in real time to be consistent with the user's health needs, and at the same time provide accurate data support for other modules of the system.

[0053] In this embodiment, the core of the personalized diet database is the management and real-time update of multi-level structured data. Generally, this database includes three core parts: user profiles, nutritional data, and historical records. User profiles mainly cover basic health information, nutritional data is responsible for storing the nutritional components of standardized foods, and historical records are the dynamic storage of the user's interaction data and intake records.

[0054] In a possible implementation, the establishment of user profiles is based on the user's initial input and combined with the integration of wearable devices and external health data. The user's basic information can include height, weight, age, gender, etc., and the extended information covers allergy history, dietary taboos, and long-term health goals. For example, after the user enters their weight goal, the database can mark it in real time and provide parameters for subsequent diet plan generation.

[0055] Specifically, the user's health record can be represented by the following formula: Among them, BMI: Body Mass Index, used to measure whether an individual's weight is within a healthy range; Weight (kg): The weight of an individual, measured in kilograms; Height (m): The height of an individual, measured in meters.

[0056] The BMI value will directly affect the calculation of the daily calorie requirement in the diet plan. The subsequent daily energy intake requirement is determined by the following formula: E = 10·W + 6.25·H - 5·A + S Where, E is the total daily energy requirement (kcal); W, H, and A are the weight (kg), height (cm), and age (years) respectively; S is the gender correction coefficient, +5 for males and -161 for females.

[0057] The nutrition data part stores the standardized nutrition information of foods, including calories, proteins, fats, carbohydrates, and micronutrients, etc. Generally, these data are sourced from authoritative food databases, such as the USDA (United States Department of Agriculture) database or domestic nutrition standard databases. In one option, the system supports users to input custom food data. For example, users can input the nutritional components of home-made foods, and the database will automatically calculate the standardized values and store them.

[0058] In order to make effective use of the user's historical records in the system, a dynamic interactive storage structure is designed in this embodiment. Specifically, each user intake record is stored as time series data, defined as: R(t) = {F 1 , F 2 , …, F n} Where, R(t) represents the diet record at time t; F i represents the intake amount of each food (in grams). Historical data can be used to analyze the user's long-term eating habits. For example, to statistics whether the user's protein intake meets the standard or whether the sugar intake exceeds the standard.

[0059] In some embodiments, the dynamic nature of the personalized diet database is achieved through real-time data collection and algorithm optimization. Generally, the real-time data provided by the sensor network is stored in the historical record module of the database. For example, when the smart food scale detects that the user's breakfast contains 100 grams of oats and 200 milliliters of milk, these data will be uploaded immediately and compared with the user's daily intake target.

[0060] In a possible implementation, the dynamic update of the database relies on the support of AI algorithms. For example, the collaborative filtering algorithm is used to generate personalized dietary recommendations for users. The collaborative filtering algorithm constructs a recommendation model based on the user's historical records and the dietary preferences of similar users. The specific algorithm process is as follows: The historical record matrix of user u is defined as R u，i , where i is the food item.

[0061] Similar users are calculated through the cosine similarity formula: where Sim(u, v): the similarity value between user u and user v; R u，i : the rating of user u for item i; R v，i : the rating of user v for item i.

[0062] Based on the user similarity matrix, high-rated but unattempted food items are recommended for the target user u.

[0063] In some embodiments, the database can also dynamically adjust the recommended content through the seasonal ingredient optimization algorithm. For example, in summer, the database will give priority to recommending foods with high water content (such as watermelon and cucumber), while in winter, it will recommend high-calorie foods (such as sweet potato and chestnut).

[0064] In another implementation, the database can also predict the user's health trend in real time. For example, when the historical records show that the user has recently consumed excessive sugar, the system will issue a warning based on the health risk model in the database and recommend alternative foods. For example, recommend sugar-free drinks to replace carbonated drinks.

[0065] To ensure the performance of the database, a distributed architecture is adopted for storage and calculation in this embodiment. Generally, the user's health records are stored in the cloud, and only partial cache data is stored on the device side to improve the response speed. The update of the database is implemented through the incremental update mode, that is, only the changed part is uploaded, thereby reducing the network transmission volume.

[0066] As an extended design, the personalized diet database can be interconnected with a third-party health management platform. For example, share the user's health data with the medical platform to optimize the diet recommendations related to diseases.

[0067] Through the above description, the personalized diet database realizes the full-link function from static data storage to dynamic data update. Combining the user profile, real-time records, and recommendation algorithms, the database not only provides users with a scientific diet plan but also can be adjusted in real time according to the user's health goals, meeting the personalized and dynamic needs of users.

[0068] Enable an Augmented Reality Simulation Tool to Demonstrate the Health Impact of Food In the present invention, the main function of the augmented reality (AR) simulation tool is to present the health impact of dietary choices to users in a dynamic and intuitive manner. This step receives food and nutrition information from a personalized diet database, and at the same time combines the semantic parsing results of user needs and the virtual food models generated by the AR engine to construct a visual display of the health impact. Through this tool, users can intuitively feel the potential impact of different dietary choices on their physical health, such as blood sugar fluctuations, body fat changes, etc., thereby enhancing their understanding of health management.

[0069] In some embodiments, the augmented reality simulation tool is designed as a real-time dynamic demonstration system. Generally, this system calculates the impact of dietary combinations on key health indicators through the combination of a physical health model and a nutritional component model. As an option, this tool can also dynamically generate health curves or simulation animations to show the long-term health trends of users, such as the cumulative effect of a long-term high-sugar diet on the blood sugar curve.

[0070] In this embodiment, the main implementation of the AR simulation tool includes the following: First, the AR simulation tool receives food nutrition data from a personalized diet database and the user's current health profile. The nutritional model calculates the immediate impact of the dietary combination on key health indicators. Taking blood sugar fluctuations as an example, the simulation tool calculates the change in blood sugar after a meal through the following formula: Where: ΔBG is the blood sugar change value (mmol / L); GI is the glycemic index of the food; CHO is the carbohydrate content ingested (g); BW is the user's body weight (kg).

[0071] Specifically, if the user selects a combination of 100 grams of rice and 50 grams of fruit, the tool will calculate the total intake of carbohydrates in each part, simulate its impact on blood sugar through this formula, and generate a corresponding dynamic curve.

[0072] To enhance intuitiveness, in this embodiment, the health impact is also visually presented in a graphical manner. For example, after the user selects a certain high-sugar food, the tool will generate a dynamic fluctuation graph of the blood sugar curve and mark the risk range with a gradient color. Generally, the green area represents the healthy range, and the red area represents blood sugar values outside the healthy range.

[0073] In a possible implementation, the AR simulation tool can also integrate time-based dynamic simulation functions. For example, after the user selects a one-week diet plan, the tool generates a risk map of cumulative weight change or nutritional imbalance within a week through superimposed simulation. This function is predicted by the following formula: Where: ΔW is the weight change value (kg); E in is the total daily energy intake (kcal); E out is the total daily energy expenditure (kcal); 7700 is the unit fat calorie (kcal / kg).

[0074] Specifically, when the user's daily energy intake exceeds their total energy expenditure, the simulation tool dynamically displays the cumulative effect of fat storage and presents it to the user as a visual fat accumulation animation.

[0075] To adapt to different user needs, the augmented reality simulation tool supports various forms of interaction. In some embodiments, the user can adjust the simulation process through gesture operations, such as dragging the timeline to view the health status at different time points or switching different health indicators (such as cholesterol level, blood sugar fluctuations, etc.). Generally, this interaction function is implemented through the touch screen of the terminal device or the gesture capture module of the AR glasses.

[0076] Specifically, this embodiment also provides an extended function that stores the simulation results of different diet choices in a database by dynamically constructing the user's health simulation profile. This function can be used for the user's long-term health management. For example, the user can review the simulation profile to understand the specific impact of their diet decisions on health over a period of time and adjust their future diet plan based on the simulation results.

[0077] To improve the accuracy of the simulation results, in this embodiment, the AR simulation tool introduces machine learning algorithms for data optimization. For example, the tool compares the user's actual intake data with the predicted health indicators and optimizes the accuracy of the simulation results by continuously training the model. The specific training process can be achieved through the following methods: Use actual user data (such as blood glucose monitoring data) as the training set.

[0078] Adopt random forest or support vector machine algorithms to construct a non-linear regression relationship between health indicators and diet combinations.

[0079] Regularly update the model weights to adapt to the individual differences of users.

[0080] In one option, the tool also supports statistical analysis of group data. For example, when there are significant deviations between the user's simulation results and the health trends of the same population, the tool can mark the diet combination as a potential risk and suggest alternative options in the prompt.

[0081] As an extended design, the AR simulation tool can be combined with social functions. For example, users can generate short videos of their simulation results and share them on social platforms. This function can enhance users' health awareness and motivate better diet choices by comparing with others' simulation results.

[0082] Through the above steps, the augmented reality simulation tool of the present invention realizes the full - process functions from data reception, health calculation to dynamic visualization. Combining the personalized diet database with the user's health profile, the tool can not only provide scientific health simulation results, but also improve the user experience through diverse interaction methods. This tool design enhances users' understanding and participation in healthy behaviors and provides strong data support for other modules.

[0083] Set up a sensor network to monitor the user's food intake in real time In this step of the present invention, by deploying a sensor network, the real - time monitoring of the user's food intake is realized, providing accurate basic data support for the dynamic adjustment of the diet plan. This step is closely connected to the personalized diet database and the AR simulation tool, and feeds the collected data back to the core module of the system for calibrating the execution effect of the diet plan and generating health suggestions. The sensor network design fully considers various data sources, including the user's eating behavior, food ingredients, and body physiological parameters, ensuring the comprehensiveness and accuracy of data collection.

[0084] In this embodiment, the core components of the sensor network include a smart food scale, a wearable device, and a smart kitchen device. Generally, these devices are connected to the central server through Bluetooth, Wi - Fi or other communication methods to achieve real - time data transmission and synchronization.

[0085] Specifically, the smart food scale is used to measure the food intake. Before meals, the user weighs the ingredients on the scale, and the initial mass m 0 of each ingredient is recorded. After meals, the scale records the remaining mass m r , and then calculates the intake m c : m c = m 0 - m r In a possible implementation, the system automatically matches the measurement results with the food nutrition information stored in the personalized diet database to generate a report on the user's ingested components. For example, if the user ingests 150 grams of rice, the system calculates the calories, carbohydrates, and trace elements ingested based on the nutritional composition standards of rice and records them in the user's diet profile.

[0086] As an option, the smart fridge can also be part of the sensor network for tracking the inventory status of food and the user's usage. Specifically, the fridge records the amount taken out of each ingredient through built-in cameras and weight sensors and correlates it with the user's intake. For example, if the user takes out 200 grams of milk from the fridge, and the food scale records only 180 grams ingested, the system marks a difference of 20 grams and issues a verification prompt according to the set rules.

[0087] In some embodiments, wearable devices are used to collect the user's physiological data, such as heart rate, steps, calories burned, etc. Generally, these data are combined with the user's food intake to calculate the actual energy balance status. The calculation formula for energy balance is as follows: E b =E in -E out Where: E b is the energy balance value; E in is the total energy intake; E out is the total energy consumption (provided by the wearable device, including basal metabolism and exercise consumption).

[0088] As an implementation, the real-time monitoring results of energy balance are transmitted to the personalized diet database for adjusting subsequent diet plans. For example, when the user's energy consumption is significantly higher than the intake, the system automatically recommends adding foods rich in carbohydrates to maintain physical strength consumption.

[0089] To improve the accuracy of data collection, visual analysis technology is also introduced in this embodiment. Through the camera installed in the kitchen area, the system can identify the user's eating actions and monitor the eating behavior in real time. For example, the camera captures the user's chewing times through motion analysis and calculates the eating speed in combination with the timestamp. This process can be implemented through a convolutional neural network (CNN) model, and the specific training methods include: Using the dining video dataset to label action categories, such as "chewing", "pausing", "drinking water"; Input the feature frame data X = {x 1 ,x 2 ,…,x n} to the network model and output the classification result y; Optimize the model weights to achieve a high action recognition accuracy in real-time scenarios.

[0090] In another possible implementation, the sensor network can also integrate environmental data to analyze the impact of the dining environment on eating behavior. For example, temperature and humidity sensors record environmental conditions, and the system combines the user's eating habits to analyze whether there are abnormal intake situations caused by environmental factors. For example, in a high-temperature environment, users may tend to reduce the intake of high-calorie foods and increase the intake of water or fruits and vegetables.

[0091] To ensure data security and privacy, all sensor data in this embodiment is transmitted through encryption. The data storage adopts a distributed storage architecture, and the user's personal information is anonymized. Generally, users can choose whether to authorize data sharing through the system interface and set privacy preferences at the same time.

[0092] In some extended designs, the sensor network also supports interconnection with other smart home devices. For example, users can query the daily energy intake and consumption through a voice assistant, or automatically generate a shopping list for the next day based on the intake records. As a possible implementation, smart kitchen devices (such as ovens and microwaves) can receive intake suggestions and automatically set cooking parameters according to user selections, such as adjusting cooking time and temperature to meet health requirements.

[0093] Through the above implementation methods, the sensor network of the present invention can not only collect the user's food intake in real-time, but also efficiently interact with other modules to achieve dynamic adjustment and comprehensive management of eating behavior. The diverse design of the sensor network enables it to adapt to different user needs and environmental conditions, providing users with a more intelligent and accurate diet management solution.

[0094] Configure a reminder system to guide users to adjust their behavior In this step of the present invention, the reminder system is designed to give timely feedback on deviations in user behavior and guide users to readjust their diet decisions. This module is closely integrated with the previous sensor network. Based on the real-time collected intake and body status data, it compares the deviation information with the target plan in the personalized diet database to generate specific reminder content. The reminder system not only needs to achieve fast and accurate feedback, but also adjust the reminder method according to the user's preferences and interaction habits to enhance the practical applicability of the system.

[0095] In this embodiment, the reminder system mainly includes a data analysis module, a rule trigger module, and an output module. Generally, the data analysis module receives the real-time data provided by the sensor network and compares it with the diet target; the rule trigger module generates reminder events according to the deviation situation; and the output module selects an appropriate output method according to the reminder type, such as text notification, voice broadcast, or visual warning.

[0096] Specifically, this embodiment realizes the reminder logic in the following way: when the user's real-time intake I(t) deviates from the recommended intake R(t), the reminder system calculates the deviation value D(t): D(t) = I(t) - R(t) Where: D(t) > 0 indicates over-standard, triggering a reminder to reduce intake; D(t) < 0 indicates insufficiency, triggering a reminder to increase intake.

[0097] In a possible implementation, the reminder system can also provide refined reminders according to the nutritional components ingested. For example, when the fat content F(t) ingested by the user exceeds the daily target value F max , the reminder system will trigger a targeted warning and recommend reducing the choice of high-fat foods. For this situation, the system will calculate the fat over-standard ratio in real time: When P F > 1.1 (more than 10% over-standard), a red warning is triggered; when P F is between 1.0 and 1.1, a yellow warning is triggered.

[0098] As an option, the reminder system can provide dynamic adjustment suggestions when the intake is abnormal. For example, when the user's fat intake exceeds the standard at lunch, the system will adjust the recommended foods for dinner and preferentially select a food combination with low fat and high fiber. The new plan generated by the system will recalculate the daily fat intake target based on the following formula: F new = F max - F lunch Where: F new is the adjusted target value; F lunch is the actual intake at lunch.

[0099] In another possible implementation, the reminder system can also monitor the user's intake time interval and give a prompt for abnormal meal times. For example, when the user has not eaten for a long time (no record for more than 4 hours), the system will suggest an appropriate snack. The meal interval T g can be judged by the following conditions: T g = T current - T last If T g > T max (e.g., 4 hours), the system triggers a reminder.

[0100] Generally, the output method of the reminder system is customized according to user preferences. In some embodiments, the reminder content is pushed in the form of a pop-up window on a mobile device, such as "Your sugar intake today has exceeded the standard. Please pay attention to reducing the intake of sugary drinks." In another implementation, the reminder system can provide reminders through voice announcements, which is especially suitable for elderly users or users with visual impairments. The voice reminder content can be directly generated from the deviation data, such as "The fat you just ingested is close to the upper limit. Please choose low-fat foods." To improve the applicability of the reminder, this embodiment also designs a contextual reminder mechanism. For example, when the user is engaged in high-intensity exercise, the system will first recommend an appropriate increase in energy intake instead of prompting a reduction in food choices. This mechanism is achieved by integrating the user's exercise data. The calculation formula for the exercise intensity M is: M = HR · t Where: HR is the average heart rate during exercise; t is the duration of exercise.

[0101] In some embodiments, the reminder system can adjust the tone of the reminder content in combination with the user's emotional state. For example, when the user's emotion analysis module determines that the user is in a stressed state (such as a rapid tone or an impatient facial expression), the system will reduce accusatory language and adopt a softer reminder method, such as "Today's dietary deviation is a bit large, but it doesn't matter. We can adjust it back tomorrow." As an extended design, the reminder system can provide trend-based suggestions by integrating historical data. For example, when the user exceeds the standard multiple times during dinner, the system will mark this behavior as a high-risk point and generate a targeted optimization plan. This plan may include a pre-planned low-calorie meal plan or prompt the user to reduce the purchase of certain high-calorie foods through a reminder tool.

[0102] Through the above content, the reminder system has achieved a complete function from real-time monitoring, deviation calculation to diversified output. Combining the user's personalized needs and health records, this system can guide users to optimize their dietary behaviors in a flexible manner, while providing specific adjustment suggestions to ensure the sustainability of health management.

[0103] Example Two: As part of this application, the present invention also provides a nutrition management system. Based on modular design, the system includes the following components: The voice recognition module is responsible for capturing and parsing user instructions. By the collaborative operation of ASR technology and NLP technology, it realizes the accurate recognition of user intentions.

[0104] Augmented Reality Engine Provides 3D modeling and dynamic rendering functions for generating virtual food models and superimposing them onto the user's actual environment.

[0105] Supports multiple devices (smartphones, tablets, AR glasses).

[0106] Personalized Diet Database Stores user health profiles, dietary preferences, exercise data, and historical intake records.

[0107] The database is dynamically optimized through AI analysis.

[0108] Augmented Reality Simulation Tool Provides a dynamic visualization demonstration function of the health impact of food, enhancing users' awareness of health management.

[0109] Sensor Network Used to collect real-time data on user intake and physical health indicators.

[0110] Reminder Module Provides rule-based health reminders and behavior suggestions.

[0111] As part of this application, the present invention also provides a nutrition management terminal.

[0112] The terminal device can be a smartphone, a tablet, or an embedded device, and includes a processor, a memory, a display screen, a microphone, and a camera.

[0113] Users interact with the terminal via voice or touch. The terminal runs the method of the present invention. The present invention also provides a storage medium storing computer program instructions, which, when the program runs, implement all the functions of the above-mentioned nutrition management method.

[0114] The storage medium includes hard disks, USB flash drives, SD cards, etc.

[0115] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A nutrition management method, characterized in that: The following steps are involved: Integrated speech recognition module to capture user's voice commands and analyze user needs; Deploy an augmented reality engine to generate virtual food models based on user needs and display the visual effects of food combinations in real time; Establish a personalized diet database, comprehensively analyze the user's health data and living habits, and generate a dynamic diet plan; Use augmented reality simulation tools to demonstrate the impact of different food combinations on users' health; Set up a sensor network to monitor the user's food intake in real time and calibrate the diet plan based on the monitoring results; Enable the reminder system to issue reminders to guide users to adjust their behavior when it is detected that the user's behavior deviates from the diet plan.

2. A nutrition management method according to claim 1, characterized in that: Also includes: Based on the user's geographic location and social activity schedule, adjust the meal plan to suit the user's activity scenarios and dietary needs.

3. A nutrition management method according to claim 1, characterized in that: The emotional state of the user is captured through the emotion perception module, and the interface display and dietary suggestion content are adjusted when the user is in a negative mood.

4. A nutrition management system, characterized in that: A method for nutrition management according to any one of claims 1 to 3, comprising: Speech recognition module, used to capture user voice commands and analyze user needs; Augmented reality engine to generate virtual food models and display visual effects of food combinations; Personalized diet database, used to store and analyze user health data and lifestyle habits, and generate dynamic diet plans; an augmented reality simulation tool to demonstrate the health effects of food combinations; a network of sensors to monitor the user’s food intake and calibrate the diet plan; The reminder module is used to send reminders when the user's behavior deviates from the diet plan.

5. A nutrition management system according to claim 4, characterized in that: It also includes an emotion perception module, which is used to capture the user's facial expressions and changes in voice tone, and dynamically adjust the user interaction interface and dietary recommendations based on the emotional state.

6. A nutrition management system according to claim 4, characterized in that: The augmented reality engine further includes a multimedia display module for displaying the nutritional value of food, cooking suggestions and food matching effects.

7. A nutrition management terminal, characterized in that: include: A voice input module, used to capture user's voice commands; a data processing module, communicating with the personalized diet database, for analyzing the user's health data and generating a diet plan; A display module, used for displaying virtual food models and diet plans through augmented reality technology; A monitoring module, for communicating with the sensor network to collect food intake data of the user in real time; The prompt module is used to provide dietary adjustment suggestions based on the monitoring results.

8. A nutrition management terminal according to claim 7, characterized in that: The data processing module further includes: a scenario matching algorithm module for analyzing the user's geographical location, social activity schedule and seasonal food information, and generating a diet plan adapted to a specific scenario.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a computer, the method according to any one of claims 1 to 4 is implemented.

10. A computer-readable storage medium according to claim 9, characterized in that: The program also includes instructions for adjusting dietary recommendations and the interactive interface based on the user's emotional state.

Images