Dietary Inference System and Method Combining Blood Glucose Monitoring and Mobile Phone Behavior
By combining blood glucose monitoring with mobile phone behavior and using a deep learning model to fuse physiological signals and behavioral data, a non-intrusive, continuous, and highly accurate inference of food intake is achieved. This solves the problems of poor user experience and limited accuracy in traditional methods and is suitable for large-scale health management.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-02-14
- Publication Date
- 2026-06-26
AI Technical Summary
Existing methods for recording food intake have poor user experience, are burdensome to participate in, and have limited inference accuracy, making it impossible to accurately determine eating events and food types.
By combining blood glucose monitoring with mobile phone behavior, and integrating physiological signals and behavioral context information through smartphones and continuous blood glucose monitoring devices, a deep learning model is used to infer dietary intake, achieving a seamless, continuous, and highly accurate inference of food intake.
It improves the accuracy and granularity of dietary inference, reduces the false alarm rate, achieves 24/7 unburdened automatic monitoring, adapts to individual differences among different users, and is low-cost and easy to promote.
Smart Images
Figure CN122290890A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the interdisciplinary fields of smart healthcare, health informatics, and artificial intelligence. More specifically, it relates to a dietary inference system and method that combines blood glucose monitoring with mobile phone behavior. Background Technology
[0002] Accurate food tracking is crucial for diabetes management, weight control, nutrition research, and personal health management. However, traditional food tracking methods (such as manual diaries and photo logs) suffer from problems such as heavy user burden, memory bias, subjective inaccuracies, and difficulty in long-term adherence. Existing automatic or semi-automatic food monitoring methods include image recognition-based technologies, technologies based on dedicated wearable devices, and technologies based on continuous glucose monitoring data. Image recognition-based technologies typically involve users taking photos of food with their mobile phones, using computer vision algorithms to identify food types and estimate portions. Their drawbacks include the need for active user participation, the inability to capture unphotographed eating behaviors (such as snacking), and the influence of shooting angle and lighting on portion estimation accuracy. Technologies based on dedicated wearable devices, such as wrist-worn devices that monitor chewing and swallowing activities using electromyography or acoustic sensors, require users to wear additional specialized equipment, resulting in poor comfort, high costs, and limited adoption by the general population. Continuous glucose monitoring devices can provide high temporal resolution glucose data. By analyzing the glucose fluctuation curve, specific foods or habits that cause glucose fluctuations can be deduced. However, because glucose changes have significant lag and individual differences, it is difficult to determine the starting point of eating, distinguish between meals and snacks, and identify food types based solely on glucose signals. Furthermore, they are easily affected by other factors such as exercise, stress, and medication, leading to ambiguous inference results and a high false alarm rate.
[0003] In summary, current technologies or user experiences are often unsatisfactory, burdensome to participate in, or offer limited information dimensions and inference accuracy. Therefore, there is an urgent need for a new method that can seamlessly, continuously, and with high accuracy in inferring dietary intake. Summary of the Invention
[0004] To address the aforementioned issues, this invention proposes a dietary inference system and method that combines blood glucose monitoring with mobile phone behavior. Utilizing a user's everyday smartphone and a pre-worn medical-grade continuous glucose monitoring device, it achieves unobtrusive monitoring without requiring active recording or additional dedicated sensors. By fusing physiological signals (blood glucose trend) with behavioral contextual information (eating-related actions, environment), it accurately determines the timing of eating events, distinguishes between main meals and snacks, and preliminarily infers the macronutrient types of ingested foods (e.g., primarily carbohydrates, primarily protein / fat) and their relative intake levels, improving the accuracy and granularity of dietary inference. Using behavioral data sensed by the smartphone as leading and corroborating information, it effectively filters out non-eating blood glucose fluctuations caused by factors such as exercise and emotional fluctuations, significantly improving system specificity and reducing false alarm rates. Through a small amount of initial user calibration data (e.g., records of several known meals), it establishes a personalized blood glucose response and behavioral pattern model for each user, enhancing the individual adaptability of the inference.
[0005] To address at least one of the aforementioned technical problems, according to one aspect of the present invention, a diet inference system combining blood glucose monitoring and mobile phone behavior is provided, comprising:
[0006] Data acquisition module;
[0007] Data processing and feature extraction module;
[0008] Multimodal fusion and joint inference module:
[0009] User interaction and calibration module;
[0010] The results output and visualization module presents the inference results to the user or uploads them to the health management platform.
[0011] Furthermore, the data acquisition module includes:
[0012] The continuous glucose monitoring data unit receives minute-by-minute blood glucose values and their derived characteristics from the user's wearable continuous glucose monitoring device via wireless communication.
[0013] The smartphone's behavior sensing unit utilizes the mobile terminal's built-in sensors to collect the following multimodal data:
[0014] Accelerometers / gyroscopes are used to detect typical eating-related gestures and equipment postures.
[0015] A microphone is used to sample audio for short periods of time to analyze the sound characteristics of the environment related to eating.
[0016] An ambient light sensor helps determine whether a user is in a dark environment, and is associated with the act of ordering more food.
[0017] Screen status / usage log.
[0018] Furthermore, the data processing and feature extraction module includes:
[0019] Continuous blood glucose monitoring feature extraction calculates blood glucose features within a sliding window, including: current blood glucose value, rate of blood glucose rise over the past N minutes, area under the blood glucose curve, and deviation from individual baseline blood glucose.
[0020] Behavioral feature extraction;
[0021] Action recognition extracts periodic patterns from accelerometer data to identify "eating gestures";
[0022] Audio event detection;
[0023] Contextual features are used to extract the time, duration, and ambient light intensity of the behavior.
[0024] Furthermore, the multimodal fusion and joint inference module includes: a time alignment and segmentation module, which aligns the continuous blood glucose monitoring data stream and the behavioral data stream based on a unified time axis; when the behavioral features exceed a threshold, a "suspected eating period" is marked;
[0025] Two-level inference model, including:
[0026] Behavioral triggering and initial screening: Based on behavioral perception data, a lightweight classifier is used to initially determine the confidence level of eating behavior during the period, filter out high-probability eating events, and exclude most non-eating activities.
[0027] Physiological verification and precise inference: For each suspected time period after initial screening, extract the corresponding time window and continuous blood glucose monitoring features for a subsequent period; input the "behavioral features" and "continuous blood glucose monitoring features" into a deep learning model for joint analysis;
[0028] Model task:
[0029] Event Confirmation: Determine if it is a genuine eating event;
[0030] Meal categories: distinguish between "main meals" and "snacks";
[0031] Food type inference: Based on blood glucose response patterns and duration of behavior, infer whether the diet is "carbohydrate-based" or "protein / fat-based";
[0032] Relative intake estimation: Based on the duration of the activity, the incremental area under the continuous blood glucose monitoring curve, and the magnitude of the increase, the intake is divided into "low", "medium", and "high" levels.
[0033] Furthermore, the user interaction and calibration module provides a user interface for users to input a small amount of rough information about known meals during the initial use of the system or when necessary; the system uses labeled data to fine-tune the user-irrelevant parts of the joint inference model.
[0034] Furthermore, the user interaction and calibration module presents results in one or more ways, such as structured logs and visual charts.
[0035] According to one aspect of the present invention, a method for dietary inference combining blood glucose monitoring and mobile phone behavior is provided, comprising the following steps: synchronously collecting continuous blood glucose monitoring time series data of the user and multimodal behavioral perception data from the mobile terminal; constructing a joint inference model by aligning the timestamps, extracting features and fusing the two types of data; the joint inference model first uses the behavioral perception data to detect suspected eating behavior segments, and then combines the synchronous continuous blood glucose monitoring data change patterns for verification, classification and quantitative inference, and finally outputs inference results including eating time, meal type, food category and relative intake level.
[0036] According to another aspect of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the dietary inference method combining blood glucose monitoring and mobile phone behavior of the present invention.
[0037] According to another aspect of the present invention, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the steps of the dietary inference method combining blood glucose monitoring and mobile phone behavior of the present invention.
[0038] Compared with existing technologies, the beneficial effects of the above-described method of the present invention are as follows:
[0039] High accuracy and low false alarms: It integrates the physiological response of continuous blood glucose monitoring with direct behavioral observation of smartphones, and utilizes both the "behavioral precursors" and "physiological consequences" of events to make the detection of eating events more reliable and effectively eliminate blood glucose fluctuations caused by exercise or stress alone.
[0040] It provides richer information granularity: it not only detects "whether or not it ate", but also distinguishes "when it ate" (main meal / snack), "approximately what type of food it ate" (carbohydrate level), and provides relative intake levels, providing more valuable information for nutritional analysis than simple event detection.
[0041] Seamless user experience: By making full use of the smartphones that users already carry and the continuous glucose monitoring devices they may be wearing (especially for diabetic patients), it achieves all-day, burden-free automatic monitoring, which greatly improves user compliance.
[0042] The system combines universality and personalization: the core model is universal, while a lightweight calibration mechanism can adapt to the different dietary habits and blood glucose metabolism differences of different users, thus improving the applicability of the method.
[0043] Low cost and easy to promote: No dedicated hardware is required; it only relies on smartphones and the increasingly widespread continuous glucose monitoring devices, which is conducive to health management applications for large populations. Attached Figure Description
[0044] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments will be briefly described below. Obviously, the drawings described below only relate to some embodiments of the present invention and are not intended to limit the present invention.
[0045] Figure 1 This is a system architecture diagram of a preferred embodiment of the present invention;
[0046] Figure 2 This is a timing diagram of the multimodal data fusion and joint inference process according to a preferred embodiment of the present invention;
[0047] Figure 3 This is a structural diagram of a two-level inference model according to a preferred embodiment of the present invention. Detailed Implementation
[0048] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention.
[0049] Unless otherwise defined, the technical or scientific terms used herein shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.
[0050] like Figure 1-3 As shown, this invention provides a method for dietary inference combining blood glucose monitoring and mobile phone behavior, including the following steps: synchronously collecting continuous blood glucose monitoring time series data of users and multimodal behavioral perception data from their smartphones; constructing a joint inference model by aligning the timestamps, extracting features and fusing the two types of data; the model first uses behavioral perception data to detect suspected eating behavior segments, and then combines the synchronous continuous blood glucose monitoring data change patterns for verification, classification and quantitative inference, and finally outputs inference results including eating time, meal type (main meal / snack), food category and relative intake level.
[0051] A dietary inference system combining blood glucose monitoring and mobile phone behavior includes the following modules:
[0052] Data acquisition module:
[0053] Continuous Glucose Monitoring Data Unit: Receives minute-by-minute blood glucose values and their derived characteristics (such as the rate of blood glucose change) from the user's worn continuous glucose monitoring device via wireless communication methods such as Bluetooth.
[0054] Smartphone Behavior Sensing Unit: Utilizes the smartphone's built-in sensors, operating in the background with low power consumption, to collect the following multimodal data:
[0055] Accelerometer / Gyroscope: Detects typical eating-related gestures (such as reciprocating hand-to-mouth movements) and device posture (a phone placed flat on a table may indicate eating).
[0056] Microphone (in user-authorized and privacy-secure mode): Performs short-term audio sampling during specific time periods (such as after a behavior is triggered) to analyze environmental sound characteristics related to eating, such as chewing and swallowing;
[0057] Ambient light sensor: Helps determine whether the user is in a dark environment (such as at night), which may be associated with the act of eating more food;
[0058] Screen status / usage log: A screen that is off and inactive may indicate that the user is focused on eating.
[0059] Data processing and feature extraction module:
[0060] Continuous blood glucose monitoring feature extraction: Calculate blood glucose features within the sliding window, including but not limited to: current blood glucose value, rate of blood glucose rise over the past N minutes, area under the blood glucose curve, and deviation from personal baseline blood glucose.
[0061] Behavioral feature extraction:
[0062] Motion recognition: Extract periodic patterns from accelerometer data to recognize "eating gestures";
[0063] Audio event detection: Identify sound events and their density such as "chewing", "swallowing", and "clattering of utensils" from microphone data;
[0064] Contextual features: Extract the time when the behavior occurs (e.g., whether it is during regular mealtimes), duration, ambient light intensity, etc.
[0065] Multimodal fusion and joint inference module:
[0066] Time alignment and segmentation: Align continuous blood glucose monitoring data streams with behavioral data streams based on a unified time axis. When behavioral features (such as continuously detected eating gestures) exceed a threshold, a "suspected eating period" is marked.
[0067] Two-level inference model:
[0068] Behavior triggering and initial screening: Based on behavior perception data, lightweight classifiers (such as decision trees and support vector machines) are used to initially determine the confidence level of eating behavior during this period. This step mainly filters out high-probability eating events and excludes most non-eating activities.
[0069] Physiological verification and precise inference: For each suspected time period after initial screening, extract the continuous blood glucose monitoring features of the corresponding time window and its subsequent period. The "behavioral features" and "continuous blood glucose monitoring features" are then jointly input into a deep learning model (such as a hybrid model of Temporal Convolutional Network (TCN) or Long Short-Term Memory Network (LSTM) with fully connected layers) for joint analysis.
[0070] Model task:
[0071] a. Event Confirmation: Determine whether it is a real eating event (output yes / no).
[0072] b. Meal Classification: Distinguish between "main meals" and "snacks".
[0073] c. Food type inference: Based on the blood glucose response pattern (such as rapid rise or gradual rise) and the duration of the behavior, infer whether the diet is "mainly carbohydrate" or "mainly protein / fat".
[0074] d. Relative intake estimation: Based on the duration of the activity, the incremental area under the continuous blood glucose monitoring curve, and the magnitude of the increase, the intake is divided into "low", "medium", and "high" levels.
[0075] User interaction and calibration module:
[0076] A user interface is provided, allowing users to input rough information about a small amount of known meals (such as "breakfast, sandwich") during the initial stages of system use or when necessary. The system uses this labeled data to fine-tune the user-irrelevant parts of the joint inference model, achieving personalized adaptation.
[0077] Results Output and Visualization Module
[0078] The inference results (time, meal type, type, level) will be presented to users or uploaded to the health management platform in the form of structured logs, visual charts (such as food icons on a timeline overlaid with blood glucose curves).
[0079] In summary, this invention employs multimodal time-series data fusion technology to solve the problem of asynchronous matching of behavioral signals (short-term, event-driven) and physiological signals (continuous, delayed response) on the time scale.
[0080] Based on a behavior-driven triggering mechanism, low-power behavior perception is used as a "sentinel" to activate more complex analysis and in-depth processing of continuous blood glucose monitoring data only during suspicious periods, thereby optimizing computation and energy consumption.
[0081] Personalized transfer learning employs a "pre-training + fine-tuning" strategy. A general model is first trained on group data, and then fine-tuned using a small amount of user calibration data to fine-tune the last layer or some parameters of the model, quickly adapting to individual user differences.
[0082] Example 1:
[0083] Data Acquisition: The user wears a continuous glucose monitoring sensor and installs the corresponding mobile app of this invention. The app runs in the background in low-power mode, continuously monitoring accelerometer data, and triggers a 10-second audio sampling segment when a specific gesture pattern (such as wrist rotation once every 2 seconds) is detected for more than 15 seconds.
[0084] Feature extraction: For the triggering time period T (e.g., 15:00-15:05), extract behavioral features: gesture cycle frequency (e.g., 0.5Hz), gesture regularity, and chewing event count in the audio segment (e.g., 8 times). Simultaneously extract continuous blood glucose monitoring data: data from 5 minutes before the start of time period T to 60 minutes after its end. Calculate features: average rate of change of blood glucose (ROG) within time period T, maximum blood glucose increment (ΔG) within 30 minutes after the end of T, and blood glucose iAUC.
[0085] Joint inference:
[0086] Initial screening: Behavioral features are input into a pre-trained random forest model, and the "eating confidence" is 85% (>threshold 60%), which passes the initial screening.
[0087] Fine inference: The fused feature vector [gesture frequency = 0.5 Hz, number of chews = 8, average blood glucose change rate = 0.1 mg / dL / min, ΔG = 45 mg / dL, iAUC = 1200 mg·min / dL] is input into the personalized fine-tuned LSTM model.
[0088] Model output: Event confirmation probability 98%; Meal category: "Snack" (because it is in the afternoon, not a regular meal time); Food type inference: "Mainly carbohydrates" (because ROG and ΔG are high); Intake level: "Medium" (based on iAUC and behavior duration).
[0089] Output results: The app automatically adds a record to the health log: "Around 15:00, snack, probably mainly carbohydrates, medium portion." At the same time, a snack icon is marked at 15:00 on the blood sugar curve.
[0090] When a user first uses the system, during the first week, the system prompts the user to simply label several main meals (e.g., selecting "This is lunch" in an app notification). The system records multimodal data corresponding to these labeled periods. After one week, this labeled data is used to fine-tune the last fully connected layer of the joint inference model, enabling the model to learn the user's typical "lunch" behavioral patterns (such as longer midday hand gestures) and unique blood glucose response curves (e.g., the user's post-meal blood glucose peak typically occurs at 45 minutes). After calibration, the system's accuracy in inferring similar pattern events for the user is significantly improved.
[0091] Example 2:
[0092] The computer-readable storage medium of this embodiment stores a computer program that, when executed by a processor, implements the steps in the dietary inference method combining blood glucose monitoring and mobile phone behavior of Embodiment 1.
[0093] The computer-readable storage medium in this embodiment can be an internal storage unit of the terminal, such as the terminal's hard disk or memory; the computer-readable storage medium in this embodiment can also be an external storage device of the terminal, such as a plug-in hard disk, smart memory card, secure digital card, flash memory card, etc. equipped on the terminal; furthermore, the computer-readable storage medium can include both the terminal's internal storage unit and external storage devices.
[0094] The computer-readable storage medium of this embodiment is used to store computer programs and other programs and data required by the terminal. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.
[0095] Example 3:
[0096] The computer device of this embodiment includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the dietary inference method combining blood glucose monitoring and mobile phone behavior of Embodiment 1.
[0097] In this embodiment, the processor can be a central processing unit, or other general-purpose processors, digital signal processors, application-specific integrated circuits, off-the-shelf programmable gate arrays or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. The memory can include read-only memory and random access memory, and provides instructions and data to the processor. A portion of the memory can also include non-volatile random access memory. For example, the memory can also store device type information.
[0098] Those skilled in the art will understand that the content disclosed in the embodiments can be provided as a method, system, or computer program product. Therefore, this solution can take the form of a hardware embodiment, a software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this solution can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage) containing computer-usable program code.
[0099] This solution is described with reference to flowchart illustrations and / or block diagrams of methods and computer program products according to embodiments of this solution. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing device, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0100] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0101] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0102] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.
[0103] The examples described herein are merely preferred embodiments of the invention and are not intended to limit the concept and scope of the invention. Any modifications and improvements made by those skilled in the art to the technical solutions of the invention without departing from the design concept of the invention should fall within the protection scope of the invention.
[0104] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the specific embodiments described above. The specific embodiments and descriptions in the specification are merely for further illustrating the principles of the invention. Various changes and modifications can be made to the present invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the claims and their equivalents.
Claims
1. A dietary inference system combining blood glucose monitoring and mobile phone behavior, characterized in that, include: Data acquisition module; Data processing and feature extraction module; Multimodal fusion and joint inference module: User interaction and calibration module; The results output and visualization module presents the inference results to the user or uploads them to the health management platform.
2. The system as described in claim 1, characterized in that, The data acquisition module includes: The continuous glucose monitoring data unit receives minute-by-minute blood glucose values and their derived characteristics from the user's wearable continuous glucose monitoring device via wireless communication. The smartphone's behavior sensing unit utilizes the mobile terminal's built-in sensors to collect the following multimodal data: Accelerometers / gyroscopes are used to detect typical eating-related gestures and equipment postures. A microphone is used to sample audio for short periods of time to analyze the sound characteristics of the environment related to eating. An ambient light sensor helps determine whether a user is in a dark environment, and is associated with the act of ordering more food. Screen status / usage log.
3. The system as described in claim 2, characterized in that, The data processing and feature extraction module includes: Continuous blood glucose monitoring feature extraction calculates blood glucose features within a sliding window, including: current blood glucose value, rate of blood glucose rise over the past N minutes, area under the blood glucose curve, and deviation from individual baseline blood glucose. Behavioral feature extraction; Action recognition extracts periodic patterns from accelerometer data to identify "eating gestures"; Audio event detection; Contextual features are used to extract the time, duration, and ambient light intensity of the behavior.
4. The system as described in claim 3, characterized in that, The multimodal fusion and joint inference module includes: a time alignment and segmentation module, which aligns the continuous blood glucose monitoring data stream and the behavioral data stream based on a unified time axis; when the behavioral features exceed a threshold, a "suspected eating period" is marked; Two-level inference model, including: Behavioral triggering and initial screening: Based on behavioral perception data, a lightweight classifier is used to initially determine the confidence level of eating behavior during the period, filter out high-probability eating events, and exclude most non-eating activities. Physiological verification and precise inference: For each suspected time period after initial screening, extract the corresponding time window and continuous blood glucose monitoring features for a subsequent period; input the "behavioral features" and "continuous blood glucose monitoring features" into a deep learning model for joint analysis; Model task: Event Confirmation: Determine if it is a genuine eating event; Meal categories: distinguish between "main meals" and "snacks"; Food type inference: Based on blood glucose response patterns and duration of behavior, infer whether the diet is "carbohydrate-based" or "protein / fat-based"; Relative intake estimation: Based on the duration of the activity, the incremental area under the continuous blood glucose monitoring curve, and the magnitude of the increase, the intake is divided into "low", "medium", and "high" levels.
5. The system as described in claim 4, characterized in that, The user interaction and calibration module provides a user interface for users to input rough information about a small amount of known food during the initial use of the system or when necessary; the system uses labeled data to fine-tune the user-irrelevant parts of the joint inference model.
6. The system as described in claim 5, characterized in that, The user interaction and calibration module presents results in one or more ways, such as structured logs and visual charts.
7. A method for inferring diet by combining blood glucose monitoring and mobile phone behavior, characterized in that: The dietary inference system combining blood glucose monitoring and mobile phone behavior, applied to any one of claims 1 to 6, comprises the following steps: The system simultaneously collects continuous blood glucose monitoring time-series data from users and multimodal behavioral perception data from mobile terminals. By aligning the timestamps, extracting features, and fusing the two types of data, a joint inference model is constructed. The joint inference model first uses the behavioral perception data to detect suspected eating behavior segments, and then combines the change patterns of the synchronous continuous blood glucose monitoring data for verification, classification, and quantitative inference. Finally, it outputs inference results including eating time, meal type, food category, and relative intake level.
8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by the processor, it implements the steps in the dietary inference method combining blood glucose monitoring and mobile phone behavior as described in claim 7.
9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the dietary inference method combining blood glucose monitoring and mobile phone behavior as described in claim 7.