Method, device and equipment for evaluating metabolism flexibility based on multi-modal data, medium and product

Abstract

The invention discloses a method, device and equipment for evaluating metabolism flexibility based on multi-modal data, a medium and a product. The method comprises the following steps: collecting first data of a subject, and inputting the first data into a first model to obtain a respiratory entropy value; collecting second data of the subject, fusing the second data and the respiratory entropy value to obtain a first feature vector, and inputting the first feature vector into a second model to obtain a glycolipid metabolism ratio; repeating the above steps to obtain a glycolipid metabolism ratio determined based on the time sequence, and collecting third data of the multi-mode time sequence at the same time; and performing feature extraction on the glucose and lipid metabolism proportion and the third data, fusing the extracted features into a high-dimensional second feature vector, and inputting the second feature vector into the metabolism flexibility scoring model to obtain a metabolism flexibility score of the subject. The health of the subject can also be fed back and intervened based on the metabolic flexibility score.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence, and in particular to a method, device, equipment, medium and product for evaluating metabolic flexibility based on multimodal data. Background Art

[0002] As people's health awareness increases, metabolic health, as a key indicator of an individual's overall health, is gaining increasing attention. Metabolic flexibility, which refers to the body's ability to adjust the proportion of metabolic substrates utilized under varying environmental conditions, is a key indicator of an individual's health and metabolic status. Metabolic flexibility is closely associated with a variety of health issues, including but not limited to obesity, diabetes, cardiovascular disease, and metabolic syndrome.

[0003] Currently, the method for assessing metabolic flexibility mainly relies on indirect calorimetry, which measures the composition of respiratory gases to infer the body's energy source. This is the common respiratory quotient (RQ) test. The respiratory quotient value can reflect the body's dependence on carbohydrates or fats at a certain moment. A lower RQ value usually indicates that fat metabolism is dominant, while a higher RQ value indicates that carbohydrate metabolism is dominant. Although the RQ test can provide certain metabolic status information, existing respiratory quotient detection methods mostly rely on large equipment, are complex to operate, and can usually only provide static, single-point measurement results, lacking the ability to monitor dynamic changes in metabolic status over the long term.

[0004] Existing technologies usually rely on complex laboratory equipment, such as metabolic carts and indirect calorimeters. Although these devices can provide relatively accurate metabolic parameters, they are expensive, bulky, and complex to operate, making them unsuitable for the needs of primary healthcare institutions or family health management. In addition, existing technologies are mostly invasive tests, such as collecting blood samples or other biological samples for metabolic analysis, which not only increases patient discomfort, but also increases the risk of infection and limits the possibility of high-frequency monitoring. Secondly, existing metabolic flexibility assessment methods often focus on a single physiological parameter and lack a comprehensive analysis of multimodal data.

[0005] Finally, existing technologies that leverage metabolic flexibility for health feedback and intervention lack real-time and personalization. Relying on traditional rules or simple statistical models, they are unable to dynamically adjust feedback and intervention strategies based on real-time individual data, resulting in interventions failing to take effect in a timely manner. Existing technologies also have poor dynamic adaptability and rely on fixed algorithms or preset rules, making them unable to respond to dynamic changes in an individual's health status, leading to reduced intervention effectiveness. Summary of the Invention

[0006] In order to solve the technical problems of existing health feedback systems such as lack of real-time and personalization, defects in multi-source data fusion and intelligent processing, poor dynamic adaptability and insufficient hardware adaptability, the present invention provides a method for evaluating metabolic flexibility based on multimodal data. This method can achieve comprehensive data fusion and precise feature extraction, non-invasive and highly accurate metabolic status assessment, and form a self-optimizing health intervention mechanism, thereby improving the practical applicability and user experience of the system in population health management.

[0007] The technical solution adopted by the present invention to solve its technical problem is: providing a method for evaluating metabolic flexibility based on multimodal data, including: collecting first data of the subject, inputting the first data into a first model to obtain the respiratory entropy value of the subject; wherein the first data includes the difference in carbon dioxide concentration between inhalation and exhalation, the resting state of the subject, the basal metabolic rate and the environmental data around the subject; collecting second data of the subject, fusing the second data and the respiratory entropy value to obtain a first eigenvector, and inputting the first eigenvector into a second model to obtain the glucose and lipid metabolism ratio; wherein the second data includes the subject's body mass index, body fat composition, muscle composition, dietary characteristics and exercise characteristics; repeating the above steps based on specific time intervals to obtain the glucose and lipid metabolism ratio determined based on time series, and simultaneously collecting third data, wherein the third data is multimodal time series data; performing feature extraction on the glucose and lipid metabolism ratio and the third data, and fusing the extracted features into a high-dimensional second eigenvector, and inputting the second eigenvector into a metabolic flexibility scoring model to obtain the metabolic flexibility score of the subject.

[0008] Preferably, the present invention further includes: determining different metabolic levels based on the scores, and establishing a mapping table between the metabolic levels and preset recommended behaviors; and using the subject's behavior and the metabolic level or the score as feedback to adjust the recommended behavior.

[0009] Furthermore, the first data includes the difference in carbon dioxide concentration between inhalation and exhalation, the subject's resting state, basal metabolic rate, and environmental data around the subject; and the second data includes the subject's body mass index, body fat composition, muscle composition, dietary characteristics, and exercise characteristics.

[0010] Optionally, the mathematical representation of fusing the second data and the respiratory entropy value to obtain the first eigenvector includes: F = W1·RQ+W2·BMI+W3·C fat +W4·C muscle +W5·L diet +W6·L exercise, where F is the first eigenvector, RQ is the respiratory entropy value, BMI is the body mass index, C fat is the body fat component, C muscle is the muscle component, L diet is the dietary characteristic, L exercise is the motion feature, and W1, W2, W3, W4, W5, and W6 are different weights.

[0011] Preferably, the first model is a random forest regression model, and the second model is any one of a random forest regression model, a multiple linear regression model and a BP neural network.

[0012] Furthermore, when the second model is a multiple linear regression model, it is described by the following formula: Y1 = β0 + β1·RQ + β2·BMI + β3·C fat +β4·C muscle +β5·L diet +β6·L exercise , where Y1 represents the glucose and lipid metabolism ratio, β0 is the bias term, RQ is the respiratory entropy value, BMI is the body mass index, C fat is the body fat component, C muscle is the muscle component, L diet is the dietary characteristic, L exercise is the motion feature, and β0, β1, β2, β3, β4, β5, and β6 are different regression coefficients.

[0013] Furthermore, the third data includes the subject's body mass index, body fat composition, muscle composition, dietary characteristics and exercise characteristics collected based on time series.

[0014] Optionally, the metabolic flexibility scoring model is a multi-layer perceptron model.

[0015] This application also proposes a device for evaluating metabolic flexibility based on multimodal data. The device includes:

[0016] a first acquisition unit, configured to acquire first data of a subject and input the first data into a first model to obtain a respiratory entropy value of the subject; wherein the first data includes a difference in carbon dioxide concentration between inspiration and expiration, a resting state of the subject, a basal metabolic rate, and environmental data surrounding the subject;

[0017] a second acquisition unit, configured to acquire second data of the subject, fuse the second data with the respiratory entropy value to obtain a first eigenvector, and input the first eigenvector into a second model to obtain a glucose and lipid metabolism ratio; wherein the second data includes the subject's body mass index, body fat composition, muscle composition, dietary characteristics, and exercise characteristics;

[0018] a third acquisition unit, configured to repeat the operations of the first acquisition unit and the second acquisition unit at specific time intervals to obtain the glucose and lipid metabolism ratio determined based on a time series, and simultaneously acquire third data; wherein the third data is multimodal time series data;

[0019] A metabolic scoring unit is used to extract features from the glycolipid metabolism ratio and the third data, fuse the extracted features into a high-dimensional second feature vector, and input the second feature vector into a metabolic flexibility scoring model to obtain a metabolic flexibility score for the subject.

[0020] The beneficial effects of the present invention are:

[0021] First, the present invention realizes the accurate prediction of respiratory entropy value from non-invasively collected raw physiological data through a prediction model of respiratory entropy and lipid metabolism ratio based on random forest, overcoming the shortcomings of traditional respiratory entropy detection equipment that is bulky, expensive and complicated to operate, making metabolic status assessment more convenient and real-time.

[0022] Secondly, the present invention adopts a multimodal data fusion lipid metabolism assessment network, which effectively integrates multi-source data such as respiratory entropy, BMI, body composition and lifestyle habits, overcomes the limitations of a single data model, and comprehensively reflects the individual's metabolic characteristics through weighted calculation, providing high-quality input for subsequent evaluation.

[0023] Third, the present invention adopts a staged modeling strategy, so that the model of each stage focuses on learning local complex relationships, reducing error propagation, and thus improving the overall prediction accuracy. The evaluation method based on the multivariate linear regression model has the advantages of simple calculation and high real-time performance. It can quickly evaluate the ratio of glucose and lipid metabolism and is suitable for real-time health monitoring and a wide range of scenarios. At the same time, the evaluation method based on the BP neural network makes full use of the nonlinear mapping capability, can accurately capture the complex relationship between the ratio of glucose and lipid metabolism and multimodal characteristics, and achieve high-precision prediction of metabolic status.

[0024] Fourth, the metabolic flexibility score prediction method of the present invention is based on time series data analysis. It combines long-term recorded multimodal time series data, uses long short-term memory networks (LSTM) to extract features from the data, and predicts the metabolic flexibility score through a multi-layer perceptron (MLP) neural network, thereby realizing a dynamic and comprehensive assessment of individual metabolic flexibility.

[0025] Fifth, the present invention integrates a metabolic flexibility assessment model with an intervention strategy module, creating a self-optimizing "suggestion-feedback-re-suggestion" mechanism. Based on the user's metabolic trends and behavioral results, the system adjusts intervention content for the next cycle in real time, enabling personalized health management. By balancing suggestion stability and responsiveness, this model prevents users from experiencing "intervention fatigue" due to frequent changes in suggestions and improves user compliance.

[0026] In summary, the present invention achieves accurate assessment of individual metabolic flexibility and personalized health intervention through multimodal data fusion, non-invasive detection and intelligent algorithm analysis, overcoming the problems of complex metabolic flexibility assessment equipment, incomplete data analysis, lack of real-time and personalization in the existing technology, and provides a new technical means for the prevention of metabolic-related diseases and health management. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 4 is a flow chart of a method for evaluating metabolic flexibility based on multimodal data according to an embodiment of the present application.

[0028] Figure 2 This is a flowchart of providing health feedback and behavioral intervention to users based on metabolic flexibility scores according to an embodiment of the present application.

[0029] Figure 3 Schematic diagram of a device for evaluating metabolic flexibility based on multimodal data according to an embodiment of the present application.

[0030] Figure 4 is a schematic diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0031] The technical solutions of the present invention will be described clearly and completely below with reference to the accompanying drawings and embodiments. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without inventive effort are also within the scope of protection of the present invention.

[0032] First reference Figure 1 The method for evaluating metabolic flexibility based on multimodal data is described in the flowchart shown in FIG. The method includes:

[0033] Step 101 collects first data from the subject and inputs it into a first model to obtain the subject's respiratory entropy value. The respiratory entropy value reflects the current metabolic substrate utilization: an RQ of ≈ 1.0 indicates that sugar is the primary metabolic substrate; an RQ of ≈ 0.7 indicates that fat is the primary metabolic substrate; intermediate values ​​represent a mixed glucose and lipid metabolism state. Specifically, the first data includes the difference in carbon dioxide concentration between inspiration and expiration (referred to as the respiratory CO2 concentration difference), the subject's resting state, basal metabolic rate, and environmental data surrounding the subject, all of which can be collected non-invasively. The respiratory CO2 concentration difference serves as the primary input variable, while the subject's resting state, basal metabolic rate, and environmental data surrounding the subject serve as auxiliary variables. In one embodiment, a portable monitoring device can first be used to collect the user's physiological data in real time, including body mass index (BMI), body composition (e.g., fat percentage, muscle percentage), and lifestyle habits (e.g., dietary patterns, exercise frequency, and sleep duration). The collected data undergoes denoising, normalization, and missing value imputation to ensure data integrity and consistency. This data serves as model input for subsequent analysis.

[0034] The difference in respiratory CO2 concentration is collected by a portable device under a preset standardized breathing process (inhale-hold your breath-blow, specifically, take a deep breath first - hold your breath for a few seconds - blow slowly). The portable device integrates the volume of carbon dioxide in the exhaled gas collected during the entire blowing process to obtain the total amount of carbon dioxide emitted during this exhalation process. The resting state, for example, may refer to physiological data measured when the subject keeps a quiet rest for at least 30 minutes before measurement, avoiding strenuous exercise or emotional fluctuations. The basal metabolic rate can be determined, for example, by indirect calorimetry, that is, by measuring the subject's oxygen consumption and carbon dioxide production. Finally, all the first data collected above are input into the first model to obtain the respiratory entropy value. The first model is preferably a random forest regression model, which models this nonlinear relationship, can effectively improve the estimation accuracy of the glucose and lipid metabolism ratio, and adapt to metabolic differences between different individuals.

[0035] Step 102 : collecting second data of the subject, fusing the second data with the respiratory entropy value to obtain a first eigenvector, and inputting the first eigenvector into a second model to obtain a glucose and lipid metabolism ratio.

[0036] The second data may include the subject's body mass index, body fat composition, muscle composition, dietary characteristics, and exercise characteristics. The body mass index (BMI) is calculated by dividing the subject's weight by the square of his height. Body fat composition and muscle composition can be measured by a bioelectrical impedance analyzer. Dietary characteristics may include the user's diet composition, nutritional components, and the user's dining habits, for example, the daily intake ratio of carbohydrates, protein, and fat, as well as total calorie intake, etc., and the features of all data are extracted as dietary characteristics. Exercise characteristics can be collected through corresponding sports equipment, recording the type, frequency, intensity, and duration of the subject's exercise during a daily or weekly period, and extracting the features of all data as exercise characteristics.

[0037] The fusion of the first eigenvector can be performed by the following mathematical expression:

[0038] F=W1·RW+W2·BMI+W3·C fat +W4·C muscle +W5·L diet +W6·L exercise

[0039] Among them, F is the first eigenvector, RQ is the respiratory entropy value, BMI is the body mass index, C fat is the body fat composition, C muscle For muscle components, L diet For dietary characteristics, L exercise is the motion feature, and W1, W2, W3, W4, W5, and W6 are different weights. The above formula realizes the weighted fusion of multimodal data to generate a unified feature vector for the prediction of glucose and lipid metabolism ratio.

[0040] The second model receives the first eigenvector as input and outputs the glucose and lipid metabolism ratio, which reflects the relative proportion of carbohydrates and fats used as energy sources in the subject and is a key indicator for evaluating metabolic flexibility. This second model can be implemented using a variety of artificial intelligence models, such as a random forest regression model, a multivariate linear regression model, and a BP neural network model.

[0041] When the second model is a multiple linear regression model, it is described by the following formula:

[0042] Y1=β0+β1·RQ+β2·BMI+β3·C fat +β4·C muscle +β5·L diet +β6·L exercise

[0043] Among them, Y1 represents the ratio of glucose and lipid metabolism, β0 is the bias term, and β0, β1, β2, β3, β4, β5, and β6 are different regression coefficients. These different regression coefficients can be calculated using the least squares method. The specific optimization formula is as follows:

[0044]

[0045] Among them, X is the input variable matrix, Y0 is the target value vector, is the estimated regression coefficient vector.

[0046] When the second model is a BP neural network model, it is described by the following formula:

[0047] First, its input layer is mapped as follows: input data X' = [RQ, BMI, C fat ,C muscle ,L diet ,L exercise ].

[0048] Then, the hidden layer output is calculated as follows:

[0049]

[0050] Among them, H j is the output of the jth node in the hidden layer, is the weight matrix from the input layer to the hidden layer, is the bias term. f is the activation function, usually ReLU(x)=max(0,x).

[0051] Finally, the output layer of the BP neural network model is calculated by the following formula:

[0052]

[0053] Among them, the output layer value Y2 is the ratio of glucose and lipid metabolism, is the weight from the hidden layer to the output layer, b (2) is the output layer bias.

[0054] The evaluation method based on the multivariate linear regression model offers the advantages of computational simplicity and high real-time performance. It can rapidly assess the proportion of glucose and lipid metabolism, making it suitable for real-time health monitoring and a wide range of applications. Furthermore, the evaluation method based on the BP neural network leverages nonlinear mapping capabilities to accurately capture the complex relationship between glucose and lipid metabolism and multimodal features, enabling highly accurate prediction of metabolic status.

[0055] Step 103 repeats the above steps based on specific time intervals to obtain a glucose and lipid metabolism ratio determined based on a time series, and simultaneously collects third data; wherein the third data is multimodal time series data. In order to achieve real-time prediction and evaluation of the subject's metabolic flexibility, the data used to predict metabolic flexibility needs to be time series data. The third data can be obtained by simultaneously acquiring multiple continuous time series values ​​for the above data on a continuous time axis. In one embodiment, predictions can be made once or multiple times daily, and measurements can be taken continuously for 14 days, for example, at intervals of 8 hours per day or once per week, thereby obtaining multiple glucose and lipid metabolism ratio data based on a time series. In addition, simultaneously collecting the third data means that each time the glucose and lipid metabolism ratio is predicted, the third data is collected at the same time point or within an allowable time period, resulting in the third data being multimodal time series data, including the subject's body mass index, body fat composition, muscle composition, dietary characteristics, and exercise characteristics collected based on the time series.

[0056] The above-mentioned assessment of lipid metabolism ratio based on multimodal data fusion effectively integrates multi-source data such as respiratory entropy, BMI, body composition and lifestyle habits, overcomes the limitations of a single data model, and comprehensively reflects the individual's metabolic characteristics through weighted calculation, providing high-quality input data for subsequent metabolic flexibility scoring.

[0057] Step 104 , extracting features from the glucose and lipid metabolism ratio and the third data, fusing the extracted features into a high-dimensional second feature vector, and inputting the second feature vector into a metabolic flexibility scoring model to obtain a metabolic flexibility score of the subject.

[0058] Because the glucose and lipid metabolism ratio and the third data are time series data, a long short-term memory network model (LSTM) is used for feature extraction. LSTM can effectively capture long-term dependencies when processing time series data, especially when the subject's metabolic state changes over time, and can accurately predict future metabolic flexibility trends. Its mathematical description is:

[0059] h t =f(W·h t-1 +U·x t +b)

[0060] Among them, h t is the hidden state at time t, x t is the input at time t (i.e., glucose and lipid metabolism ratio, exercise, sleep, and other data), W and U are weight matrices, b is the bias term, and f is the activation function. This model can generate long-term feature representations for data such as glucose and lipid metabolism ratio. Fusion of these extracted features yields a high-dimensional second feature vector.

[0061] The second eigenvector is then input into the trained metabolic flexibility scoring model to obtain the subject's metabolic flexibility score. The metabolic flexibility scoring model uses a multilayer perceptron (MLP) neural network to learn the complex nonlinear relationship between input features and metabolic flexibility scores through a multi-layer fully connected network. Its mathematical representation is as follows:

[0062] Y3=f(W1·X1+W2·X2+…+W n ·X n +b)

[0063] Among them, Y3 is the metabolic flexibility score, X1, X2, ..., X n Multimodal features of the input, W1, W2, ..., W n are the weights corresponding to different multimodal features, b is the bias term, and f is the activation function. The metabolic flexibility score ranges from 0 to 100, with higher scores indicating better metabolic flexibility and lower scores indicating poorer metabolic flexibility.

[0064] Then, this method can also provide health feedback and behavioral intervention to users based on metabolic flexibility scores through a health advice intervention model, such as Figure 2 Shown include:

[0065] Step 201 : determining different metabolic levels based on the scores, and creating a mapping table between the metabolic levels and preset recommended behaviors.

[0066] The metabolic level may be, for example, excellent, good, medium, poor, or bad, and a mapping table between the metabolic level and the recommended behavior may be as follows:

[0067] Excellent level: Maintain the current lifestyle and appropriately increase the intensity and diversity of exercise;

[0068] Good level: appropriate physical training, optimized diet structure, and increased intake of certain nutrients;

[0069] Moderate level: Adjust eating habits, increase intermittent fasting, eat small and frequent meals, and perform moderate to high intensity exercise several times a week;

[0070] Poor: Reduce refined carbohydrate intake, increase healthy fat intake, and perform aerobic exercise every day;

[0071] Poor level: Strictly control total calorie intake, increase daily activity, and perform metabolic adjustment training under professional guidance.

[0072] In step 202, the subject's behavior and metabolic level or score are used as feedback for health intervention to adjust the recommended behavior. The mapping table is not fixed once it is determined, but will be adjusted according to actual conditions to obtain real-time optimal personalized recommendations.

[0073] For example, a reinforcement learning algorithm can be introduced to adjust recommended behaviors, using the subject's behavior and metabolic level or score as feedback. This allows the user's health management process to be modeled as a "state-behavior-reward" reinforcement learning environment. The state can be a combination of metabolic flexibility level, the user's recent behavioral performance, and physiological parameters (such as weight / heart rate changes); the behavior can be the health intervention behavior recommendations provided; and the reward can be goal achievement (such as weight loss, improvement in RQ, and changes in score level). Specifically, the subject's execution of the recommended behavior can be recorded, including frequency, quality, and subjective experience, while monitoring changes in the metabolic flexibility score. If the level or score improves, the current recommendation is maintained or appropriately strengthened. If the level or score does not improve significantly, the recommendation content is adjusted, perhaps by increasing exercise intensity or adjusting the diet. If the score decreases, the subject's situation is reassessed, and the difficulty of the recommendation may be reduced or the type of recommendation changed. This feedback adjustment mechanism ensures the personalization and effectiveness of the recommended behavior.

[0074] Furthermore, a feedback monitoring and threshold determination mechanism is introduced to dynamically determine whether to intervene: if the metabolic score does not change significantly, the current intervention is maintained; if a preset deviation threshold is reached, an intervention update is triggered, which involves modifying the recommended behavior to intervene in the user's behavior to improve the metabolic score or metabolic level. This creates an "active learning health intervention system" with state-of-care capabilities, rather than a traditional static recommendation push system.

[0075] Finally, to prevent users from experiencing "intervention fatigue" due to frequent changes in recommended behavior, the present invention also incorporates a confidence and adjustment rhythm controller (strategy smoother) into the health recommendation intervention model. This controller can adjust the strength and frequency of recommendations based on factors such as historical recommendation trajectories and user feedback weights. This model thus achieves a balance between stability and responsiveness, improving the system's practical applicability and user experience in population health management.

[0076] Another aspect of the present application provides a device 300 for evaluating metabolic flexibility based on multimodal data. Figure 3 As shown, the device 300 includes:

[0077] The first acquisition unit 301 is used to collect first data of the subject and input the first data into the first model to obtain the respiratory entropy value of the subject; wherein the first data includes the difference in carbon dioxide concentration between inhalation and exhalation, the resting state of the subject, the basal metabolic rate and the environmental data around the subject.

[0078] The second acquisition unit 302 is used to collect second data of the subject, fuse the second data and the respiratory entropy value to obtain a first eigenvector, and input the first eigenvector into the second model to obtain the glucose and lipid metabolism ratio; wherein the second data includes the subject's body mass index, body fat composition, muscle composition, dietary characteristics, and exercise characteristics.

[0079] The third acquisition unit 303 is configured to repeat the operations of the first acquisition unit and the second acquisition unit at specific time intervals to obtain a glucose and lipid metabolism ratio determined based on a time series, and simultaneously acquire third data; wherein the third data is multimodal time series data;

[0080] The metabolic scoring unit 304 is used to extract features from the glucose and lipid metabolism ratio and the third data, fuse the extracted features into a high-dimensional second feature vector, and input the second feature vector into the metabolic flexibility scoring model to obtain a metabolic flexibility score of the subject.

[0081] It is understandable that the device 300 for evaluating metabolic flexibility based on multimodal data can run on various computers and servers, and can use the method for evaluating metabolic flexibility based on multimodal data corresponding to the present application to evaluate the user's metabolic flexibility.

[0082] Now refer to Figure 4 , which is a block diagram of an electronic device 400 according to one embodiment of the present application. The electronic device 400 may include one or more processors 402, a system control logic 408 connected to at least one of the processors 402, a system memory 404 connected to the system control logic 408, a non-volatile memory (NVM) 406 connected to the system control logic 408, and a network interface 410 connected to the system control logic 408.

[0083] The processor 402 may include one or more single-core or multi-core processors. The processor 402 may include any combination of general-purpose processors and special-purpose processors (e.g., graphics processors, application processors, baseband processors, etc.). In the embodiments herein, the processor 402 may be configured to execute one or more of the various embodiments proposed in this application.

[0084] In some embodiments, system control logic 408 may include any suitable interface controller to provide any suitable interface to at least one of processors 402 and / or any suitable device or component in communication with system control logic 408 .

[0085] In some embodiments, the system control logic 408 may include one or more memory controllers to provide an interface to the system memory 404. The system memory 404 may be used to load and store data and / or instructions. In some embodiments, the system memory 404 of the electronic device 400 may include any suitable volatile memory, such as a suitable dynamic random access memory (DRAM).

[0086] The non-volatile memory 406 may include one or more tangible, non-transitory computer-readable media for storing data and / or instructions. In some embodiments, the non-volatile memory 406 may include any suitable non-volatile memory such as flash memory and / or any suitable non-volatile storage device, such as at least one of an HDD (Hard Disk Drive), a CD (Compact Disc) drive, and a DVD (Digital Versatile Disc) drive.

[0087] The non-volatile memory 406 may include a portion of storage resources installed on the device of the electronic device 400, or it may be accessible to the device but not necessarily a part of the device. For example, the non-volatile memory 406 may be accessed over a network via the network interface 410.

[0088] In particular, system memory 404 and non-volatile memory 406 may each include a temporary copy and a permanent copy of instructions 420. Instructions 420 may include instructions that, when executed by at least one of processors 402, cause electronic device 400 to implement the methods provided herein. In some embodiments, instructions 420, hardware, firmware, and / or software components thereof may additionally or alternatively be located in system control logic 408, network interface 410, and / or processor 402.

[0089] In some embodiments, the network interface 410 may be integrated with other components of the electronic device 400. For example, the network interface 410 may be integrated with at least one of the processor 402, the system memory 404, the non-volatile memory 406, and a firmware device (not shown) having instructions, and when at least one of the processors 402 executes the instructions, the electronic device 400 implements one or more of the various embodiments described herein. The network interface 410 may further include any suitable hardware and / or firmware to provide a multiple-input multiple-output radio interface.

[0090] In one embodiment, at least one of the processors 402 may be packaged together with logic for one or more controllers of the system control logic 408 to form a system-in-package (SiP). In one embodiment, at least one of the processors 402 may be integrated on the same die with logic for one or more controllers of the system control logic 408 to form a system-on-chip (SoC).

[0091] The electronic device 400 may further include an input / output (I / O) device 412. The input / output (I / O) device 412 may include a user interface to enable a user to interact with the electronic device 400; and a peripheral component interface may be designed to enable peripheral components to interact with the electronic device 400.

[0092] In some embodiments, the user interface may include, but is not limited to, a display (e.g., an LCD display, a touch screen display, etc.), a speaker, a microphone, one or more cameras (e.g., a still image camera and / or a video camera), a flashlight (e.g., an LED flash), and a keyboard.

[0093] In some embodiments, the peripheral component interface may include, but is not limited to, a non-volatile memory port, an audio jack, and a power interface.

[0094] It should be understood that the structure illustrated in the embodiment of the present invention does not constitute a specific limitation on the electronic device 400. In other embodiments of the present application, the electronic device 400 may include more or fewer components than shown, or may combine or separate certain components, or arrange the components differently. The illustrated components may be implemented in hardware, software, or a combination of software and hardware.

[0095] Program code can be applied to input instructions to perform the functions described herein and generate output information. The output information can be applied to one or more output devices in a known manner. For the purposes of this application, a processing system includes any system having a processor such as, for example, a digital signal processor (DSP), a microcontroller, an application specific integrated circuit (ASIC), or a microprocessor.

[0096] Program code can be implemented with a high-level programming language or an object-oriented programming language to communicate with the processing system. Where necessary, program code can also be implemented with assembly language or machine language. In fact, the mechanism described herein is not limited to the scope of any particular programming language. In either case, the language can be a compiled language or an interpreted language.

[0097] One or more aspects of at least one embodiment may be implemented as representative instructions stored on a computer-readable storage medium, which represent various logic within a processor and, when read by a machine, causes the machine to fabricate logic for performing the techniques described herein. These representations, known as "IP cores," may be stored on a tangible, computer-readable storage medium and supplied to various customers or manufacturing facilities to load into fabrication machines that actually manufacture the logic or processor.

[0098] An embodiment of the present application discloses a computer-readable medium storing one or more programs executable by one or more processors to implement the method of the present application.

[0099] An embodiment of the present application discloses a computer program product, including a computer program, which implements the method of the present application when executed by a processor.

[0100] The above is an explanation of the embodiments of the present application by specific embodiments. Those skilled in the art can easily understand other advantages and effects of the present application from the contents disclosed in this specification. Although the description of the present application will be introduced in conjunction with the preferred embodiment, this does not mean that the features of this invention are limited to this embodiment. In addition, in order to avoid confusion or blurring the focus of the present application, some specific details will be omitted in the description. It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other unless there is a conflict.

[0101] Furthermore, various operations will be described as multiple discrete operations in a manner that is most helpful in understanding the illustrative embodiments; however, the order of description should not be construed as implying that these operations are necessarily order dependent. In particular, these operations do not need to be performed in the order presented.

[0102] Unless the context dictates otherwise, the terms "comprising," "having," and "including" are synonymous. The phrase "A / B" means "A or B." The phrase "A and / or B" means "(A and B) or (A or B)."

[0103] As used herein, the term "module" or "unit" may refer to, be or include: an application specific integrated circuit (ASIC), an electronic circuit, a (shared, dedicated or group) processor and / or memory that executes one or more software or firmware programs, a combinational logic circuit and / or other suitable components that provide the described functionality.

[0104] In the accompanying drawings, some structural or method features are shown in a specific arrangement and / or order. However, it should be understood that such specific arrangement and / or order may not be required. In some embodiments, these features may be arranged in a manner and / or order different from that shown in the illustrative drawings. In addition, the inclusion of structural or method features in a particular figure does not imply that such features are required in all embodiments, and in some embodiments, these features may not be included or may be combined with other features.

[0105] It should be understood that although the terms "first," "second," and the like may be used herein to describe various elements or data, these elements or data should not be limited by these terms. These terms are used only to distinguish one feature from another. For example, a first feature may be referred to as a second feature, and similarly, a second feature may be referred to as a first feature without departing from the scope of the exemplary embodiments.

[0106] It should be noted that in this specification, similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings.

[0107] While the present invention has been shown and described with reference to certain preferred embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention.

Claims

1. A method for evaluating metabolic flexibility based on multimodal data, characterized in that: include: Collecting first data of a subject and inputting the first data into a first model to obtain a respiratory entropy value of the subject; wherein the first data includes a difference in carbon dioxide concentration between inspiration and expiration, the subject's resting state, basal metabolic rate, and environmental data surrounding the subject; collecting second data of the subject, fusing the second data with the respiratory entropy value to obtain a first eigenvector, and inputting the first eigenvector into a second model to obtain a glucose and lipid metabolism ratio; wherein the second data includes the subject's body mass index, body fat composition, muscle composition, dietary characteristics, and exercise characteristics; Repeating the above steps based on specific time intervals to obtain the glucose and lipid metabolism ratio determined based on the time series, and simultaneously collecting third data; wherein the third data is multimodal time series data; Feature extraction is performed on the glycolipid metabolism ratio and the third data, and the extracted features are fused into a high-dimensional second feature vector. The second feature vector is input into a metabolic flexibility scoring model to obtain a metabolic flexibility score of the subject.

2. The method according to claim 1, characterized in that Also includes: determining different metabolic levels based on the scores, and establishing a mapping table between the metabolic levels and preset recommended behaviors; The subject's behavior and the metabolic level or the score are used as feedback to adjust the recommended behavior.

3. The method according to claim 1, characterized in that The step of fusing the second data and the respiratory entropy value to obtain a mathematical representation of a first eigenvector includes: <h2 style=";text-align:left;direction:ltr">F = W1 RQ + W2 BMI + W3 C<h2 style=";text-align:left;direction:ltr"> fat <h2 style=";text-align:left;direction:ltr"> +W4·C<h2 style=";text-align:left;direction:ltr"> muscle <h2 style=";text-align:left;direction:ltr"> +W5·L<h2 style=";text-align:left;direction:ltr"> diet <h2 style=";text-align:left;direction:ltr"> +W6·L<h2 style=";text-align:left;direction:ltr"> exercise Wherein, F is the first eigenvector, RQ is the respiratory entropy value, BMI is the body mass index, C fat is the body fat component, C muscle is the muscle component, L diet is the dietary characteristic, L exercise is the motion feature, and W1, W2, W3, W4, W5, and W6 are different weights.

4. The method according to claim 1, wherein The first model is a random forest regression model, and the second model is any one of a random forest regression model, a multiple linear regression model and a BP neural network.

5. The method according to claim 4, characterized in that When the second model is a multiple linear regression model, it is described by the following formula: Y1=β0+β1·RQ+β2·BMI+β3·C fat +β4·C muscle +β5·L diet +β6·L exercise Wherein, Y1 represents the glucose and lipid metabolism ratio, β0 is the bias term, RQ is the respiratory entropy value, BMI is the body mass index, C fat is the body fat component, C muscle is the muscle component, L diet is the dietary characteristic, L exercise is the motion feature, and β0, β1, β2, β3, β4, β5, and β6 are different regression coefficients.

6. The method according to claim 1, characterized in that The third data includes the subject's body mass index, body fat composition, muscle composition, dietary characteristics, and exercise characteristics collected based on time series.

7. The method according to claim 6, characterized in that The metabolic flexibility scoring model is a multi-layer perceptron model.

8. A device for evaluating metabolic flexibility based on multimodal data, characterized in that: include: a first acquisition unit, configured to acquire first data of a subject and input the first data into a first model to obtain a respiratory entropy value of the subject; wherein the first data includes a difference in carbon dioxide concentration between inspiration and expiration, a resting state of the subject, a basal metabolic rate, and environmental data surrounding the subject; a second acquisition unit, configured to acquire second data of the subject, fuse the second data with the respiratory entropy value to obtain a first eigenvector, and input the first eigenvector into a second model to obtain a glucose and lipid metabolism ratio; wherein the second data includes the subject's body mass index, body fat composition, muscle composition, dietary characteristics, and exercise characteristics; a third acquisition unit, configured to repeat the operations of the first acquisition unit and the second acquisition unit at specific time intervals to obtain the glucose and lipid metabolism ratio determined based on a time series, and simultaneously acquire third data; wherein the third data is multimodal time series data; A metabolic scoring unit is used to extract features from the glycolipid metabolism ratio and the third data, fuse the extracted features into a high-dimensional second feature vector, and input the second feature vector into a metabolic flexibility scoring model to obtain a metabolic flexibility score for the subject.

9. An electronic device, characterized in that: The device comprises a memory storing computer-executable instructions and a processor; when the instructions are executed by the processor, the device implements the method according to any one of claims 1 to 7.

10. A computer-readable medium, characterized in that The computer-readable medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the method according to any one of claims 1 to 7.

11. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images