A method for generating dynamic adaptive dietary recommendations based on real-time biomarker monitoring

Abstract

The present invention discloses a method for generating dynamic adaptive dietary advice based on real-time biomarker monitoring, comprising the following steps: S1, a user wears a biomarker monitoring device to monitor the user's biomarkers in real time; S2, the monitored biomarker data is transmitted to a data processing center; S3, a machine learning algorithm is used to process and analyze the collected biomarker data; S4, according to the results of the analysis in S3, in combination with the user's eating habits, health status and lifestyle, a dietary advice is generated; S5, according to the user's response to the dietary advice, the dietary advice is adjusted regularly by continuously monitoring the changes in biomarkers. The present invention can provide personalized dietary advice that immediately responds to the user's current health status by real-time monitoring of the dynamic changes in the user's biomarkers, which not only improves the accuracy of the advice, but also ensures the timeliness and relevance of the advice.

Description

Technical Field

[0001] The present invention relates to the field of intelligent health technology, and in particular to a method for generating dynamic adaptive dietary advice based on real-time biomarker monitoring. Background Art

[0002] In the prior art, personalized health consultation and dietary advice services are usually based on users' static health records and generalized nutritional guidelines. These services often ignore multi-dimensional information such as real-time changes in individual biomarkers, a person's specific lifestyle, psychological conditions, and a person's long-term health plan, resulting in a lack of accuracy and personalization in dietary advice.

[0003] Traditional health consultation services mainly rely on historical medical records and standardized nutritional advice. This approach has the following obvious defects: Traditional methods cannot monitor and analyze changes in user biomarkers, such as blood sugar, cholesterol, and blood pressure, in real time. The real-time data of these biomarkers is crucial for accurately adjusting dietary advice because they directly reflect the user's immediate health status. At the same time, the failure to fully integrate multiple aspects of data such as users' biomarker data, historical health records, lifestyle, and mental health information has led to a one-sided understanding of the user's health status and failed to provide comprehensive personalized health advice. Moreover, traditional dietary advice often adopts a "one-size-fits-all" approach and fails to provide customized services based on the user's personal preferences, dietary restrictions, and living habits. The user's personal preferences and living habits are crucial for developing a sustainable diet plan. Finally, existing health consultation services usually do not include an adaptive feedback loop to adjust subsequent advice based on the user's actual response to previous dietary advice. This static service model limits the real-time and adaptability of dietary advice.

[0004] Therefore, how to provide a method for generating dynamic adaptive dietary recommendations based on real-time biomarker monitoring is an urgent problem that needs to be solved by technicians in this field. Summary of the invention

[0005] One object of the present invention is to propose a method for generating dynamic adaptive dietary recommendations based on real-time biomarker monitoring. The present invention can provide personalized dietary recommendations that immediately respond to the user's current health status by real-time monitoring of the dynamic changes in the user's biomarkers, which not only improves the accuracy of the recommendations, but also ensures the timeliness and relevance of the recommendations.

[0006] A method for generating dynamic adaptive dietary advice based on real-time biomarker monitoring according to an embodiment of the present invention comprises the following steps:

[0007] S1. The user wears a biomarker monitoring device to monitor the user's biomarkers in real time;

[0008] S2, transmitting the monitored biomarker data to the data processing center;

[0009] S3, using machine learning algorithms to process and analyze the collected biomarker data;

[0010] S4, generating dietary recommendations based on the results of the analysis in S3 and in combination with the user's dietary habits, health status and lifestyle;

[0011] S5. Regularly adjust dietary recommendations based on the user's response to dietary recommendations by continuously monitoring biomarker changes.

[0012] Optionally, the S1 specifically includes:

[0013] S11. The biomarker monitoring device is internally equipped with sensors for real-time monitoring of the user's corresponding biomarkers, including blood sugar, cholesterol, blood pressure, intestinal flora composition and metabolite levels, and the collected blood sugar, cholesterol, blood pressure, intestinal flora composition and metabolite levels constitute a set B;

[0014] S12. The biomarker monitoring device continuously collects data of each indicator in the biomarker set at a preset frequency. i The monitoring data of ∈B all consider the corresponding errors, i∈ blood sugar, cholesterol, blood pressure, intestinal flora composition, metabolite level, and form an error model:

[0015] B′ i =B i +∈ i ;

[0016] Among them, B i represents the actual biomarker value, B′ i represents the value of the monitored biomarker, ∈ i Indicates monitoring error;

[0017] S13, the biomarker monitoring device transmits the collected biomarker data to the data processing center in real time via wireless communication;

[0018] S14. The biomarker monitoring device has a built-in function for detecting any biomarker B in set B. i The abnormal data detection algorithm triggers an alarm when the collected biomarker data exceeds the preset normal range.

[0019] Optionally, the abnormal data detection algorithm includes:

[0020] The system analyzes a set of biomarkers B of the user, identifies potential associations and patterns between biomarkers, and performs a comprehensive correlation assessment:

[0021]

[0022] Among them, B′ i and B′ j are the real-time monitoring values ​​of biomarkers, and is the historical average of the biomarker;

[0023] Using a linear regression model, combined with the comprehensive correlation evaluation Corr(B′), the user’s historical data HD and the global health trend GT, the normal range threshold of each biomarker is dynamically adjusted:

[0024]

[0025] Among them, β0, β1, β2 and β3 are model parameters learned by comprehensive correlation evaluation Corr(B′), user’s historical data HD and global health trend GT;

[0026] The system uses a linear regression model to predict the future trend of biomarkers based on the comprehensive correlation evaluation Corr(B′) and the user's historical data:

[0027] B i,future =γ0+γ1×Corr(B′) t-1 +γ2×HD t-1 ;

[0028] Among them, γ0, γ1 and γ2 are model parameters, and the subscript t-1 means using the data at the previous time point for prediction.

[0029] Optionally, the S2 specifically includes:

[0030] S21. Clean the collected biomarker data, remove noise and outliers, and perform normalization;

[0031] S22. Combining correlation assessment and time series forecasting model for trend analysis:

[0032] T i =β0+β1×t+γ1×Corr(B′) t-1 ;

[0033] Where t represents time, β0 and β1 are linear regression parameters, which represent the initial state and rate of change of the trend. The future changes of biomarkers can be predicted by identifying the trend. Corr(B′) t-1 It indicates the correlation between biomarkers at previous time points and is used to adjust trend analysis;

[0034] S23. Periodic pattern recognition based on Fourier transform:

[0035]

[0036] Among them, A k 、f k and φ k They represent the amplitude, frequency, and phase of the periodic pattern, respectively, capturing the repetitive pattern of physiological cycle or seasonal changes in biomarkers, and Δt represents the predicted time offset, which is used to detect and analyze periodic changes in biomarker data.

[0037] Optionally, the S3 specifically includes:

[0038] S31. Use the LSTM model to analyze the preprocessed biomarker time series data B′(t), process and memorize the changes in biomarker data over time, and predict the user's health status and nutritional needs:

[0039] h(B′(t))=LSTM(B′(t);Θ);

[0040] Where B′(t) represents the biomarker data at time t, Θ is the parameter set of the LSTM model;

[0041] S32. Use the historical data set HD to train and optimize the LSTM model. Use the back propagation algorithm to adjust the model parameters to reduce the error between the predicted output and the actual health status. The loss function L is evaluated using the mean square error:

[0042]

[0043] Among them, y is the actual health status label, y i Represents at time t i actual health condition or nutritional needs.

[0044] Optionally, the S4 specifically includes:

[0045] S41. Based on the trends, periodic patterns, mutation points and LSTM model output data of biomarkers, the user's historical health records, nutritional intake, lifestyle and mental health status are integrated for comprehensive data analysis. Through multimodal data integration and correlation analysis, a comprehensive understanding of the user's health status is achieved;

[0046] S42. Use a hybrid algorithm based on rules and machine learning to generate dietary recommendations by combining comprehensive data analysis results and users’ personal preferences;

[0047] S43. Adapt in real time to changes in the user's health status, lifestyle adjustments, and fluctuations in mental health. By continuously monitoring biomarkers and combining real-time feedback from users and changes in living habits, dietary recommendations can be dynamically adjusted.

[0048] Optionally, the multimodal data integration and association analysis includes:

[0049] D integrated =α1·g(B′)+α2·h(HD)+α3·i(L lifestyle )+α4·j(M mental );

[0050] Among them, g, h, i and j are processing functions for different data types to extract the key features of the data, α1, α2, α3 and α4 are weighting parameters used to adjust the impact of each type of data in the integrated analysis, g(B′) represents the processed biomarker data, capturing the physiological indicators of health, h(HD) represents the converted historical health records, reflecting the past health status and disease history, i(L lifestyle ) represents the processed lifestyle information, j(M mental ) means processed mental health data;

[0051] The hybrid algorithm uses the multimodal data integration result D integrated And the results of advanced correlation analysis C analysis and user's personal preference P user Generate dietary recommendations:

[0052] R=λ·D integrated +(1-λ)·C analysis +μ·P user ;

[0053] Among them, C analysis represents the key health associations and patterns discovered from the data, P user It contains the user's personal preferences and dietary restrictions, and λ and μ are weighting parameters to balance the influence of different data sources;

[0054] The diet recommendation algorithm uses the output of the hybrid model R, the user's actual health results, and the consistency between the user's long-term health goals to calculate the loss function:

[0055]

[0056] Where N is the number of samples, t is the sample index, and R(t; Θ hybrid ) is the dietary recommendation at time t according to the hybrid algorithm, H actual (t) is the actual health result of the user at time t, H goalis the user’s long-term health goal, and λ is the regularization parameter used to balance the importance of short-term and long-term health goals.

[0057] Optionally, the lifestyle information i(L lifestyle ) Combined with exercise intensity, exercise type, sleep duration and sleep quality factors, a nonlinear mixed model was used to describe:

[0058]

[0059] in, and Represent different types of exercise and sleep parameters, ω k and ρ m is the weight of each parameter, v1 and v2 are coefficients for adjusting the overall impact, and the hyperbolic tangent function tanh and exponential function are used to capture complex nonlinear relationships;

[0060] The mental health data j(M mental ) Integrate the dynamic and nonlinear characteristics of mental health data to describe the combined effects of stress and emotional state:

[0061]

[0062] in, and Represent the parameters of different stress and emotional states, σ q and τ s is the weight coefficient, ξ1 and ξ2 are used to adjust the overall impact, and the logarithmic and fractional forms are used to capture the complexity and nonlinear response of psychological states.

[0063] Optionally, the S5 specifically includes:

[0064] S51. The system will collect and analyze the user's feedback on previous dietary recommendations. user , including user satisfaction ratings, frequency of following recommendations, and self-reports of any health changes. User feedback will be combined with real-time biomarker monitoring data:

[0065] F integrated =η·F user +(1-η)·D integrated ;

[0066] Among them, η is a parameter that balances the influence of user feedback and biomarker data;

[0067] S52, combined with integrated user feedback and biomarker data, using a hybrid algorithm f hybrid Dietary adjustments:

[0068] R adjusted=R+θ·f hybrid (F integrated ,R previous ; Θ hybrid );

[0069] Where R is the previous diet recommendation, θ is the learning rate, and Θ hybrid are algorithm parameters;

[0070] S53. Use optimized dietary advice R adjusted Minimize the prediction error of the user's health outcomes and consider the user's long-term health goals:

[0071]

[0072] Optionally, it also includes establishing an expert diet library, which includes working with nutritionists, doctors and health experts to collect dietary recommendations for different health conditions, physiological indicators and nutritional needs, classifying and labeling the collected dietary recommendations according to different health conditions and nutritional goals, and creating a data model for matching the user's biomarker data, historical health records and lifestyle with the recommendations in the expert diet library;

[0073] The establishment of the expert diet library includes the use of time series analysis and statistical methods to extract key health characteristics:

[0074]

[0075] Among them, α, β, and γ are weight parameters used to adjust the contribution of different data sources. j and They represent specific values ​​of historical health data and lifestyle respectively;

[0076] Combine the extracted features and LSTM model to evaluate the user's current health status:

[0077]

[0078] Where σ is the activation function, W and b are the weights and biases of the LSTM model, and p is the number of features.

[0079] Combine health status assessment results with user preferences to select and optimize dietary recommendations:

[0080]

[0081] Among them, R(t; Θ select ) are dietary recommendations selected from an expert diet library, H actual (t) and H goal are the user’s actual health status and long-term health goal at time t, N is the number of samples, and λ is the regularization parameter.

[0082] The beneficial effects of the present invention are:

[0083] (1) The present invention can provide personalized dietary recommendations that immediately respond to the user's current health status by monitoring the dynamic changes of the user's biomarkers in real time, which not only improves the accuracy of the recommendations but also ensures the timeliness and relevance of the recommendations.

[0084] (2) The present invention uses multimodal data integration and association analysis to comprehensively consider the user's biomarker data, historical health records, lifestyle and mental health information, revealing the deep-level correlation of the user's health status and providing strong support for formulating comprehensive personalized dietary recommendations. By considering the user's eating habits, health status, lifestyle and personal preferences, the dietary recommendations are personalized, which helps encourage users to follow a healthy diet plan.

[0085] (3) The present invention can dynamically adjust dietary recommendations by analyzing the user's response to previous dietary recommendations and combining it with continuous biomarker monitoring, ensuring that dietary recommendations evolve as the user's health status and lifestyle habits change, thereby improving the long-term effectiveness of the recommendations and user compliance. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0087] Figure 1 This is an overall flow chart of a method for generating dynamic adaptive dietary recommendations based on real-time biomarker monitoring proposed by the present invention;

[0088] Figure 2 This is a flow chart of a personalized diet recommendation system in a dynamic adaptive diet recommendation generation method based on real-time biomarker monitoring proposed in the present invention. DETAILED DESCRIPTION

[0089] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0090] refer to Figure 1-Figure 2 , a method for generating dynamic adaptive dietary recommendations based on real-time biomarker monitoring, comprising the following steps:

[0091] S1. The user wears a biomarker monitoring device to monitor the user's biomarkers in real time;

[0092] S2, transmitting the monitored biomarker data to the data processing center;

[0093] S3, using machine learning algorithms to process and analyze the collected biomarker data;

[0094] S4, generating dietary recommendations based on the results of the analysis in S3 and in combination with the user's dietary habits, health status and lifestyle;

[0095] S5. Regularly adjust dietary recommendations based on the user's response to dietary recommendations by continuously monitoring biomarker changes.

[0096] In this implementation, S1 specifically includes:

[0097] S11. The biomarker monitoring device is internally equipped with sensors for real-time monitoring of the user's corresponding biomarkers, including blood sugar, cholesterol, blood pressure, intestinal flora composition and metabolite levels, and the collected blood sugar, cholesterol, blood pressure, intestinal flora composition and metabolite levels constitute a set B;

[0098] S12. The biomarker monitoring device continuously collects data of each indicator in the biomarker set at a preset frequency. i The monitoring data of ∈B all consider the corresponding errors, i∈ blood sugar, cholesterol, blood pressure, intestinal flora composition, metabolite level, and form an error model:

[0099] B′ i =B i +∈ i ;

[0100] Among them, B i represents the actual biomarker value, B′ i represents the value of the monitored biomarker, ∈ i Indicates monitoring error;

[0101] S13, the biomarker monitoring device transmits the collected biomarker data to the data processing center in real time via wireless communication;

[0102] S14. The biomarker monitoring device has a built-in function for detecting any biomarker B in set B. i The abnormal data detection algorithm triggers an alarm when the collected biomarker data exceeds the preset normal range.

[0103] In this implementation, the abnormal data detection algorithm includes:

[0104] The system analyzes a set of biomarkers B of the user, identifies potential associations and patterns between biomarkers, and performs a comprehensive correlation assessment:

[0105]

[0106] Among them, B′ i and B′ j are the real-time monitoring values ​​of biomarkers, and is the historical average of the biomarker;

[0107] Using a linear regression model, combined with the comprehensive correlation evaluation Corr(B′), the user’s historical data HD and the global health trend GT, the normal range threshold of each biomarker is dynamically adjusted:

[0108]

[0109] Among them, β0, β1, β2 and β3 are model parameters learned by comprehensive correlation evaluation Corr(B′), user’s historical data HD and global health trend GT;

[0110] The system uses a linear regression model to predict the future trend of biomarkers based on the comprehensive correlation evaluation Corr(B′) and the user's historical data:

[0111] B i,future =γ0+γ1×Corr(B′) t-1 +γ2×HD t-1 ;

[0112] Among them, γ0, γ1 and γ2 are model parameters, and the subscript t-1 means using the data at the previous time point for prediction.

[0113] In this implementation, S2 specifically includes:

[0114] S21. Clean the collected biomarker data, remove noise and outliers, and perform normalization;

[0115] S22. Combining correlation assessment and time series forecasting model for trend analysis:

[0116] T i =β0+β1×t+γ1×Corr(B′) t-1 ;

[0117] Where t represents time, β0 and β1 are linear regression parameters, which represent the initial state and rate of change of the trend. The future changes of biomarkers can be predicted by identifying the trend. Corr(B′) t-1 It indicates the correlation between biomarkers at previous time points and is used to adjust trend analysis;

[0118] S23. Periodic pattern recognition based on Fourier transform:

[0119]

[0120] Among them, A k 、f k and φ k They represent the amplitude, frequency, and phase of the periodic pattern, respectively, capturing the repetitive pattern of physiological cycle or seasonal changes in biomarkers, and Δt represents the predicted time offset, which is used to detect and analyze periodic changes in biomarker data.

[0121] In this implementation, S3 specifically includes:

[0122] S31. Use the LSTM model to analyze the preprocessed biomarker time series data B′(t), process and memorize the changes in biomarker data over time, and predict the user's health status and nutritional needs:

[0123] h(B′(t))=LSTM(B′(t);Θ);

[0124] Where B′(t) represents the biomarker data at time t, Θ is the parameter set of the LSTM model;

[0125] S32. Use the historical data set HD to train and optimize the LSTM model. Use the back propagation algorithm to adjust the model parameters to reduce the error between the predicted output and the actual health status. The loss function L is evaluated using the mean square error:

[0126]

[0127] Among them, y is the actual health status label, y i Represents at time t i actual health condition or nutritional needs.

[0128] In this implementation, S4 specifically includes:

[0129] S41. Based on the trends, periodic patterns, mutation points and LSTM model output data of biomarkers, the user's historical health records, nutritional intake, lifestyle and mental health status are integrated for comprehensive data analysis. Through multimodal data integration and correlation analysis, a comprehensive understanding of the user's health status is achieved;

[0130] S42. Use a hybrid algorithm based on rules and machine learning to generate dietary recommendations by combining comprehensive data analysis results and users’ personal preferences;

[0131] S43. Adapt in real time to changes in the user's health status, lifestyle adjustments, and fluctuations in mental health. By continuously monitoring biomarkers and combining real-time feedback from users and changes in living habits, dietary recommendations can be dynamically adjusted.

[0132] In this embodiment, multimodal data integration and association analysis include:

[0133] D integrated =α1·g(B′)+α2·h(HD)+α3·i(L lifestyle )+α4·j(M mental );

[0134] Among them, g, h, i and j are processing functions for different data types to extract the key features of the data, α1, α2, α3 and α4 are weighting parameters used to adjust the impact of each type of data in the integrated analysis, g(B′) represents the processed biomarker data, capturing the physiological indicators of health, h(HD) represents the converted historical health records, reflecting the past health status and disease history, i(L lifestyle ) represents the processed lifestyle information, j(M mental ) means processed mental health data;

[0135] The hybrid algorithm uses the multimodal data integration results D integrated And the results of advanced correlation analysis C analysis and the user's personal preference P user Generate dietary recommendations:

[0136] R=λ·D integrated +(1-λ)·C analysis +μ·P user ;

[0137] Among them, C analysis represents the key health associations and patterns discovered from the data, P user It contains the user's personal preferences and dietary restrictions, and λ and μ are weighting parameters to balance the influence of different data sources;

[0138] The diet recommendation algorithm uses the output of the hybrid model R, the user's actual health results, and the consistency between the user's long-term health goals to calculate the loss function:

[0139]

[0140] Where N is the number of samples, t is the sample index, and R(t; Θ hybrid ) is the dietary recommendation at time t according to the hybrid algorithm, H actual (t) is the actual health result of the user at time t, H goalis the user’s long-term health goal, and λ is the regularization parameter used to balance the importance of short-term and long-term health goals.

[0141] In this embodiment, the lifestyle information i(L lifestyle ) Combined with exercise intensity, exercise type, sleep duration and sleep quality factors, a nonlinear mixed model was used to describe:

[0142]

[0143] in, and Represent different types of exercise and sleep parameters, ω k and ρ m is the weight of each parameter, v1 and b2 are the coefficients for adjusting the overall impact, and the hyperbolic tangent function tanh and exponential function are used to capture complex nonlinear relationships;

[0144] Mental health dataj(M mental ) Integrate the dynamic and nonlinear characteristics of mental health data to describe the combined effects of stress and emotional state:

[0145]

[0146] in, and Represent the parameters of different stress and emotional states, σ q and τ s is the weight coefficient, ξ1 and ξ2 are used to adjust the overall impact, and the logarithmic and fractional forms are used to capture the complexity and nonlinear response of psychological states.

[0147] In this implementation, S5 specifically includes:

[0148] S51. The system will collect and analyze the user's feedback on previous dietary recommendations. user , including user satisfaction ratings, frequency of following recommendations, and self-reports of any health changes. User feedback will be combined with real-time biomarker monitoring data:

[0149] F integrated =η·F user +(1-η)·D integrated ;

[0150] Among them, η is a parameter that balances the influence of user feedback and biomarker data;

[0151] S52, combined with integrated user feedback and biomarker data, using a hybrid algorithm f hybrid Dietary adjustments:

[0152] R adjusted=R+θ·f hybrid (F integrated ,R previous ; Θ hybrid );

[0153] Where R is the previous diet recommendation, θ is the learning rate, and Θ hybrid are algorithm parameters;

[0154] S53. Use optimized dietary advice R adjusted Minimize the prediction error of the user's health outcomes and consider the user's long-term health goals:

[0155]

[0156] In this embodiment, it also includes establishing an expert diet library, which includes working with nutritionists, doctors and health experts to collect dietary recommendations for different health conditions, physiological indicators and nutritional needs, classifying and labeling the collected dietary recommendations according to different health conditions and nutritional goals, and creating a data model for matching the user's biomarker data, historical health records and lifestyle with the recommendations in the expert diet library;

[0157] The establishment of the expert diet library includes the use of time series analysis and statistical methods to extract key health characteristics:

[0158]

[0159] Among them, α, β, and γ are weight parameters used to adjust the contribution of different data sources. j and They represent specific values ​​of historical health data and lifestyle respectively;

[0160] Combine the extracted features and LSTM model to evaluate the user's current health status:

[0161]

[0162] Where σ is the activation function, W and b are the weights and biases of the LSTM model, and p is the number of features.

[0163] Combine health status assessment results with user preferences to select and optimize dietary recommendations:

[0164]

[0165] Among them, R(t; Θ select ) are dietary recommendations selected from an expert diet library, H actual (t) and H goal are the user’s actual health status and long-term health goal at time t, N is the number of samples, and λ is the regularization parameter.

[0166] Embodiment 1:

[0167] In Example 1, it is shown how a 35-year-old office worker Xiao Wang can improve his health through the personalized diet suggestion system of the present invention. In a recent physical examination, Xiao Wang's initial physical examination showed that his fasting blood sugar level was 7.9mmol / L, his total cholesterol level was 6.5mmol / L, and his blood pressure was 133 / 85mmHg, reflecting mild hypertension. His intestinal flora composition and metabolite levels showed no abnormalities. In his daily work, Xiao Wang often worked overtime, had irregular diet, and lacked sufficient physical exercise. Xiao Wang hopes to improve his health by adjusting his eating habits, but he does not know where to start.

[0168] After learning about the experience of accepting the personalized diet suggestion system of the present invention, Xiao Wang is first provided with a wearable device for monitoring his biomarkers around the clock. The system first analyzes his health status based on Xiao Wang's initial biomarker data using a machine learning algorithm, and generates the first version of personalized diet suggestions based on his living habits and dietary preferences. The suggestions include daily calorie intake, recommended food types, and dietary frequency, as shown in Table 1 below:

[0169] Table 1 Xiao Wang's four-week personalized dietary advice health improvement record

[0170]

[0171] As can be seen from Table 1 above, in the first week: the system recommended that Xiao Wang consume 2000 kcal of calories per day, mainly from whole grains, lean meat, vegetables and low-fat dairy products. It was recommended that he eat at least 500 grams of vegetables and fruits every day and limit the intake of high-sugar and high-fat foods. After one week, Xiao Wang's blood sugar dropped from 7.9mmol / L to 6.8mmol / L, cholesterol dropped from 6.5mmol / L to 5.9mmol / L, and blood pressure remained at 132 / 83mmHg.

[0172] Week 2: The system recommended that Xiao Wang consume 1950 kcal of calories per day. In view of the decrease in blood sugar and cholesterol, the system continued to maintain a low-sugar and low-fat diet, and added foods rich in omega-3 such as deep-sea fish to further lower cholesterol, and increased the proportion of coarse grains to control blood sugar. At the end of the second week, Xiao Wang's blood sugar dropped to 6.2mmol / L, cholesterol dropped to 5.5mmol / L, and blood pressure dropped slightly to 128 / 83mmHg.

[0173] Week 3: The system recommended that Xiao Wang consume 1900 kcal of calories per day. According to Xiao Wang's feedback, he said he wanted to try more diverse foods, so beans and nuts were added to the recommendation while keeping the total calories unchanged. The system also added regular physical activity recommendations, such as brisk walking three times a week. By the weekend, Xiao Wang's blood sugar further dropped to 5.9mmol / L, cholesterol reached 5.2mmol / L, and blood pressure stabilized at 127 / 80mmHg.

[0174] Week 4: The system recommends that Xiao Wang consume 1,900 kcal of calories per day. Xiao Wang reports that his energy level has improved and his work efficiency has also improved. The system optimizes the suggestions based on this and introduces intermittent fasting to encourage Xiao Wang's metabolic level to improve. At the same time, in order to better adapt to Xiao Wang's new lifestyle, the system recommends that he try food diversification, including more types of vegetables, fruits and plant protein sources. At the end of the fourth week, Xiao Wang's blood sugar stabilized at a normal level of 5.6mmol / L, cholesterol dropped to an ideal level of 4.9mmol / L, and blood pressure also dropped to an ideal level of 125 / 78mmHg.

[0175] Over a one-month period, the system continuously monitored changes in Xiao Wang's biomarkers and adjusted dietary recommendations based on real-time data and Xiao Wang's feedback. In this way, Xiao Wang's blood sugar level remained within the normal range, his cholesterol level was effectively controlled, his metabolic level improved, and his intestinal flora composition and blood pressure remained at a healthy level without adverse changes. In addition, Xiao Wang reported that he felt more energetic and his sleep quality improved.

[0176] The personalized dietary advice system in Example 1 demonstrates its effectiveness in real-time monitoring of biomarkers, generation of personalized advice, and adaptive adjustment. Xiao Wang's case demonstrates the effectiveness of the system in improving the health of users, adapting to personal eating habits and lifestyles, and promoting long-term health management. By monitoring the dynamic changes of user biomarkers in real time, personalized dietary advice that immediately responds to the user's current health status can be provided, which not only improves the accuracy of the advice, but also ensures the timeliness and relevance of the advice.

[0177] The present invention utilizes multimodal data integration and association analysis to comprehensively consider the user's biomarker data, historical health records, lifestyle, and mental health information, revealing the deep-level correlation of the user's health status, and providing strong support for formulating comprehensive personalized dietary recommendations. By considering the user's eating habits, health status, lifestyle, and personal preferences, a personalized level is achieved when providing dietary recommendations, which helps encourage users to follow a healthy diet plan.

[0178] By analyzing the user's response to previous dietary advice and combining it with continuous biomarker monitoring, the present invention can dynamically adjust dietary advice, ensuring that dietary advice evolves as the user's health status and lifestyle habits change, thereby improving the long-term effectiveness of the advice and user compliance.

[0179] In summary, our method has achieved remarkable results in improving the accuracy of dietary advice, user satisfaction, and long-term health management through high data integration capabilities, personalized services, and adaptive feedback mechanisms. These beneficial effects make our method have important innovative value and practicality in the field of personalized health consultation and dietary advice.

[0180] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for generating dynamic adaptive dietary recommendations based on real-time biomarker monitoring, characterized in that: The steps include: S1. The user wears a biomarker monitoring device to monitor the user's biomarkers in real time; S2, transmitting the monitored biomarker data to the data processing center; S3, using machine learning algorithms to process and analyze the collected biomarker data; S4, generating dietary recommendations based on the results of the analysis in S3 and in combination with the user's dietary habits, health status and lifestyle; S5. Regularly adjust dietary recommendations based on the user's response to dietary recommendations by continuously monitoring biomarker changes; The S1 specifically includes: S11. The biomarker monitoring device is internally equipped with sensors for real-time monitoring of the user's corresponding biomarkers, including blood sugar, cholesterol, blood pressure, intestinal flora composition and metabolite levels, and the collected blood sugar, cholesterol, blood pressure, intestinal flora composition and metabolite levels constitute a set B; S12. The biomarker monitoring device continuously collects data of each indicator in the biomarker set at a preset frequency. i The monitoring data of ∈B all consider the corresponding errors, i∈ blood sugar, cholesterol, blood pressure, intestinal flora composition, metabolite level, and form an error model: B′ i =B i +∈ i ; Among them, B i represents the actual biomarker value, B′ i represents the value of the monitored biomarker, ∈ i Indicates monitoring error; S13, the biomarker monitoring device transmits the collected biomarker data to the data processing center in real time via wireless communication; S14. The biomarker monitoring device has a built-in function for detecting any biomarker B in set B. i Abnormal data detection algorithm that triggers an alarm when the collected biomarker data exceeds the preset normal range; The abnormal data detection algorithm includes: The system analyzes a set of biomarkers B of the user, identifies potential associations and patterns between biomarkers, and performs a comprehensive correlation assessment: Among them, B′ i and B′ j are the real-time monitoring values ​​of biomarkers, and is the historical average of the biomarker; Using a linear regression model, combined with the comprehensive correlation evaluation Corr(B′), the user’s historical data HD and the global health trend GT, the normal range threshold of each biomarker is dynamically adjusted: Among them, β0, β1, β2 and β3 are model parameters learned by comprehensive correlation evaluation Corr(B′), user’s historical data HD and global health trend GT; The system uses a linear regression model to predict the future trend of biomarkers based on the comprehensive correlation evaluation Corr(B′) and the user's historical data: B i,future =γ0+γ1×Corr(B′) t-1 +γ2×HD t-1 ; Among them, γ0, γ1 and γ2 are model parameters, and the subscript t-1 means using the data at the previous time point for prediction.

2. A method for generating dynamic adaptive dietary advice based on real-time biomarker monitoring according to claim 1, characterized in that: The S2 specifically includes: S21. Clean the collected biomarker data, remove noise and outliers, and perform normalization; S22. Combining correlation assessment and time series forecasting model for trend analysis: T i =β0+β1×t+γ1×Corr(B′) t-1 ; Where t represents time, β0 and β1 are linear regression parameters, which represent the initial state and rate of change of the trend. The future changes of biomarkers can be predicted by identifying the trend. Corr(B′) t-1 It indicates the correlation between biomarkers at previous time points and is used to adjust trend analysis; S23. Periodic pattern recognition based on Fourier transform: Among them, A k 、f k and φ k They represent the amplitude, frequency, and phase of the periodic pattern, respectively, capturing the repetitive pattern of physiological cycle or seasonal changes in biomarkers, and Δt represents the predicted time offset, which is used to detect and analyze periodic changes in biomarker data.

3. The method for generating dynamic adaptive dietary advice based on real-time biomarker monitoring according to claim 1, characterized in that: The S3 specifically includes: S31. Use the LSTM model to analyze the preprocessed biomarker time series data B′(t), process and memorize the changes in biomarker data over time, and predict the user's health status and nutritional needs: h(B′(t))=LSTM(B′(t);Θ); Where B′(t) represents the biomarker data at time t, Θ is the parameter set of the LSTM model; S32. Use the historical data set HD to train and optimize the LSTM model. Use the back propagation algorithm to adjust the model parameters to reduce the error between the predicted output and the actual health status. The loss function L is evaluated using the mean square error: Among them, y is the actual health status label, y i Represents at time t i actual health condition or nutritional needs.

4. A method for generating dynamic adaptive dietary advice based on real-time biomarker monitoring according to claim 3, characterized in that: The S4 specifically includes: S41. Based on the trends, periodic patterns, mutation points and LSTM model output data of biomarkers, the user's historical health records, nutritional intake, lifestyle and mental health status are integrated for comprehensive data analysis. Through multimodal data integration and correlation analysis, a comprehensive understanding of the user's health status is achieved; S42. Use a hybrid algorithm based on rules and machine learning to generate dietary recommendations by combining comprehensive data analysis results and users’ personal preferences; S43. Adapt in real time to changes in the user's health status, lifestyle adjustments, and fluctuations in mental health. By continuously monitoring biomarkers and combining real-time feedback from users and changes in living habits, dietary recommendations can be dynamically adjusted.

5. A method for generating dynamic adaptive dietary advice based on real-time biomarker monitoring according to claim 4, characterized in that: The multimodal data integration and association analysis includes: D integrated =α1·g(B′)+α2·h(HD)+α3·i(L lifestyle )+α4·j(M mental ); Among them, g, h, i and j are processing functions for different data types to extract the key features of the data, α1, α2, α3 and α4 are weighting parameters used to adjust the impact of each type of data in the integrated analysis, g(B′) represents the processed biomarker data, capturing the physiological indicators of health, h(HD) represents the converted historical health records, reflecting the past health status and disease history, i(L lifestyle ) represents the processed lifestyle information, j(M mental ) means processed mental health data; The hybrid algorithm uses the multimodal data integration result D integrated And the results of advanced correlation analysis C analysis and the user's personal preference P user Generate dietary recommendations: R=λ·D integrated +(1-λ)·C analysis +μ·P user ; Among them, C analysis represents the key health associations and patterns discovered from the data, P user It contains the user's personal preferences and dietary restrictions, and λ and μ are weighting parameters to balance the influence of different data sources; The diet recommendation algorithm uses the output of the hybrid model R, the user's actual health results, and the consistency between the user's long-term health goals to calculate the loss function: Where N is the number of samples, t is the sample index, and R(t; Θ hybrid ) is the dietary recommendation at time t according to the hybrid algorithm, H actual (t) is the actual health result of the user at time t, H goal is the user’s long-term health goal, and λ is the regularization parameter used to balance the importance of short-term and long-term health goals.

6. A method for generating dynamic adaptive dietary advice based on real-time biomarker monitoring according to claim 5, characterized in that: The lifestyle information i(L lifestyle ) Combined with exercise intensity, exercise type, sleep duration and sleep quality factors, a nonlinear mixed model was used to describe: in, and Represent different types of exercise and sleep parameters, ω k and ρ m is the weight of each parameter, v1 and v2 are coefficients for adjusting the overall impact, and the hyperbolic tangent function tanh and exponential function are used to capture complex nonlinear relationships; The mental health data j(M mental ) Integrate the dynamic and nonlinear characteristics of mental health data to describe the combined effects of stress and emotional state: in, and Represent the parameters of different stress and emotional states, σ q and τ s is the weight coefficient, ξ1 and ξ2 are used to adjust the overall impact, and the logarithmic and fractional forms are used to capture the complexity and nonlinear response of psychological states.

7. The method for generating dynamic adaptive dietary advice based on real-time biomarker monitoring according to claim 1, characterized in that: The S5 specifically includes: S51. The system will collect and analyze the user's feedback on previous dietary recommendations. user , including user satisfaction ratings, frequency of following recommendations, and self-reports of any health changes. User feedback will be combined with real-time biomarker monitoring data: F integrated =η·F user +(1-n)·D integrated ; Among them, η is a parameter that balances the influence of user feedback and biomarker data; S52, combined with integrated user feedback and biomarker data, using a hybrid algorithm f hybrid Dietary adjustments: R adjusted =R+θ·f hybrid (F integrated ,R previous ;I hybrid ); Where R is the previous diet recommendation, θ is the learning rate, and Θ hybrid are algorithm parameters; S53. Use optimized dietary advice R adjusted Minimize the prediction error of the user's health outcomes and consider the user's long-term health goals:

8. The method for generating dynamic adaptive dietary advice based on real-time biomarker monitoring according to claim 1, characterized in that: It also includes establishing an expert diet library, which includes working with nutritionists, doctors and health experts to collect dietary recommendations for different health conditions, physiological indicators and nutritional needs, classifying and labeling the collected dietary recommendations according to different health conditions and nutritional goals, and creating a data model for matching the user's biomarker data, historical health records and lifestyle with the recommendations in the expert diet library; The establishment of the expert diet library includes the use of time series analysis and statistical methods to extract key health characteristics: Among them, α, β, and γ are weight parameters used to adjust the contribution of different data sources. j and They represent specific values ​​of historical health data and lifestyle respectively; Combine the extracted features and LSTM model to evaluate the user's current health status: Where σ is the activation function, W and b are the weights and biases of the LSTM model, and p is the number of features. Combine health status assessment results with user preferences to select and optimize dietary recommendations: Among them, R(t; Θ select ) are dietary recommendations selected from an expert diet library, H actual (t) and H goal are the user’s actual health status and long-term health goal at time t, N is the number of samples, and λ is the regularization parameter.

Images