Personalized digital health recommendation method based on digital twin and generative ai
By constructing personalized digital twin models and generative AI models, and combining multi-source data and real-time feedback, the shortcomings of existing technologies in personalized health management are addressed. This enables dynamic adjustment and real-time monitoring of personalized health recommendations, meeting the individual differences of users.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- GUANGZHOU HUYUN HOSPITAL MANAGEMENT CO LTD
- Filing Date
- 2026-03-16
- Publication Date
- 2026-06-12
AI Technical Summary
Existing personalized health management methods lack sufficient consideration of individual differences, making it difficult to meet the personalized needs of different users. They also have limitations in integrating multi-source health data, monitoring users' health status in real time, and dynamically adjusting health recommendations.
By collecting user status data through multiple data acquisition channels, a personalized digital twin model is constructed. A generative AI model is used to conduct health risk assessments, and personalized health recommendations are generated by combining professional medical knowledge. Feedback information is received in real time for dynamic adjustments.
It enables personalized health recommendations based on individual differences, integrates multi-source health data, monitors users' health status in real time and dynamically adjusts recommendations to meet the personalized needs of different users.
Smart Images
Figure CN122201777A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of digital health technology, and in particular to a personalized digital health recommendation method based on digital twins and generative AI. Background Technology
[0002] With increasing health awareness and rapid advancements in medical technology, personalized health management has gradually become an important means of improving public health. In recent years, digital twin technology and generative AI technology have been widely applied in various fields, bringing new opportunities for personalized health management.
[0003] Digital twin technology is a technique that creates virtual digital models of physical entities and uses real-time data to drive model updates, thereby achieving dynamic simulation and optimization of the physical entities. In the healthcare field, digital twin technology can be used to build digital twin models of patients. These models can comprehensively reflect multi-dimensional information such as the patient's physiological state, medical history, and lifestyle habits, providing strong support for precision medicine and personalized health management. Generative AI technology has powerful data processing and content generation capabilities, capable of generating text, images, and other content that meet specific needs based on input data. In the health field, generative AI can generate personalized health recommendations based on patient health data, providing users with more accurate and scientific health guidance.
[0004] However, current personalized health management methods still have some shortcomings. Traditional health advice is often based on general health guidelines, lacking sufficient consideration for individual differences and failing to meet the personalized needs of different users. Furthermore, existing health management methods also have limitations in integrating multi-source health data, real-time monitoring of user health status, and dynamic adjustment of health advice. Therefore, we propose a personalized digital health advice method based on digital twins and generative AI. Summary of the Invention
[0005] The main objective of this invention is to provide a personalized digital health recommendation method based on digital twins and generative AI, which can effectively solve the problems in the background art.
[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: Personalized digital health recommendation approaches based on digital twins and generative AI include: Step 1: Collect user status data through multiple data acquisition channels, including health monitoring devices, mobile apps, and electronic medical records; the status data includes physiological indicator data, lifestyle habit data, and genetic information data. Perform preprocessing on the collected multi-dimensional data, including cleaning, noise reduction, and standardization, to remove invalid and erroneous data, and convert it into a unified format and standard. Step 2: Using the collected multi-dimensional status data, construct a personalized digital twin model that reflects the user's physiological structure, function, and health status. The personalized digital twin model is dynamically updated based on the user's status data collected in real time. Step 3: Based on the user's current health status, a generative AI model is used to comprehensively assess the user's health risks. By comparing industry standards, medical research results, and big data analysis, the potential health risks that the user may face are identified. Step 4: Based on the user's health status and risk assessment results, combined with professional medical knowledge and the latest research findings, generate personalized health recommendations that cover one or more of the following: dietary adjustments, exercise plans, psychological adjustment, drug treatment, and regular physical examinations.
[0007] Furthermore, the method also includes: Step 5: The generative AI model receives feedback information in real time after the user adopts personalized health advice. The feedback information includes the user's execution status and changes in health status. The personalized health advice is dynamically adjusted based on the obtained feedback information.
[0008] Furthermore, the method also includes: Step Six: Dynamically update the status data based on changes in the user's health status and synchronously feed it back to the personalized digital twin model. Adjust the parameters of the personalized digital twin model based on the feedback information to optimize the model's assessment performance of the user's physiological structure, function, and health status.
[0009] Furthermore, in step five, the specific process for dynamically adjusting the personalized health recommendations based on the obtained feedback information includes the following steps: Obtain personalized health advice information for users, including the first personalized health advice, the second personalized health advice, ..., the nth personalized health advice; The execution status of the i-th personalized health suggestion is classified into two categories: executed and not executed. The execution status of the user for all personalized health suggestions is obtained, where i=1,...,n; Changes in health risks are categorized into three types: increased health risk, unchanged health risk, and decreased health risk. The changes in users' health risks are determined based on the feedback information obtained. Based on the implementation status of the acquired personalized health advice and the changes in health risks, the personalized health advice is classified into four categories: one category of advice that is implemented and the health risk decreases; two categories of advice that is implemented and the health risk increases or remains unchanged; three categories of advice that is not implemented and the health risk decreases; and four categories of advice that is not implemented and the health risk increases or remains unchanged. Based on the classification results of personalized health recommendations, adjustments are made to generate recommended adjustment strategies.
[0010] Furthermore, the specific details of the adjustment strategy are as follows: For recommendations that have been implemented and have reduced health risks, the strategy is to retain and strengthen their specific content while further refining the details. For Category II recommendations that have been implemented and whose health risks have increased or remained unchanged, the adjustment strategy is to reassess the scientific validity and feasibility of the recommendations and adjust the recommendations accordingly. For the three types of recommendations that were not implemented but whose health risks have decreased, the adjustment strategy is as follows: analyze the reasons why users did not implement them, and adjust the wording and implementation steps of the recommendations. For the four categories of recommendations that were not implemented and whose health risks increased or remained unchanged, the adjustment strategy is to change the content of the recommendations or add incentives to increase users' willingness to implement them.
[0011] Furthermore, in step three, the specific assessment process for the health risks that users may face includes the following steps: Identify the target disease that the user needs to assess, obtain the status data of patients with the target disease, and extract risk characteristics related to the target disease from the massive status data of patients with the target disease through data mining technology; A risk assessment model for the target disease is constructed, and the obtained risk characteristic data is used as the input data for the risk assessment model. The model is trained until its prediction accuracy meets the set expected value, and the trained risk assessment model is obtained. The user's status data is input into the trained risk assessment model, and the acquired risk assessment model is used to assess the user's health risk for the target disease.
[0012] Furthermore, the risk characteristics include uncontrollable factors, controllable factors, environmental factors, medical history, and socioeconomic factors; among which, The uncontrollable factors include at least one of the following: age, sex, family medical history, and genes; The controllable factors include at least one of the following: lifestyle habits, physiological indicators, and sleep and psychological factors; The environmental factors include at least one of occupational exposure and environmental pollution; The medical history features include at least one of the following: past illnesses and medication use; The socioeconomic factors mentioned include at least one of education level and income level.
[0013] The present invention has the following beneficial effects: Compared to existing technologies, this method collects user status data through multiple data acquisition channels, performs preprocessing on the collected multi-dimensional data including cleaning, noise reduction, and standardization, and uses this multi-dimensional status data to construct a personalized digital twin model reflecting the user's physiological structure, function, and health status. Based on the user's current health status, a generative AI model is used to comprehensively assess the user's health risks. By comparing industry standards, medical research results, and big data analysis, potential health risks faced by the user are identified. Based on the user's health status and risk assessment results, combined with professional medical knowledge and the latest research findings, personalized health recommendations are generated, covering one or more combinations of dietary adjustments, exercise plans, psychological adjustment, drug treatment, and regular physical examinations. This method can integrate multi-source health data and, based on full consideration of individual differences, formulate health recommendations that meet the personalized needs of different users, thereby achieving real-time monitoring and dynamic adjustment of the user's health status. Attached Figure Description
[0014] Figure 1 This is a flowchart illustrating the personalized digital health recommendation method based on digital twins and generative AI of the present invention. Detailed Implementation
[0015] The present invention will be further described below with reference to specific embodiments. The accompanying drawings are for illustrative purposes only and are schematic diagrams, not actual pictures. They should not be construed as limiting the present invention. In order to better illustrate the specific embodiments of the present invention, some parts in the drawings may be omitted, enlarged or reduced, and do not represent the actual product size.
[0016] The specific implementation process of the technical solution of this invention includes the following steps: Step 1: Collect user status data through multiple data acquisition channels, perform preprocessing on the collected multi-dimensional data including cleaning, noise reduction and standardization, remove invalid and erroneous data, and convert it into a unified format and standard.
[0017] The data collection channels include health monitoring devices, mobile apps, and electronic medical records, such as wearable devices and home medical devices. Status data includes physiological indicators (such as heart rate, blood pressure, blood sugar, etc.), lifestyle data (such as diet, exercise, sleep, etc.) and genetic information data.
[0018] Step 2: Using the collected multi-dimensional status data, construct a personalized digital twin model that reflects the user's physiological structure, function, and health status. The personalized digital twin model is dynamically updated based on the user's status data collected in real time.
[0019] Step 3: Based on the user's current health status, use a generative AI model to comprehensively assess the user's health risks. By comparing industry standards, medical research findings, and big data analysis, identify the health risks the user may face.
[0020] The specific assessment process for potential health risks faced by users includes the following steps: Identify the target disease that the user needs to assess, obtain the status data of patients with the target disease, extract risk features related to the target disease from the massive status data of patients with the target disease through data mining techniques, and reduce data dimensionality and improve model processing efficiency by using feature selection algorithms (such as principal component analysis PCA). To construct a risk assessment model for the target disease, the choice of risk assessment model can be as follows: Machine learning model: Choose an appropriate machine learning algorithm based on the specific application scenario, such as logistic regression, decision tree, random forest, support vector machine (SVM), etc.; Deep learning model: For complex data patterns, deep learning models can be used, such as convolutional neural network (CNN), recurrent neural network (RNN), etc. The risk characteristic data obtained is used as input data for the risk assessment model. The model is trained until its prediction accuracy meets the set expectation value, and then the trained risk assessment model is obtained. The specific training process is as follows: a) Dataset partitioning: The dataset is divided into training set and test set, usually in an 8:2 or 7:3 ratio; b) Model training: The selected model is trained using the training set to optimize the model parameters; c) Model validation: The model performance is evaluated using the test set. Commonly used evaluation metrics include accuracy, recall, precision, F1 score, and AUC score.
[0021] The user's status data is input into the trained risk assessment model, and the acquired risk assessment model is used to assess the user's health risk for the target disease. Specifically, the following steps can be taken: Data collection and integration Electronic Health Record (EHR) analysis: Using detailed information in electronic health records, including medical history, diagnosis, treatment process, medication use, etc., potential health risks can be identified through big data analytics.
[0022] Wearable device data: Analyzing data from wearable devices, such as activity levels, sleep patterns, and heart rate, can help identify chronic disease risks or early signs of disease.
[0023] Genomics data: By analyzing genomic data, we can identify genetically related disease risks, such as certain cancers and cardiovascular diseases.
[0024] Industry standard comparison Clinical guidelines and standards: Compare users' health data with industry standards and clinical guidelines, such as using standards published by organizations like the World Health Organization (WHO) or the American Heart Association (AHA) to assess users' cardiovascular disease risk.
[0025] Risk assessment model: Validated risk assessment models, such as the Framingham Cardiovascular Risk Score, are used to quantify the user's risk of developing cardiovascular disease.
[0026] Application of medical research findings Latest research findings: Referencing the latest medical research findings, such as those on the prediction of diabetes complications and mental health interventions, to identify specific health risks that users may face.
[0027] Clinical trial data: Using clinical trial data to assess a user’s response to a specific drug or treatment, thereby predicting potential health risks.
[0028] Big data analytics Predictive analytics: Through big data analytics techniques, such as machine learning and data mining, patterns and trends are identified from large amounts of patient data to predict potential health risks for users.
[0029] Real-time monitoring and early warning: Utilize real-time big data analysis to continuously monitor users' health status, promptly detect abnormalities, and issue early warnings.
[0030] Comprehensive assessment and personalized recommendations Multi-dimensional assessment: Combining the analysis results of the above methods, a comprehensive assessment of the user's health risks is conducted to identify high-risk areas.
[0031] Personalized recommendations: Based on the assessment results, provide users with personalized health advice, including lifestyle adjustments, medication, and psychological interventions, to reduce health risks.
[0032] By using the methods described above, we can more accurately identify the health risks that users may face and provide them with targeted health management advice.
[0033] It should be noted that characteristics related to health risks can be classified and identified from multiple dimensions, mainly including characteristics of uncontrollable factors, characteristics of controllable factors, characteristics of environmental factors, characteristics of medical history, and socioeconomic factors. The following is a detailed explanation of risk characteristics: Uncontrollable factors Age: Age is a significant risk factor for many diseases. For example, the risk of cardiovascular disease, cancer, and diabetes typically increases with age.
[0034] Gender: Some diseases are more common in certain genders. For example, women are more likely to develop breast cancer, while men are more likely to develop prostate cancer.
[0035] Family history: Having certain diseases in one's family increases an individual's risk of developing the same disease. For example, if there is a family history of cardiovascular disease or diabetes, the individual's risk of developing these diseases will also increase.
[0036] Genes: Certain gene mutations or genetic traits increase the risk of specific diseases. For example, BRCA gene mutations are associated with an increased risk of breast and ovarian cancer.
[0037] Controllable factors characteristics Lifestyle, specifically, can be: Diet: Unhealthy eating habits, such as a diet high in sugar, salt, and fat, can increase the risk of obesity, cardiovascular disease, and diabetes.
[0038] Exercise: Lack of exercise is a risk factor for many chronic diseases. Regular aerobic exercise and strength training can reduce the risk of cardiovascular disease, diabetes, and certain cancers.
[0039] Sleep: Insufficient sleep or poor sleep quality can affect mental health and physiological function, and increase the risk of cardiovascular disease, diabetes and obesity.
[0040] Smoking and drinking: Smoking is a major risk factor for many cancers, cardiovascular diseases, and respiratory diseases. Excessive alcohol consumption also increases the risk of liver disease, cardiovascular disease, and certain cancers.
[0041] Physiological indicators, specifically: Blood pressure: High blood pressure is a significant risk factor for cardiovascular disease and stroke.
[0042] Blood sugar: High blood sugar is a major risk factor for diabetes, and long-term high blood sugar can also increase the risk of cardiovascular disease and kidney disease.
[0043] Blood lipids: High cholesterol and high triglyceride levels increase the risk of cardiovascular disease.
[0044] Body Mass Index (BMI): Obesity is a risk factor for many chronic diseases, including cardiovascular disease, diabetes, and certain cancers.
[0045] Psychological factors can be specifically: Psychological stress: Long-term psychological stress can affect mental health and increase the risk of cardiovascular disease, diabetes, and certain mental illnesses.
[0046] Emotional state: Emotional problems such as depression and anxiety can affect lifestyle and health behaviors, increasing the risk of chronic diseases.
[0047] Environmental factors characteristics Occupational exposure: Long-term exposure to occupational environments with harmful chemicals or dust increases the risk of certain diseases, such as occupational cancers.
[0048] Environmental pollution: Environmental factors such as air pollution, water pollution, and soil pollution can affect health and increase the risk of cardiovascular disease, respiratory diseases, and certain cancers.
[0049] Medical history characteristics Pre-existing conditions: Having certain pre-existing conditions increases the risk of related complications or recurrence. For example, patients with a history of cardiovascular disease have a higher risk of recurrence.
[0050] Medication use: Long-term use of certain medications may increase the risk of certain diseases, such as long-term use of glucocorticoids, which may increase the risk of osteoporosis and diabetes.
[0051] Socioeconomic factors characteristics Education level: Lower education level may be associated with unhealthy lifestyle and poor health.
[0052] Income level: Lower income levels may limit access to health resources and increase the risk of chronic diseases.
[0053] By taking these characteristics into account, we can more comprehensively identify the health risks that users may face and develop personalized health management plans.
[0054] Step 4: Based on the user's health status and risk assessment results, combined with professional medical knowledge and the latest research findings, generate personalized health recommendations that cover one or more of the following: dietary adjustments, exercise plans, psychological adjustment, drug treatment, and regular physical examinations.
[0055] Personalized health recommendations can include: Dietary adjustment recommendations Personalized Diet Plans: Based on the user's health status and risk assessment results, and in conjunction with nutritional research, personalized diet plans are developed. For example, for users at risk of diabetes, a low-sugar, high-fiber diet is recommended; for obese users, it is recommended to control calorie intake and increase the proportion of vegetables and whole grains.
[0056] Precision nutrition support: Referencing the latest research findings, such as the Precision Health Nutrition Project of the National Institutes of Health, machine learning and statistical models are used to identify the personal characteristics most relevant to dietary choices and predict which foods and dietary patterns are more beneficial to individuals.
[0057] Exercise plan suggestions Exercise Goal Setting: Based on the user's health condition and lifestyle, specific exercise goals can be set. For example, it is recommended to engage in aerobic exercise at least three times a week, each time lasting more than 30 minutes, to improve cardiopulmonary function.
[0058] Personalized exercise plans: Based on the user's interests and physical condition, suitable types of exercise are recommended, such as swimming, yoga, and running, and suggestions are provided on exercise intensity and frequency.
[0059] Psychological adjustment suggestions Psychological state assessment: Through psychological tests and questionnaires, assess the user's psychological state and identify potential psychological stress and emotional problems.
[0060] Psychological adjustment strategies: Provide targeted psychological adjustment suggestions, such as stress management techniques, emotion regulation methods, relaxation training, etc., to help users maintain a good mental state.
[0061] Drug treatment recommendations Drug treatment plans: Based on the user's health status and risk assessment results, and referring to the latest medical research findings, we provide personalized drug treatment plans for users who require drug intervention. For example, for patients with high cholesterol, we recommend the use of novel PCSK9 inhibitors, such as ricarcisumab, due to its longer half-life and good adherence.
[0062] Medication usage instructions: Provide detailed instructions for use, including dosage, timing of administration, precautions, etc., to ensure that users use the medication correctly.
[0063] Recommendations for a regular physical examination plan Customized health checkup programs: Personalized health checkup programs are tailored based on the user's health status and risk assessment results. For example, for women at risk of breast cancer, individualized screening based on the risk of developing the disease is recommended.
[0064] Recommended frequency of physical examinations: Based on the user's specific circumstances, we recommend an appropriate frequency of physical examinations to ensure that potential health problems are detected in a timely manner.
[0065] Step 5: The generative AI model receives feedback information in real time after the user adopts personalized health advice. The feedback information includes the user's implementation status and changes in health status. The personalized health advice is dynamically adjusted based on the feedback information.
[0066] The specific process for dynamically adjusting personalized health recommendations based on the feedback information obtained includes the following steps: Obtain personalized health advice information for users, including the first personalized health advice, the second personalized health advice, ..., the nth personalized health advice; The execution status of the i-th personalized health suggestion is classified into two categories: executed and not executed. The execution status of the user for all personalized health suggestions is obtained, where i=1,...,n; Changes in health risks are categorized into three types: increased health risk, unchanged health risk, and decreased health risk. The changes in users' health risks are determined based on the feedback information obtained. Based on the implementation status of the obtained personalized health advice and the changes in health risks, the personalized health advice is classified into four categories: Category 1 advice that is implemented and the health risk decreases; Category 2 advice that is implemented and the health risk increases or remains unchanged; Category 3 advice that is not implemented and the health risk decreases; and Category 4 advice that is not implemented and the health risk increases or remains unchanged. Based on the classification results of personalized health recommendations, adjustments are made to generate suggested adjustment strategies, the specific content of which is as follows: For recommendations that have been implemented and have reduced health risks, the strategy is to retain and strengthen their specific content, while further optimizing the details of the content to improve user compliance and effectiveness. For Category II recommendations that have been implemented and whose health risks have increased or remained unchanged, the adjustment strategy is to reassess the scientific validity and feasibility of the recommendations and adjust them in conjunction with the latest medical research and expert opinions. For the three types of recommendations that were not implemented but whose health risks have decreased, the strategy is to analyze the reasons why users did not implement them (such as excessive difficulty, lack of motivation, etc.), and adjust the wording and implementation steps of the recommendations to make them easier to understand and implement. For the four categories of recommendations that were not implemented and whose health risks increased or remained unchanged, the adjustment strategy is to change the content of the recommendations or add incentives to increase users' willingness to implement them.
[0067] Step 6: Dynamically update the status data based on changes in the user's health status and synchronously feed it back to the personalized digital twin model. Adjust the parameters of the personalized digital twin model based on the feedback information to optimize the model's assessment performance of the user's physiological structure, function, and health status.
[0068] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.
Claims
1. A personalized digital health recommendation method based on digital twins and generative AI, characterized in that, include: Step 1: Collect user status data through multiple data acquisition channels, including health monitoring devices, mobile apps, and electronic medical records; the status data includes physiological indicator data, lifestyle habit data, and genetic information data. Perform preprocessing on the collected multi-dimensional data, including cleaning, noise reduction, and standardization, to remove invalid and erroneous data, and convert it into a unified format and standard. Step 2: Using the collected multi-dimensional status data, construct a personalized digital twin model that reflects the user's physiological structure, function, and health status. The personalized digital twin model is dynamically updated based on the user's status data collected in real time. Step 3: Based on the user's current health status, a generative AI model is used to comprehensively assess the user's health risks. By comparing industry standards, medical research results, and big data analysis, the potential health risks that the user may face are identified. Step 4: Based on the user's health status and risk assessment results, combined with professional medical knowledge and the latest research findings, generate personalized health recommendations that cover one or more of the following: dietary adjustments, exercise plans, psychological adjustment, drug treatment, and regular physical examinations.
2. The personalized digital health recommendation method based on digital twins and generative AI according to claim 1, characterized in that, The method further includes: Step 5: The generative AI model receives feedback information in real time after the user adopts personalized health advice. The feedback information includes the user's execution status and changes in health status. The personalized health advice is dynamically adjusted based on the obtained feedback information.
3. The personalized digital health recommendation method based on digital twins and generative AI according to claim 2, characterized in that, The method further includes: Step Six: Dynamically update the status data based on changes in the user's health status and synchronously feed it back to the personalized digital twin model. Adjust the parameters of the personalized digital twin model based on the feedback information to optimize the model's assessment performance of the user's physiological structure, function, and health status.
4. The personalized digital health recommendation method based on digital twins and generative AI according to claim 2, characterized in that, In step five, the specific process of dynamically adjusting the personalized health recommendations based on the obtained feedback information includes the following steps: Obtain personalized health advice information for users, including the first personalized health advice, the second personalized health advice, ..., the nth personalized health advice; The execution status of the i-th personalized health suggestion is classified into two categories: executed and not executed. The execution status of the user for all personalized health suggestions is obtained, where i=1,...,n; Changes in health risks are categorized into three types: increased health risk, unchanged health risk, and decreased health risk. The changes in users' health risks are determined based on the feedback information obtained. Based on the implementation status of the acquired personalized health advice and the changes in health risks, the personalized health advice is classified into four categories: one category of advice that is implemented and the health risk decreases; two categories of advice that is implemented and the health risk increases or remains unchanged; three categories of advice that is not implemented and the health risk decreases; and four categories of advice that is not implemented and the health risk increases or remains unchanged. Based on the classification results of personalized health recommendations, adjustments are made to generate recommended adjustment strategies.
5. The personalized digital health recommendation method based on digital twins and generative AI according to claim 4, characterized in that, The specific details of the adjustment strategy are as follows: For recommendations that have been implemented and have reduced health risks, the strategy is to retain and strengthen their specific content while further refining the details. For Category II recommendations that have been implemented and whose health risks have increased or remained unchanged, the adjustment strategy is to reassess the scientific validity and feasibility of the recommendations and adjust the recommendations accordingly. For the three types of recommendations that were not implemented but whose health risks have decreased, the adjustment strategy is as follows: analyze the reasons why users did not implement them, and adjust the wording and implementation steps of the recommendations. For the four categories of recommendations that were not implemented and whose health risks increased or remained unchanged, the adjustment strategy is to change the content of the recommendations or add incentives to increase users' willingness to implement them.
6. The personalized digital health recommendation method based on digital twins and generative AI according to claim 1, characterized in that, In step three, the specific assessment process for the health risks that users may face includes the following steps: Identify the target disease that the user needs to assess, obtain the status data of patients with the target disease, and extract risk characteristics related to the target disease from the massive status data of patients with the target disease through data mining technology; A risk assessment model for the target disease is constructed, and the obtained risk characteristic data is used as the input data for the risk assessment model. The model is trained until its prediction accuracy meets the set expected value, and the trained risk assessment model is obtained. The user's status data is input into the trained risk assessment model, and the acquired risk assessment model is used to assess the user's health risk for the target disease.
7. The personalized digital health recommendation method based on digital twins and generative AI according to claim 6, characterized in that, The risk characteristics include uncontrollable factors, controllable factors, environmental factors, medical history, and socioeconomic factors; among which, The uncontrollable factors include at least one of the following: age, sex, family medical history, and genes; The controllable factors include at least one of the following: lifestyle habits, physiological indicators, and sleep and psychological factors; The environmental factors include at least one of occupational exposure and environmental pollution; The medical history features include at least one of the following: past illnesses and medication use; The socioeconomic factors mentioned include at least one of education level and income level.