Chronic disease risk prediction and early intervention method based on AI

Abstract

The invention relates to the technical field of chronic disease examination, in particular to an AI-based chronic disease risk prediction and early intervention method, which comprises the following steps of: acquiring physiological data, living habit data, environmental data and historical health data of a user through various devices and platforms, and then preprocessing the acquired data to obtain an AI-based chronic disease risk prediction and early intervention result; the method includes the steps of data cleaning, missing value filling and standardization, then conducting feature selection through a principal component analysis method, extracting key features related to chronic disease risks, building a chronic disease risk prediction model through an improved random forest algorithm on the basis of the key features, and finally predicting the chronic disease risks in combination with a model prediction result. Personalized intervention suggestions are provided for the user, and diet adjustment, exercise plans, drug prevention, regular physical examination and other aspects are covered; according to the method, high-risk people with chronic diseases can be accurately identified in an early stage, and accurate risk prediction and personalized intervention strategies are provided, so that occurrence and development of the chronic diseases are effectively prevented.

Description

Technical Field

[0001] The present invention relates to the technical field of chronic disease examination, and in particular to a chronic disease risk prediction and early intervention method based on AI. Background Art

[0002] Chronic diseases have become an important factor affecting public health. Chronic diseases such as cardiovascular disease, diabetes, hypertension, and cancer often have a long incubation period and are hidden. Once they occur, they often require long-term treatment, increasing the medical burden and social costs. In order to effectively respond to this challenge, early detection and intervention have become the key to reducing the incidence of chronic diseases.

[0003] There are several significant problems with existing technologies in the construction of chronic disease risk prediction models. Although traditional machine learning methods can process high-dimensional data, they are prone to overfitting the model when faced with multiple highly correlated or redundant features, thereby reducing the prediction accuracy. In addition, existing models lack intelligent feature selection methods and cannot automatically identify key features that are highly correlated with disease risk, resulting in inaccurate prediction results. Existing technologies are limited in the output results of chronic disease risk prediction. Although the model usually outputs risk levels, these results are too rough and lack detailed risk probabilities or more detailed predictions, and cannot provide sufficient decision-making basis for health managers. Summary of the invention

[0004] The present invention provides an AI-based chronic disease risk prediction and early intervention method.

[0005] The AI-based chronic disease risk prediction and early intervention method includes the following steps:

[0006] S1, data collection: collect multi-source data of users through various channels, including physiological data, living habit data, environmental data and health history data;

[0007] The physiological data include blood pressure, blood sugar, heart rate and blood lipids;

[0008] The lifestyle data includes eating habits, exercise volume and work and rest time;

[0009] The environmental data includes air quality, temperature and humidity;

[0010] The health history data includes the user's past medical history and family medical history;

[0011] S2, data preprocessing: preprocess the collected multi-source data, including data cleaning and missing value filling, and perform data standardization;

[0012] S3, feature selection: principal component analysis was used to extract features from the preprocessed multi-source data to extract key features related to chronic disease risk;

[0013] S4, model construction: Based on the extracted key features, the improved random forest algorithm was used to build a chronic disease risk prediction model;

[0014] S5, risk prediction: using the chronic disease risk prediction model, based on the user's multi-source data, predict the user's risk probability of chronic diseases in the future;

[0015] S6, Early Intervention Recommendations: Based on the predicted risk probability, provide users with personalized intervention recommendations, including diet adjustments, exercise plans, drug prevention, and regular physical examinations.

[0016] Optionally, the S1 includes:

[0017] S11, physiological data collection: collecting the user's physiological data through a variety of devices and sensors;

[0018] S12, life habit data collection: collect users’ life habit data through smart health management platform, mobile applications and user self-report;

[0019] S13, environmental data collection: collecting environmental data of the user's environment through environmental monitoring equipment and sensors;

[0020] S14, health history data collection: collect the user's health history data through the medical platform, user self-report or related health examination records.

[0021] Optionally, S2 includes:

[0022] S21, data cleaning: cleaning the collected multi-source data;

[0023] S22, missing value filling: fill in the missing parts in multi-source data;

[0024] S23, data standardization: standardize the preprocessed data to eliminate the dimensional differences between different data sources.

[0025] Optionally, the S3 includes:

[0026] S31, application of principal component analysis: principal component analysis is used to extract features from preprocessed multi-source data;

[0027] S32, Select key features: Select key features related to the risk of chronic diseases, including blood pressure, blood sugar, heart rate, diet and exercise.

[0028] Optionally, the method of constructing a chronic disease risk prediction model based on the extracted key features using an improved random forest algorithm specifically includes:

[0029] S41, feature weighting: After feature selection, different features are weighted to improve the contribution of key features to the chronic disease risk prediction model;

[0030] S42, improved tree generation strategy: adopt the tree generation strategy in random forest and improve it by introducing feature-weighted decision tree construction method, so that each tree gives priority to features with higher importance when generating;

[0031] S43, Ensemble strategy: In the ensemble learning of random forests, different tree structures are used to train the same feature set multiple times, and voting decisions are made in combination with weighted features;

[0032] S44, Model evaluation: The chronic disease risk prediction model was evaluated using the cross-validation method, and the evaluation indicators included accuracy, recall rate, and F1 value.

[0033] Optionally, the S5 includes:

[0034] S51, model prediction: Based on the collected multi-source data of users, the risk probability of users developing chronic diseases in the future is calculated through the trained chronic disease risk prediction model;

[0035] S52, output result: The chronic disease risk prediction model calculates the risk score based on the multi-source data input by the user, and outputs three levels: high risk, medium risk, and low risk, and generates corresponding risk assessment results for each level.

[0036] Optionally, S5 further includes:

[0037] S53, report generation: generating a personalized user risk report based on the risk prediction results;

[0038] S54, Report output: The risk report is output via the user’s smart device or mobile application, with options for downloading, printing or storing.

[0039] Optionally, the S6 includes:

[0040] S61, assess intervention needs: determine the user's intervention needs based on the risk prediction results;

[0041] S62, intervention suggestion generation: generating personalized intervention suggestions based on the user's risk level and health status;

[0042] S63, Intervention suggestion feedback: Feedback the generated personalized intervention suggestion to the user via a mobile application, SMS or email.

[0043] Optionally, S6 further includes:

[0044] S64, intervention implementation monitoring: monitoring the user's implementation of early intervention recommendations through smart devices, health management platforms, or user manual input;

[0045] S65, implementation feedback analysis: Analyze the degree of user compliance with intervention recommendations based on the user's implementation status;

[0046] S66, Dynamically adjust intervention recommendations: Dynamically adjust intervention recommendations based on the results of implementation analysis to generate an updated version of the personalized intervention plan;

[0047] S67, Feedback adjustment results: Feedback the updated personalized intervention plan to the user.

[0048] Beneficial effects of the present invention:

[0049] The present invention constructs a complete user health portrait by collecting multi-source data, including physiological data, living habit data, environmental data and health history data. By adopting the principal component analysis method for feature extraction, the key features that are highly correlated with the risk of chronic diseases can be accurately identified, thereby providing more accurate data support for predicting the risk of chronic diseases. Using the improved random forest algorithm, this method can not only process multidimensional data, but also automatically learn the influence weights of different features on disease risks, and achieve more accurate risk prediction. The prediction results are divided into three levels: high risk, medium risk and low risk, providing a reliable basis for subsequent personalized intervention.

[0050] The present invention constructs a chronic disease risk prediction model by adopting an improved random forest algorithm. The model is trained based on key features extracted from multi-source data, and can effectively capture the complex relationship between different health factors and chronic disease risks. The improved random forest algorithm can handle nonlinear relationships in the data through an integrated learning mechanism, and has strong resistance to noise and overfitting. Compared with traditional prediction methods, it can provide higher accuracy and stability. In addition, the algorithm also has strong interpretability, which can help health managers deeply understand the specific contribution of different factors to disease risks, and then provide a scientific basis for the formulation of personalized intervention strategies, significantly improving the accuracy, reliability and interpretability of the prediction results, and providing more precise support for subsequent health management and intervention.

[0051] The present invention generates personalized intervention suggestions based on the results of chronic disease risk prediction, including diet adjustment, exercise plan, drug prevention and regular physical examination. Especially for high-risk users, the intervention suggestions are targeted and accurate, which can help them effectively control risks and delay the occurrence of diseases. For medium and low-risk users, potential health problems can be prevented by maintaining a healthy lifestyle. Further, by monitoring the user's execution, the intervention plan is dynamically adjusted to ensure the continued effectiveness of the intervention measures. In this way, the probability of chronic diseases can be minimized and the health level of users can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0053] Figure 1 A schematic diagram of a method flow of an embodiment of the present invention;

[0054] Figure 2 Schematic diagram of S4 process according to an embodiment of the present invention. DETAILED DESCRIPTION

[0055] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0056] It should be noted that the references to "one embodiment", "an embodiment", "an exemplary embodiment", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0057] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0058] like Figure 1-Figure 2 As shown, the AI-based chronic disease risk prediction and early intervention method includes the following steps:

[0059] S1, data collection: collect multi-source data of users through various channels, including physiological data, living habit data, environmental data and health history data;

[0060] Physiological data include blood pressure, blood sugar, heart rate, and blood lipids;

[0061] Lifestyle data includes eating habits, exercise volume, and sleep schedule;

[0062] Environmental data includes air quality, temperature, and humidity;

[0063] Health history data includes the user’s past medical history and family medical history;

[0064] S2, data preprocessing: preprocess the collected multi-source data, including data cleaning and missing value filling, and perform data standardization;

[0065] S3, feature selection: principal component analysis was used to extract features from the preprocessed multi-source data to extract key features related to chronic disease risk;

[0066] S4, model construction: Based on the extracted key features, the improved random forest algorithm was used to build a chronic disease risk prediction model;

[0067] S5, risk prediction: using the chronic disease risk prediction model, based on the user's multi-source data, predict the user's risk probability of chronic diseases in the future;

[0068] S6, Early Intervention Recommendations: Based on the predicted risk probability, provide users with personalized intervention recommendations, including diet adjustments, exercise plans, drug prevention, and regular physical examinations.

[0069] S1 includes:

[0070] S11, Physiological data collection: The user's physiological data is collected through a variety of devices and sensors, including:

[0071] Blood pressure data collection: The user's systolic and diastolic blood pressure values ​​are collected through a blood pressure monitor, and the collection timestamp is recorded to analyze the fluctuation and trend of blood pressure;

[0072] Blood glucose data collection: Use a blood glucose monitor or continuous blood glucose monitoring device to collect the user's fasting blood glucose and postprandial blood glucose data. The frequency of blood glucose data collection is adjusted according to the user's health status;

[0073] Heart rate data collection: Wearable devices (such as smart watches, sports bracelets, etc.) are used to collect the user's resting heart rate and heart rate changes during exercise in real time. This data helps analyze the user's heart health and exercise intensity;

[0074] Blood lipid data collection: Obtain the user's total cholesterol, low-density lipoprotein, high-density lipoprotein and triglyceride levels through blood test equipment or regular physical examination data;

[0075] S12, Collection of life habit data: Collect users’ life habit data through the smart health management platform, mobile applications and user self-reports, including:

[0076] Eating habits data: The food tracking app records the user's diet, including total daily calories, carbohydrates, protein, fat and other nutrient intake. Eating habits data is input by the user or automatically collected through intelligent food recognition technology to ensure accuracy;

[0077] Exercise data collection: Collect users' daily exercise data through smart bracelets or sports tracking devices, including daily exercise duration, exercise type, exercise intensity, and calories consumed. The frequency of exercise data collection is dynamically adjusted according to the user's health goals and activity intensity;

[0078] Work and rest schedule data collection: Use sleep monitoring equipment (such as smart mattresses, wearable sleep monitors, etc.) to record the user's work and rest schedule, including daily sleep duration, deep sleep duration, sleep quality and other data. Work and rest schedule data helps analyze the relationship between sleep patterns and chronic disease risks;

[0079] S13, environmental data collection: Collect environmental data of the user's environment through environmental monitoring equipment and sensors, including:

[0080] Air quality data collection: Collect air quality indicators such as PM2.5, PM10, carbon dioxide concentration, and volatile organic compounds at the user's residence or workplace through air quality monitoring equipment or indoor air detectors. This data helps analyze the potential impact of air pollution on user health;

[0081] Temperature data collection: Use temperature and humidity sensors to record the temperature changes of the environment in real time and evaluate the impact of temperature on user health, especially the risk of chronic diseases;

[0082] Humidity data collection: The humidity level of the environment is collected through temperature and humidity sensors. This data helps analyze the impact of humidity on respiratory diseases or skin health, especially for users with allergic diseases.

[0083] S14, health history data collection: collect the user's health history data through the medical platform, user self-report or related health examination records, including:

[0084] Past medical history data collection: Collect the user's past medical history through the user's electronic health record system or medical record, including historical data on chronic diseases such as hypertension, diabetes, heart disease, respiratory diseases, etc. The user's medical history helps predict the risk of similar diseases in the future;

[0085] Family medical history data collection: obtain data on whether family members (such as parents, siblings, etc.) have a history of chronic diseases or hereditary diseases through health questionnaires filled out by users or from family health records.

[0086] S2 includes:

[0087] S21, data cleaning: Clean the collected multi-source data to ensure data quality, including:

[0088] De-duplicate data: De-duplicate data by comparing records in various data sources. Duplicate records may be caused by equipment failure, user operation errors, or data transmission problems. Match the unique identifier of each data point and remove all identical data records;

[0089] Outlier detection: Perform outlier detection on each data type (such as blood pressure, blood sugar, air quality, etc.), identify and mark abnormal data outside the range by setting a reasonable data range (based on historical data or medical standards);

[0090] Unified data format: Ensure that all collected data conforms to a unified format standard. Especially when collecting data from multiple sources, ensure the consistency of fields such as timestamps and units. For example, the unit of blood sugar value must be unified as mmol / L, and the unit of weight must be unified as kilograms.

[0091] S22, missing value filling: fill in the missing parts of multi-source data to reduce the impact of missing data on the analysis results. The specific filling methods include the following:

[0092] Mean / median filling method: For numerical data (such as blood sugar, heart rate, etc.), if a data item is missing, the mean or median of the data item can be used to fill it. This method is more applicable when there is less missing data.

[0093] Interpolation method: For time series data (such as blood sugar, blood pressure, etc.), linear interpolation or spline interpolation can be used to fill in missing values. The specific method is to calculate and insert missing values ​​based on the trend between the previous and next data points to ensure the continuity and rationality of the data;

[0094] Multiple imputation method: For variables with many or important missing values, multiple imputation methods are used to generate multiple possible imputation values, and analysis is performed based on the imputation results to ensure the diversity and reliability of the imputation values;

[0095] S23, data standardization: standardize the preprocessed data to eliminate the dimensional differences between different data sources and ensure that different features contribute equally to model training. Standardization includes the following sub-steps:

[0096] Z-score standardization: Z-score standardization is performed on each feature data. The formula is as follows:

[0097]

[0098] Among them, X is the original data, μ is the mean of the feature, and σ is the standard deviation of the feature. This method can transform the data into a standard normal distribution with a mean of 0 and a standard deviation of 1, so that the dimensions of different features are unified;

[0099] Minimum-maximum standardization: For some characteristics (such as air quality, temperature, etc.), the minimum-maximum standardization method is used, and the formula is as follows:

[0100]

[0101] Among them, X is the original data, X min is the minimum value of the feature, X max It is the maximum value of the feature. The standardized data value is between 0 and 1. It is suitable for situations where the data is required to be within a specific range.

[0102] S3 includes:

[0103] S31, application of principal component analysis: principal component analysis is used to extract features from the preprocessed multi-source data, and low-dimensional features with important information are extracted from high-dimensional data. This step specifically includes the following sub-steps:

[0104] S311, constructing a covariance matrix: First, the preprocessed data is centered, that is, the mean of each feature is subtracted so that the mean of each feature is 0. Then, based on the standardized data, the covariance matrix of the data set is calculated, and the formula is as follows:

[0105]

[0106] Among them, X i is the eigenvector of the i-th sample, μ is the mean vector of all samples, n is the number of samples, Σ is the covariance matrix, and the covariance matrix describes the linear relationship between different features in the data;

[0107] S312, eigenvalue and eigenvector calculation: perform eigenvalue decomposition on the covariance matrix to obtain eigenvalues ​​and eigenvectors. The eigenvalue reflects the variance of each principal component in the data, and the eigenvector corresponds to the main direction of the data. The calculation process is:

[0108] Σv=λv;

[0109] Among them, Σ is the covariance matrix, v is the eigenvector, and λ is the eigenvalue. By calculating all eigenvalues ​​and eigenvectors, the variance proportion of each principal component in the data can be determined;

[0110] S313, select principal components: select the main eigenvectors according to the size of the eigenvalues. The information carried by the principal components with larger eigenvalues ​​is more important, and usually the first k principal components with the largest eigenvalues ​​are selected. The selected principal components will be used to construct a new feature space. This process ensures that only principal components with larger variances are retained, and features with smaller variances and less contribution to disease prediction are eliminated;

[0111] The first k principal components can be selected by calculating the cumulative variance contribution rate, the formula is as follows:

[0112]

[0113] Among them, λ i is the eigenvalue of the i-th principal component, n is the number of all principal components, and by selecting an appropriate cumulative variance ratio threshold (such as 90%), we can decide how many principal components to retain;

[0114] S314, feature conversion and data reconstruction: by mapping the original data to the feature space of the selected principal component, the dimension reduction of the data is achieved. The conversion process can be completed by matrix multiplication, and the specific formula is:

[0115] X′=X·V k ;

[0116] Among them, X is the original data after preprocessing, V kis a matrix containing the first k principal component eigenvectors, and X′ is the new feature data after transformation. This step effectively transforms high-dimensional data into low-dimensional data and ensures that the most important information in the data is retained;

[0117] S32, select key features: select key features related to chronic disease risk, including blood pressure, blood sugar, heart rate, diet and exercise. The specific selection method includes:

[0118] Variance-based screening: select principal components with higher variance. Features with high variance usually represent parts of the data with greater variation. These features are more likely to be associated with the risk of chronic diseases.

[0119] Domain knowledge-guided screening: Combine professional knowledge in the medical field to identify features related to chronic disease risks, such as physiological data such as blood pressure, blood sugar, heart rate, and lifestyle data (such as diet and exercise), and further confirm key features through feedback from medical experts.

[0120] Based on the extracted key features, an improved random forest algorithm was used to build a chronic disease risk prediction model, including:

[0121] S41, feature weighting: After feature selection, different features are weighted to improve the contribution of key features to the chronic disease risk prediction model. The weighting methods include:

[0122] Weighting based on feature importance: Using the feature importance evaluation results of the random forest algorithm, features that are highly correlated with chronic disease risk are given higher weights to ensure that these key features have a greater influence on the splitting of the decision tree during model training;

[0123] The weighting method is: the weight w of each feature i Calculate, where Importance score i Calculated by the random forest algorithm, it is calculated as:

[0124]

[0125] in, represents the information gain of feature i in the jth tree, and N is the total number of trees;

[0126] S42, improved tree generation strategy: adopt the tree generation strategy in random forest, and introduce the feature weighted decision tree construction method, so that each tree gives priority to the features with higher importance when generating, and the top 30% of the feature importance scores are defined as high. The specific method is as follows:

[0127] When constructing each tree, nodes are split using weighted feature subsets, with features with higher weights being prioritized for splitting, allowing the model to focus more on key features during training.

[0128] When training each tree, node splitting is determined based on weighted features, so that features with high importance can enter the decision path earlier;

[0129] S43, Ensemble strategy: In the ensemble learning of random forests, different tree structures are used to train the same feature set multiple times, and voting decisions are made in combination with weighted features;

[0130] Voting mechanism: For each tree, its prediction result for chronic disease risk (such as high risk, medium risk or low risk) is output. Through the majority voting mechanism, the final model prediction result is the mode of all tree prediction results, that is, the risk level;

[0131] S44, Model evaluation: Use cross-validation method to evaluate the chronic disease risk prediction model. The evaluation indicators include accuracy, recall rate and F1 value to ensure that the model can effectively balance precision and recall rate when predicting chronic disease risk to avoid misdiagnosis or missed diagnosis.

[0132] S5 includes:

[0133] S51, model prediction: Based on the collected multi-source data of users, the risk probability of users developing chronic diseases in the future is calculated through the trained chronic disease risk prediction model, including:

[0134] Input data (including multiple features) into the trained model, and the model will make inferences based on the extracted key features;

[0135] Based on the weights and interrelationships of each feature, the user’s risk score is calculated;

[0136] The risk score usually ranges from 0 to 1, indicating the probability of a user developing a chronic disease;

[0137] At the output layer, the scores are divided into three levels: high risk, medium risk, and low risk;

[0138] S52, output results: The chronic disease risk prediction model calculates the risk score based on the multi-source data input by the user, and outputs three levels of high risk, medium risk, and low risk, and generates corresponding risk assessment results for each level. The specific outputs include:

[0139] High risk: The risk score is greater than or equal to 0.7, indicating that the user is more likely to develop a chronic disease in the future;

[0140] Medium risk: The risk score is between 0.4 and 0.7, indicating that the user has a medium probability of developing a chronic disease in the future;

[0141] Low risk: A risk score of less than 0.4 indicates that the user is less likely to develop a chronic disease in the future.

[0142] The S5 also includes:

[0143] S53, report generation: Generate a personalized user risk report based on the risk prediction results. The report contains the following contents:

[0144] User basic information: including the user's basic health records, such as age, gender, previous health conditions, etc.

[0145] Risk level assessment: Based on the prediction results, clearly identify the user as belonging to the high-risk, medium-risk or low-risk category.

[0146] Analysis of various risk factors: Detailed analysis of key factors that affect the user's health risks. For example, certain physiological indicators such as high blood pressure or high blood sugar may be the main reason for the user's high risk, and these factors will be highlighted in the report;

[0147] Risk prediction data: The report includes detailed data on risk scores and corresponding score ranges to help users understand the basis for different risk levels;

[0148] S54, Report output: The risk report is output through the user's smart device or mobile application, and provides download, print or storage options. The report is stored in the user's account for future reference or further analysis.

[0149] S6 includes:

[0150] S61, assess intervention needs: determine the user's intervention needs based on the risk prediction results. Specifically, if the user is classified as a high-risk or medium-risk level, further intervention suggestions are generated. If the user is classified as a low-risk level, regular health monitoring and lifestyle maintenance are recommended.

[0151] S62, Intervention suggestion generation: Generate personalized intervention suggestions based on the user's risk level and health status. The intervention suggestions include:

[0152] Diet adjustment suggestions: Diet adjustment suggestions are generated based on the user's lifestyle data (such as eating habits, weight, etc.) and the correlation with chronic diseases. High-risk users may be advised to reduce salt and sugar intake, increase dietary fiber intake, etc., while medium- and low-risk users may be advised to have a balanced diet and increase fruit and vegetable intake.

[0153] Exercise plan recommendations: Generate a personalized exercise plan based on the user's health status, age, weight and exercise data. High-risk users may be recommended to do low-intensity aerobic exercise, such as walking and swimming, and avoid strenuous exercise; medium- and low-risk users may be recommended to do higher-intensity exercise, such as running and strength training.

[0154] Drug prevention recommendations: Based on the user's physiological data (such as blood sugar, blood pressure, blood lipids, etc.), if obvious abnormal values ​​are found, drug prevention is recommended, such as antihypertensive drugs, hypoglycemic drugs or lipid-lowering drugs, to help control the risk of disease.

[0155] Regular physical examination recommendations: Based on the risk prediction results, the frequency and items of regular physical examinations are recommended. For example, high-risk users may need to undergo a comprehensive physical examination every quarter, including cardiovascular examinations, diabetes screening, etc.; medium- and low-risk users may be recommended to undergo a routine physical examination once a year.

[0156] S63, Intervention suggestion feedback: The generated personalized intervention suggestions are fed back to the user through a mobile application, SMS or email. The user can view, record and implement the intervention suggestions and record the implementation status through the application.

[0157] The S6 also includes:

[0158] S64, Intervention Implementation Monitoring: Monitor the user's implementation of early intervention recommendations through smart devices, health management platforms or user manual input. The monitoring content includes the implementation of dietary adjustments, completion of exercise plans, use of preventive medications, and participation in regular physical examinations;

[0159] S65, Implementation feedback analysis: Analyze the user's compliance with the intervention recommendations based on the user's implementation. If the user does not follow the recommendations (such as unbalanced diet, insufficient exercise, or failure to take medication on time), assess their possible health risks and decide whether the intervention plan needs to be adjusted;

[0160] S66, Dynamically adjust intervention recommendations: Based on the results of the implementation analysis, dynamically adjust the intervention recommendations and generate an updated version of the personalized intervention plan, including:

[0161] Diet adjustment optimization: If users fail to follow dietary recommendations, they may be provided with a more detailed and easy-to-implement diet plan, or recipes and food recommendations through mobile applications;

[0162] Exercise plan optimization: If the user fails to complete the exercise goal on time, the exercise intensity or frequency will be adjusted, or reminders and incentives will be used to encourage the user to persist in exercise;

[0163] Medication adjustment: If the medication is not taken well, we may recommend increasing the dosage or changing the type of medication, and provide a more professional adjustment plan through the doctor's review opinion;

[0164] Physical examination reminder: If the user fails to have a physical examination regularly, remind him / her to do so and suggest setting up a physical examination reminder function;

[0165] S67, feedback on adjustment results: Feedback the updated personalized intervention plan to users to ensure that users can understand and implement the new health management plan in a timely manner. The feedback can be provided through multiple channels such as smart health platforms, text messages, and emails.

[0166] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0167] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. AI-based chronic disease risk prediction and early intervention method, characterized by: The following steps are involved: S1, data collection: collect multi-source data of users through various channels, including physiological data, living habit data, environmental data and health history data; The physiological data include blood pressure, blood sugar, heart rate and blood lipids; The lifestyle data includes eating habits, exercise volume and work and rest time; The environmental data includes air quality, temperature and humidity; The health history data includes the user's past medical history and family medical history; S2, data preprocessing: preprocess the collected multi-source data, including data cleaning and missing value filling, and perform data standardization; S3, feature selection: principal component analysis was used to extract features from the preprocessed multi-source data to extract key features related to chronic disease risk; S4, model construction: Based on the extracted key features, the improved random forest algorithm was used to build a chronic disease risk prediction model; S5, risk prediction: using the chronic disease risk prediction model, based on the user's multi-source data, predict the user's risk probability of chronic diseases in the future; S6, Early Intervention Recommendations: Based on the predicted risk probability, provide users with personalized intervention recommendations, including diet adjustments, exercise plans, drug prevention, and regular physical examinations.

2. The AI-based chronic disease risk prediction and early intervention method according to claim 1, characterized in that: The S1 includes: S11, physiological data collection: collecting the user's physiological data through a variety of devices and sensors; S12, life habit data collection: collect users’ life habit data through smart health management platform, mobile applications and user self-report; S13, environmental data collection: collecting environmental data of the user's environment through environmental monitoring equipment and sensors; S14, health history data collection: collect the user's health history data through the medical platform, user self-report or related health examination records.

3. The AI-based chronic disease risk prediction and early intervention method according to claim 2, characterized in that: The S2 includes: S21, data cleaning: cleaning the collected multi-source data; S22, missing value filling: fill in the missing parts in multi-source data; S23, data standardization: standardize the preprocessed data to eliminate the dimensional differences between different data sources.

4. The AI-based chronic disease risk prediction and early intervention method according to claim 3, characterized in that: The S3 includes: S31, application of principal component analysis: principal component analysis is used to extract features from preprocessed multi-source data; S32, Select key features: Select key features related to the risk of chronic diseases, including blood pressure, blood sugar, heart rate, diet and exercise.

5. The AI-based chronic disease risk prediction and early intervention method according to claim 4, characterized in that: The improved random forest algorithm is used to construct a chronic disease risk prediction model based on the extracted key features, which specifically includes: S41, feature weighting: After feature selection, different features are weighted to improve the contribution of key features to the chronic disease risk prediction model; S42, improved tree generation strategy: adopt the tree generation strategy in random forest and improve it by introducing feature-weighted decision tree construction method, so that each tree gives priority to features with higher importance when generating; S43, Ensemble strategy: In the ensemble learning of random forests, different tree structures are used to train the same feature set multiple times, and voting decisions are made in combination with weighted features; S44, Model evaluation: The chronic disease risk prediction model was evaluated using the cross-validation method, and the evaluation indicators included accuracy, recall rate, and F1 value.

6. The AI-based chronic disease risk prediction and early intervention method according to claim 5, characterized in that: The S5 includes: S51, model prediction: Based on the collected multi-source data of users, the risk probability of users developing chronic diseases in the future is calculated through the trained chronic disease risk prediction model; S52, output result: The chronic disease risk prediction model calculates the risk score based on the multi-source data input by the user, and outputs three levels: high risk, medium risk, and low risk, and generates corresponding risk assessment results for each level.

7. The AI-based chronic disease risk prediction and early intervention method according to claim 6, characterized in that: The S5 further includes: S53, report generation: generating a personalized user risk report based on the risk prediction results; S54, Report output: The risk report is output via the user’s smart device or mobile application, with options for downloading, printing or storing.

8. The AI-based chronic disease risk prediction and early intervention method according to claim 7, characterized in that: The S6 includes: S61, assess intervention needs: determine the user's intervention needs based on the risk prediction results; S62, intervention suggestion generation: generating personalized intervention suggestions based on the user's risk level and health status; S63, Intervention suggestion feedback: Feedback the generated personalized intervention suggestion to the user via a mobile application, SMS or email.

9. The AI-based chronic disease risk prediction and early intervention method according to claim 8, characterized in that: The S6 further includes: S64, intervention implementation monitoring: monitoring the user's implementation of early intervention recommendations through smart devices, health management platforms, or user manual input; S65, implementation feedback analysis: Analyze the degree of user compliance with intervention recommendations based on the user's implementation status; S66, Dynamically adjust intervention recommendations: Dynamically adjust intervention recommendations based on the results of implementation analysis to generate an updated version of the personalized intervention plan; S67, Feedback adjustment results: Feedback the updated personalized intervention plan to the user.

Images