A risk assessment method and system for human health status

Abstract

The present invention provides a risk assessment method and system for human health status, which relates to the technical field of health risk assessment, including obtaining text data information of all sugar substitute products, using natural language processing technology to identify key entities of the cleaned text data information; using time series analysis algorithm to monitor and identify the sugar substitute intake data of users; combining the behavioral pattern and change trend data of the sugar substitute intake of users, and analyzing through machine learning algorithm; using gradient boosting algorithm to train the model, and using association analysis method to analyze the association rules between sugar substitute intake and health risk indicators, and calculating the risk score of individual intake of sugar substitutes; applying knowledge graph construction algorithm to generate user behavior knowledge graph; combining risk score and user behavior knowledge graph to provide personalized health advice and alarm for users. The present invention identifies the sugar substitute intake pattern and trend of users, predicts health risks, and provides health advice and real-time alarms for users.

Description

Technical Field

[0001] The present invention relates to the technical field of health risk assessment, and in particular to a method and system for risk assessment of human health status. Background Art

[0002] With the improvement of health awareness and the change of lifestyle, people are paying more and more attention to the impact of sugar intake in food on health. In the current technical field, the health risk assessment of sugar substitute products mainly relies on traditional nutrition research and epidemiological surveys. Although these methods can provide certain risk assessments, they often have problems such as incomplete data collection, backward analysis methods, and inability to update in real time. In addition, existing assessment technologies often lack consideration of individual differences, making it difficult to provide personalized health advice.

[0003] To address these issues, researchers have tried to collect more data and apply statistical models to conduct risk assessments, but these methods still have limitations in processing large-scale data sets and real-time data analysis. Summary of the invention

[0004] The purpose of the present invention is to provide a method and system for risk assessment of human health status to improve the above problems. In order to achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0005] In a first aspect, the present application provides a method for risk assessment of human health conditions, comprising:

[0006] Obtain text data information of all sugar substitute products, clean it, use natural language processing technology to identify key entities, preprocess and extract features of the cleaned text data information, where feature extraction includes word frequency, phrases, semantic roles and dependencies, and extract key information about sugar substitute products through data analysis algorithms, and convert the key information into a structured format to form the first sugar substitute product data set, where the text data information includes chemical structure, sweetness, calories and safety research, where key information includes dosage, effect description and side effect report;

[0007] Based on the first sugar substitute product data set, a time series analysis algorithm is used to monitor and identify users' sugar substitute intake data, where the sugar substitute intake data includes sugar substitute intake patterns and trends, thereby obtaining users' sugar substitute intake behavior patterns and change trend data;

[0008] Combining the behavioral patterns and changing trend data of users' sugar substitute intake, as well as the acquired eating habits and physiological data of users, the health risk prediction results of users are obtained through machine learning algorithm analysis;

[0009] According to the health risk prediction results and sugar substitute intake data of users, data integration processing is performed to form a second sugar substitute product data set. The gradient boosting algorithm is used to train the model to predict the relationship between sugar substitute intake and health risks. The association analysis method is used to analyze the association rules between sugar substitute intake and health risk indicators, and finally the risk score of individual sugar substitute intake is calculated, where the risk score includes short-term and long-term health effects;

[0010] Data integration and processing are performed on the risk scores of individual sugar substitute intake to construct a data set that includes individual sugar substitute intake frequency, intake and preferred sugar substitute types. Based on the user's sugar substitute intake behavior pattern, a knowledge graph construction algorithm is applied to generate a user behavior knowledge graph, where the user behavior knowledge graph includes the relationship between user preferences, living habits and health status;

[0011] Combining risk scores and user behavior knowledge graphs, through analysis of machine learning models and combined with real-time user feedback, we provide users with personalized health recommendations and alerts.

[0012] Preferably, based on the first sugar substitute product data set, a time series analysis algorithm is used to monitor and identify the user's sugar substitute intake data, wherein the sugar substitute intake data includes sugar substitute intake patterns and trends, thereby obtaining the user's sugar substitute intake behavior patterns and change trend data, including:

[0013] Based on the first sugar substitute product data set, the autoregressive integrated moving average model is used to construct a time series, and the residuals are autoregressive and moving average calculated to capture the statistical characteristics of sugar substitute intake data, and the time series data model of user sugar substitute intake is obtained, which contains the user's sugar substitute intake records;

[0014] Based on the time series data model of users' sugar substitute intake, Fourier transform is applied to analyze the periodic components of the intake data to identify the periodic trend in the intake pattern, and the main periodic components of the intake pattern are determined by analyzing the peak values ​​in the frequency domain, thereby obtaining the user's sugar substitute intake behavior pattern over a period of time, where the periodic trend includes daily periodicity and weekly periodicity;

[0015] The behavioral pattern of sugar substitute intake is processed through a trend analysis algorithm, which includes smoothing the time series data using a local weighted scatter plot smoothing method to obtain the behavioral trend data of users' sugar substitute intake.

[0016] Preferably, the user's behavior pattern and change trend data of sugar substitute intake, as well as the obtained eating habits and physiological data of the user are analyzed by a machine learning algorithm to obtain the user's health risk prediction results, including:

[0017] Integrate the user's sugar substitute intake behavior patterns, change trend data, user's eating habits and physiological data to build a comprehensive feature data set;

[0018] Divide the comprehensive feature dataset into feature sets X and label sets y , select a machine learning model for training, and during the training process, use an optimization algorithm to adjust the model parameters θ , iteratively update the parameters θ Until the performance of the model on the training data reaches the predetermined number of iterations, the model training is completed and a model that can map features to health risks is obtained;

[0019] The new data features of the user are input into the health risk model to predict the health risk and obtain the health risk prediction result of the user.

[0020] Preferably, the gradient boosting algorithm is used to train the model to predict the relationship between sugar substitute intake and health risks, including:

[0021] The second-generation sugar product data set was divided into a training set and a test set. After feature selection, the most important features for predicting health risks in the training set were determined. The training set was used for model training, and the test set was used for evaluating model performance. The most important features included sugar substitute intake, intake frequency, user age, gender, and weight physiological indicators.

[0022] Select the gradient boosting machine as the prediction model and initialize the model parameters to obtain the first prediction model. The most important features are input into the first prediction model, and the decision tree is gradually constructed and the model parameters are optimized. In each step, the gradient boosting machine adds a new decision tree to correct the errors of the previous model.

[0023] During the training process, the test set is used to evaluate the first prediction model, and the parameters and learning rate of the first prediction model are adjusted according to the evaluation results. After multiple iterations of training and optimization, the final gradient boosting machine model is obtained. The adjustment process includes pruning decision trees and adjusting the number and depth of trees. The final gradient boosting machine model can predict the user's health risks based on their sugar substitute intake data and other physiological indicators.

[0024] Preferably, the association analysis method is used to analyze the association rules between sugar substitute intake and health risk indicators, and finally calculate the risk score of individual sugar substitute intake, which includes:

[0025] Based on the final gradient boosting machine model, after prior algorithm and association rule confidence processing, the support and confidence between sugar substitute intake characteristics and health risks were calculated;

[0026] According to the support and confidence between sugar intake characteristics and health risks, a comprehensive model is obtained through risk scoring model construction, wherein the comprehensive model is a model used to calculate the risk score;

[0027] The comprehensive model is weighted and fused to obtain an individual risk score for sugar substitute intake, in which the confidence level and model prediction value are combined and a weighted average is used to calculate the final risk score, in which the weight is determined based on the confidence level.

[0028] In a second aspect, the present application also provides a risk assessment system for human health conditions, including:

[0029] Extraction module: used to obtain text data information of all sugar substitute products, clean it, use natural language processing technology to identify key entities, preprocess and extract features of the cleaned text data information, where feature extraction includes word frequency, phrases, semantic roles and dependencies, and extract key information about sugar substitute products through data analysis algorithms, and convert the key information into a structured format to form the first sugar substitute product data set, where the text data information includes chemical structure, sweetness, calories and safety research, and the key information includes dosage, effect description and side effect report;

[0030] Monitoring and identification module: used to monitor and identify the user's sugar substitute intake data based on the first sugar substitute product data set using a time series analysis algorithm, where the sugar substitute intake data includes sugar substitute intake patterns and trends, thereby obtaining the user's sugar substitute intake behavior patterns and change trend data;

[0031] The first analysis module is used to combine the behavior pattern and change trend data of the user's sugar substitute intake, as well as the obtained eating habits and physiological data of the user, and obtain the user's health risk prediction results through machine learning algorithm analysis;

[0032] Prediction association module: used to integrate data based on the user's health risk prediction results and sugar substitute intake data to form a second sugar substitute product data set, use the gradient boosting algorithm to train the model to predict the relationship between sugar substitute intake and health risks, and use association analysis to analyze the association rules between sugar substitute intake and health risk indicators, and finally calculate the risk score of individual sugar substitute intake, where the risk score includes short-term and long-term health effects;

[0033] Generation module: used to integrate data for risk scores of individual sugar substitute intake, build a data set containing individual sugar substitute intake frequency, intake and preferred sugar substitute types, and apply knowledge graph construction algorithm based on the user's sugar substitute intake behavior pattern to generate a user behavior knowledge graph, where the user behavior knowledge graph includes the relationship between user preferences, living habits and health status;

[0034] The second analysis module is used to combine risk scores and user behavior knowledge graphs, analyze machine learning models, and combine real-time user feedback to provide users with personalized health recommendations and alerts.

[0035] In a third aspect, the present application also provides a risk assessment device for human health conditions, including:

[0036] Memory for storing computer programs;

[0037] A processor is used to implement the steps of the risk assessment method for human health status when executing the computer program.

[0038] In a fourth aspect, the present application further provides a readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the above-mentioned risk assessment method based on human health status are implemented.

[0039] The beneficial effects of the present invention are:

[0040] The present invention aims to provide a method for assessing the risk of human health status, with a special focus on the impact of the intake of sugar substitute products on health. By integrating advanced natural language processing technology, machine learning algorithms and knowledge graph construction technology, a comprehensive analysis of individual sugar substitute intake behavior is conducted; first, natural language processing technology is used to extract key information from the text data of sugar substitute products collected on the market, and a first sugar substitute product data set is constructed; then, through time series analysis and machine learning models, the user's sugar substitute intake patterns and trends are monitored and identified, and health risks are predicted. Combined with the user behavior knowledge graph and risk score, personalized health advice and real-time alerts can be provided to users, which not only improves the accuracy and personalization level of risk assessment, but also can adapt to the rapidly changing data environment and provide users with timely health guidance.

[0041] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or be understood by implementing the embodiments of the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0043] Figure 1 A schematic diagram of a process flow of a risk assessment method for a human health condition described in an embodiment of the present invention;

[0044] Figure 2 A schematic diagram of the structure of a risk assessment system for human health conditions described in an embodiment of the present invention;

[0045] Figure 3 Schematic diagram of the structure of a risk assessment device for human health conditions described in an embodiment of the present invention.

[0046] In the figure: 701, extraction module; 702, monitoring and identification module; 703, first analysis module; 704, prediction and association module; 705, generation module; 706, second analysis module; 800, risk assessment device for human health status; 801, processor; 802, memory; 803, multimedia component; 804, I / O interface; 805, communication component. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0048] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0049] Embodiment 1:

[0050] This embodiment provides a method for risk assessment of human health status.

[0051] See also Figure 1 , the figure shows that the method includes step S100, step S200, step S300, step S400, step S500 and step S600.

[0052] S100. Obtain text data information of all sugar substitute products and clean it. Use natural language processing technology to identify key entities, perform preprocessing and feature extraction on the cleaned text data information, where feature extraction includes word frequency, phrases, semantic roles and dependencies. Use data analysis algorithms to extract key information about sugar substitute products and convert the key information into a structured format to form the first sugar substitute product data set, where the text data information includes chemical structure, sweetness, calories and safety research, and the key information includes dosage, effect description and side effect report.

[0053] It is understandable that in this step, according to the obtained text data information of the sugar substitute product, after data cleaning, irrelevant characters, HTML tags and special symbols are removed to ensure the readability and analyzability of the text. Therefore, in this step, regular expressions are usually used to identify and eliminate unnecessary elements to improve the accuracy of subsequent processing, and then identify keyword entities in the text, such as sugar substitute names, brands, ingredients, etc. Through pattern matching and machine learning models, a standardized representation of entities is obtained, in which the example in this step is to use named entity recognition technology for key entity recognition, usually using conditional random fields or deep learning models, and the specific process is as follows:

[0054] The cleaned text sequence is input into the model, represented as , where each xi Represents a word or phrase in the text, which may refer to the name, ingredients, dosage, etc. of the sugar substitute product; extract the features of each word through feature engineering, including part of speech, root, context information, etc., to form a feature vector F(xi); use the trained named entity recognition model to calculate each word xi Corresponding entity label probability distribution ,in yi Represents possible entity tags, such as "sugar substitute name", "dosage", "side effects", etc. The model's prediction score score ( xi , yi ) can be calculated by the following formula:

[0055]

[0056] In the formula, W is the weight matrix of the model, b is the bias term, F ( xi ) is a word xi The feature vector of xi Belongs to the tag yi confidence level.

[0057] The scores output by the model are then converted into probabilities through the softmax function for label selection:

[0058]

[0059] In the formula, is the exponential fraction, the denominator is the sum of the exponential fractions of all possible tags, and this probability represents the probability of xi In this case, it belongs to the tag yi Then the optimal label sequence is selected by the Viterbi algorithm , ensuring that the transfer relationship between tags complies with the predefined constraints.

[0060] Therefore, through this process, key entities can be identified from the text descriptions of sugar substitute products and classified into predefined tags, providing a basis for subsequent structured data processing and analysis.

[0061] S200. Based on the first sugar substitute product data set, a time series analysis algorithm is used to monitor and identify the user's sugar substitute intake data, wherein the sugar substitute intake data includes sugar substitute intake patterns and trends, thereby obtaining the user's sugar substitute intake behavior patterns and change trend data.

[0062] It can be understood that step S200 includes S201, S202 and S203, wherein:

[0063] S201. Based on the first sugar substitute product data set, a time series is constructed using an autoregressive integrated moving average model, and the residual is autoregressive and moving average calculated to capture the statistical characteristics of the sugar substitute intake data, thereby obtaining a time series data model of the user's sugar substitute intake, which includes the user's sugar substitute intake record;

[0064] It can be understood that in this step, data from different sources and types are integrated together to construct the feature vector X ,in X These include frequency of sugar substitute intake, total intake, dietary habit indicators (such as calorie intake, macronutrient ratio), and physiological indicators (such as weight, blood pressure, and blood sugar levels).

[0065] S302 divides the comprehensive feature data set into feature sets X and label sets y , select a machine learning model for training, and during the training process, use an optimization algorithm to adjust the model parameters θ , iteratively update the parameters θ Until the performance of the model on the training data reaches the predetermined number of iterations, the model training is completed and a model that can map features to health risk is obtained. The model training process is:

[0066]

[0067] In the formula, θ are model parameters, is the loss function, y Labels for known health risks, is the best solution found among all possible parameter values, X is the dataset of input features;

[0068] It should be noted that during the training process, we use optimization algorithms (such as gradient descent) to adjust the model parameters. θ , to minimize the loss function, this process usually involves multiple iterations, each iteration will update the parameters θ , until the performance of the model on the training data no longer improves significantly or reaches the predetermined number of iterations. After the model training is completed, the validation set (if available) is used to evaluate the performance of the model, where the validation set is a set of data that was not used in the training process, and the generalization ability of the model on unknown data is evaluated. The accuracy, recall, precision and other related indicators of the model on the validation set are calculated to ensure the predictive performance of the model. A trained and optimized machine learning model is obtained, which can predict the health risks of users based on their sugar substitute intake behavior and other related data.

[0069] S303: Input the acquired new data features of the user into the health risk model to predict the health risk and obtain the health risk prediction result of the user. The calculation formula is as follows:

[0070]

[0071] In the formula, h For a trained machine learning model, is the health risk predicted by the model, X is the input feature, θ is the model parameter.

[0072] It should be noted that new data features obtained from users include: users may enter their personal information, including eating habits, lifestyle and health indicators, through applications, websites or direct interactions with medical professionals; wearable devices such as smart watches, fitness trackers or blood glucose monitors can automatically collect users' physiological data, such as heart rate, blood pressure, activity level, etc.; users' medical records may contain their health history, medication use records, diagnosis results and other information, which can be used as feature input; users' dietary preferences, lifestyle habits and self-reported health status are collected through online or paper questionnaires.

[0073] In practical applications, these data features need to be properly processed and analyzed to ensure that they can be effectively used by machine learning models, which often includes steps such as data cleaning, standardization, missing value processing, feature selection, and feature engineering. X It will be used for model inference to predict the user's health risks.

[0074] S400. Based on the user's health risk prediction results and sugar substitute intake data, data integration processing is performed to form a second sugar substitute product data set. The model is trained using the gradient boosting algorithm to predict the relationship between sugar substitute intake and health risk. The association analysis method is used to analyze the association rules between sugar substitute intake and health risk indicators, and finally the risk score of individual sugar substitute intake is calculated, where the risk score includes short-term and long-term health effects.

[0075] It can be understood that the step S400 includes S401, S402 and S403, wherein:

[0076] S401, dividing the second sugar substitute product data set into a training set and a test set, and determining the most important features in the training set for predicting health risks through feature selection, wherein the training set is used for model training, and the test set is used for evaluating model performance, wherein the most important features include sugar substitute intake, intake frequency, user's age, gender, and body weight physiological indicators;

[0077] It should be noted that in this step, the most important features for predicting health risks are determined through feature selection processing. Feature selection may involve statistical analysis, model-based feature selection techniques, and the knowledge of domain experts to identify the most predictive features. The most important features include sugar substitute intake, intake frequency, user's age, gender, and weight physiological indicators. These features are selected because they have a strong correlation with health risks and can effectively help the model capture key information that affects health. Through this process, an optimized feature set can be obtained, which will serve as the basis for training the gradient boosting machine model to predict individual health risks. Among them, it should be noted that in terms of feature selection processing, statistical tests, model-based feature selection techniques such as recursive feature elimination (RFE), tree-based feature selection, etc. can be applied to identify the most effective features for predicting health risks from the data set. Through feature selection, the complexity of the model can be reduced, the prediction accuracy can be improved, and the risk of overfitting can be reduced.

[0078] S402, select the gradient boosting machine as the prediction model, initialize the model parameters, obtain the first prediction model, input the most important features into the first prediction model, gradually build a decision tree and optimize the model parameters. In each step, the gradient boosting machine will add a new decision tree to correct the errors of the previous model. The calculation formula is as follows:

[0079]

[0080] In the formula, θ are model parameters, l is the loss function, yi For the i The true labels of samples, is the label predicted by the model, For the i The feature vector of the samples, is the best solution found among all possible parameter values, n is the total number of samples in the training set, i is an index variable, For the model i The predicted value of samples;

[0081] S403. During the training process, the first prediction model is evaluated using a test set, and the parameters and learning rate of the first prediction model are adjusted according to the evaluation results. After multiple iterations of training and optimization, a final gradient boosting machine model is obtained, wherein the adjustment process includes pruning decision trees, adjusting the number and depth of trees, and the final gradient boosting machine model can predict the user's health risks based on the sugar substitute intake data and other physiological indicators.

[0082] It should be noted that the test set is used as a verification tool for model performance, helping to identify the model's predictive ability on unseen data. The evaluation results will guide the adjustment of model parameters, including the optimization of learning rate, to improve the generalization ability and prediction accuracy of the model. In the process of model adjustment, pruning technology is used to simplify the decision tree, reduce the complexity of the model, and prevent overfitting. At the same time, by adjusting the number and depth of decision trees, the bias and variance of the model can be balanced to ensure that the model will not be overly sensitive to training data while maintaining sufficient predictive ability. After multiple iterations of training and optimization, the final gradient boosting machine model is obtained. This model can combine the user's sugar substitute intake data and other physiological indicators, such as age, gender, weight, etc., to predict individual health risks. The output of the final model is a quantitative health risk score, which reflects the level of health risks that users may face based on their current lifestyle habits. The iterative training and parameter adjustment of the model ensure that it can maintain stable predictive performance on different data sets, thus providing users with an effective health risk assessment tool.

[0083] It can be understood that step S400 also includes S404, S405 and S406, wherein:

[0084] S404, based on the final gradient boosting machine model, after prior algorithm and association rule confidence processing, the support and confidence between the sugar substitute intake characteristics and health risks are calculated;

[0085] It should be noted that according to the user's sugar substitute intake characteristics and health risk indicators, after Apriori algorithm processing, the support between sugar substitute intake and health risk is obtained. A ∪ B ) and Support( A Support( A ∪ B ) means that all records contain the sugar substitute intake feature A and health risk indicators B The frequency of Support( A ) indicates the sugar substitute intake characteristics A Frequency of occurrence alone.

[0086] S405, according to the support and confidence between the sugar intake characteristics and the health risks, a risk scoring model is constructed to obtain a comprehensive model, wherein the comprehensive model is a model for calculating the risk score;

[0087] According to the support of the sugar substitute intake characteristics and health risk indicators obtained in step S404, the confidence level Confidence ( A ⇒ B Confidence( A ⇒ B ) indicates that the sugar substitute intake characteristics A Under the conditions that occur, health risk indicators B The conditional probability of occurrence is calculated as:

[0088]

[0089] It can be understood that the gradient boosting machine model and the confidence obtained by the above steps are processed by the risk scoring model to obtain a comprehensive model. This model combines the prediction results of the gradient boosting machine model and Confidence( A ⇒ B ) are combined to build a model for calculating risk scores.

[0090] S406. Perform weighted fusion processing on the comprehensive model to obtain a risk score for the individual's intake of sugar substitutes, wherein the confidence level and the model prediction value are combined, and a final risk score is calculated using a weighted average, wherein the weight is determined based on the confidence level.

[0091] It should be noted that the risk score model constructed in step S405 is subjected to weighted fusion processing to obtain the risk score of individual sugar substitute intake. R In the weighted fusion process, the confidence A ⇒ B ) and the prediction results of the gradient boosting machine model Different weights are assigned to each risk factor, which reflect their importance in the risk score. R The calculation formula may be as follows:

[0092]

[0093] in, and are weight coefficients, which are determined based on the statistical significance of the model’s performance and confidence. R Score the risk, To predict the results, Confidence( A ⇒ B ) is the confidence level.

[0094] Through the above steps, in this step, the correlation between the sugar substitute intake characteristics and the health risk indicators, as well as the prediction results of the gradient boosting machine model can be comprehensively considered to calculate the risk score of individual sugar substitute intake, and this risk score provides users with a quantitative health risk assessment.

[0095] S500. Perform data integration processing on the risk score of individual sugar substitute intake, construct a data set including individual sugar substitute intake frequency, intake and preferred sugar substitute types, and apply a knowledge graph construction algorithm based on the user's sugar substitute intake behavior pattern to generate a user behavior knowledge graph, wherein the user behavior knowledge graph includes the relationship between user preferences, living habits and health status.

[0096] It is understandable that in this step, data integration is first performed. This step involves summarizing and organizing data such as the frequency, amount, and preferred type of sugar substitutes consumed by individuals. These data may come from self-reports by users, health tracking applications, medical records, or other monitoring devices. Subsequently, through data cleaning and data conversion, the data set is integrated to obtain a comprehensive data set, which provides a basis for the subsequent construction of the knowledge graph. Based on the integrated data set D, apply the knowledge graph construction algorithm to generate the user behavior knowledge graph. This step involves converting the data into a graph structure, where nodes represent entities (such as users, sugar substitute types) and edges represent relationships between entities (such as preferences, intake frequency); then identify entities and relationships from the data, and use a graph database or graph processing framework (such as Neo4j, Apache Spark GraphX) to build a graph, where the user sugar substitute intake health risk assessment knowledge graph can intuitively display the relationship between user preferences, living habits and health status, providing rich contextual information for personalized health recommendations. And using the constructed user sugar substitute intake health risk assessment knowledge graph, more in-depth data analysis and pattern recognition can be performed, for example, through graph algorithms (such as shortest path, cluster analysis) to discover users' intake patterns and potential health risks, and run graph algorithms on the user sugar substitute intake health risk assessment knowledge graph to identify patterns and trends, and input the graph structure data into the machine learning model for prediction and classification, where the calculation formula for pattern recognition is as follows:

[0097]

[0098] Where P is the pattern or trend identified from the knowledge graph of health risk assessment of sugar substitute intake by users, User code Knowledge map of health risk assessment of sugar intake It is a graph structure composed of nodes (entities) and edges (relationships) to represent information such as user preferences, living habits and health status. Algorithm is a function or process. The knowledge graph of health risk assessment of sugar substitute intake is described as a graph structure composed of nodes (entities) and edges (relationships) to represent information such as user preferences, living habits and health status. The knowledge graph of health risk assessment of sugar substitute intake can be defined as a four-tuple:

[0099] Knowledge graph of health risk assessment of sugar substitute intake by users =( V , E , S , T )

[0100] in, V It is a collection of nodes, representing entities in the graph, including users, sugar substitutes, and health events; E is a set of edges, representing the relationships between entities, including "preference", "intake" and "influence"; S It is a node attribute set, representing the attributes related to the node, including the user's age, gender, and weight; T is a set of edge types, representing different types of relationships, including "frequent intake", "recommended intake", and "avoid intake". In this definition, the node v ∈ VCan be associated with one or more attributes, which can be numeric or categorical. e ∈ E Connects two nodes and can be directed (e.g., a preference from a user to a sugar substitute product) or undirected (e.g., the similarity between two sugar substitute products).

[0101] Graph Algorithm can be a collection of multiple algorithms used to perform specific analysis tasks on the knowledge graph of user sugar substitute intake health risk assessment. For example: Shortest path algorithm: find the shortest path between two nodes to identify the behavior chain of users' intake of sugar substitute products; Cluster analysis: group users based on the similarity of their living habits or health status; Community detection: identify community structures in the knowledge graph, which may represent groups of users with similar health risks.

[0102] Pattern Recognition P It can be calculated by applying graph algorithms, for example: P =Algorithm( G ,parameters)

[0103] In the formula, parameters include algorithm-specific parameters, including clustering thresholds and density requirements for community detection. In this way, meaningful patterns or trends can be identified from the knowledge graph of health risk assessment of sugar substitute intake by users. P .

[0104] For example, in some embodiments, if a community detection algorithm is used to analyze the relationship between users in a social network, Algorithm represents the community detection algorithm, and P The community structures detected may help understand the group behavior or preferences of users. In the context of health risk assessment, Algorithm may represent an algorithm for identifying health risk patterns, while P It could be a risk assessment based on user behavior and health data.

[0105] S600 combines risk scoring and user behavior knowledge graph, analyzes through machine learning models, and combines real-time user feedback to provide users with personalized health advice and alerts.

[0106] It can be understood that in this step, the risk score is a quantitative indicator, which represents the health risk level of the user's sugar substitute intake. The user behavior knowledge graph is a structured graph in which nodes represent users or sugar substitute products, and edges represent the relationship between users and sugar substitute products, such as preferences, frequencies, etc. In this embodiment, the combined data is analyzed using a machine learning model to generate personalized health recommendations and alerts. This usually involves the following processes: Feature extraction: extracting features from risk scores and knowledge graphs, such as sugar substitute intake frequency, preferred sugar substitute types, etc.; Model training: using the extracted features to train machine learning models, such as random forests, gradient boosting machines, etc.; Prediction and analysis: the model predicts the user's health status based on the training results and analyzes possible health risks.

[0107] Among them, personalized health recommendations are provided based on the user's current health risks and behavior patterns. These recommendations may include: adjusting sugar substitute intake, changing eating habits, increasing the intake of specific nutrients, and recommending regular health checks. Alerts are warnings about potential health risks, prompting users that they may need to take action. For example, if the model predicts that the user has a higher risk of diabetes, the system may issue an alert to recommend that the user reduce sugar substitute intake and consult a doctor. Therefore, by combining risk scores and user behavior knowledge graphs and applying machine learning models, users can be provided with more accurate and personalized health recommendations and alerts, which not only helps users better manage their own health, but also detects potential health problems at an early stage, thereby taking preventive measures to improve the user's overall health level. This method can also be used to study the relationship between sugar substitute intake and health risks, and provide a scientific basis for public health policies and nutritional recommendations.

[0108] Embodiment 2:

[0109] like Figure 2 As shown, this embodiment provides a risk assessment system for human health conditions, see Figure 2 The system comprises:

[0110] Extraction module 701: used to obtain text data information of all sugar substitute products, clean it, use natural language processing technology to perform key entity recognition, preprocessing and feature extraction on the cleaned text data information, where feature extraction includes word frequency, phrases, semantic roles and dependencies, and extract key information about sugar substitute products through data analysis algorithms, and convert the key information into a structured format to form a first sugar substitute product data set, where the text data information includes chemical structure, sweetness, calories and safety research, and the key information includes dosage, effect description and side effect report;

[0111] Monitoring and identification module 702: used to monitor and identify the user's sugar substitute intake data based on the first sugar substitute product data set by using a time series analysis algorithm, wherein the sugar substitute intake data includes sugar substitute intake patterns and trends, thereby obtaining the user's sugar substitute intake behavior patterns and change trend data;

[0112] The first analysis module 703 is used to combine the behavior pattern and change trend data of the user's sugar substitute intake, as well as the obtained eating habits and physiological data of the user, and obtain the user's health risk prediction result through machine learning algorithm analysis;

[0113] Prediction association module 704: used to perform data integration processing based on the health risk prediction results and sugar substitute intake data of the user to form a second sugar substitute product data set, use the gradient boosting algorithm to train the model to predict the relationship between sugar substitute intake and health risk, and use the association analysis method to analyze the association rules between sugar substitute intake and health risk indicators, and finally calculate the risk score of individual sugar substitute intake, where the risk score includes short-term and long-term health effects;

[0114] Generation module 705: for performing data integration processing on the risk score of individual intake of sugar substitutes, constructing a data set including the frequency of individual sugar substitute intake, intake and preferred sugar substitute types, and applying a knowledge graph construction algorithm based on the user's sugar substitute intake behavior pattern to generate a user behavior knowledge graph, wherein the user behavior knowledge graph includes the relationship between user preferences, living habits and health status;

[0115] The second analysis module 706 is used to combine the risk score and the user behavior knowledge graph, analyze the machine learning model, and combine the user's real-time feedback to provide personalized health advice and alerts to the user.

[0116] Specifically, the monitoring and identification module 702 includes:

[0117] The first calculation unit is used to construct a time series using an autoregressive integrated moving average model based on the first sugar substitute product data set, perform autoregressive and moving average calculations on the residuals, thereby capturing the statistical characteristics of the sugar substitute intake data, and obtaining a time series data model of the user's sugar substitute intake, which includes the user's sugar substitute intake records;

[0118] Determination unit: used to analyze the periodic components of the intake data based on the time series data model of the user's sugar substitute intake by applying Fourier transform to identify the periodic trend in the intake pattern, and determine the main periodic components of the intake pattern by analyzing the peak value in the frequency domain, so as to obtain the user's sugar substitute intake behavior pattern in a period of time, wherein the periodic trend includes daily periodicity and weekly periodicity, and the calculation formula of its Fourier transform is as follows:

[0119]

[0120] In the formula, Y ( k ) is the first k Quantity, yt is the original data point in the time domain, N is the total number of data points, k is the frequency index, t is the time point, It is the complex exponential term in Fourier transform, which is used to convert the time domain signal to the frequency domain;

[0121] The first processing unit is used to process the behavior pattern of sugar substitute intake through a trend analysis algorithm, including smoothing the time series data using a local weighted scatter plot smoothing method to obtain the user's sugar substitute intake behavior trend data.

[0122] Specifically, the first analysis module 703 includes:

[0123] Integration unit: used to integrate the behavior patterns, change trend data, dietary habits and physiological data of users’ sugar substitute intake to construct a comprehensive feature data set;

[0124] Training unit: used to divide the comprehensive feature dataset into feature sets X and label sets y , select a machine learning model for training, and during the training process, use an optimization algorithm to adjust the model parameters θ , iteratively update the parameters θ Until the performance of the model on the training data reaches the predetermined number of iterations, the model training is completed and a model that can map features to health risk is obtained. The model training process is:

[0125]

[0126] In the formula, θ are model parameters, is the loss function, y Labels for known health risks, is the best solution found among all possible parameter values, X is the dataset of input features;

[0127] Prediction unit: used to input the new data features of the user into the health risk model to predict the health risk and obtain the health risk prediction result of the user. The calculation formula is as follows:

[0128]

[0129] In the formula, h For a trained machine learning model, is the health risk predicted by the model, X is the input feature, θ is the model parameter.

[0130] Specifically, the prediction association module 704 includes:

[0131] The second processing unit is used to divide the second sugar substitute product data set into a training set and a test set, and determine the most important features in the training set for predicting health risks through feature selection processing. The training set is used for model training, and the test set is used for evaluating model performance. The most important features include sugar substitute intake, intake frequency, user's age, gender, and weight physiological indicators;

[0132] Input unit: used to select the gradient boosting machine as the prediction model, initialize the model parameters, obtain the first prediction model, input the most important features into the first prediction model, gradually build a decision tree and optimize the model parameters. In each step, the gradient boosting machine adds a new decision tree to correct the errors of the previous model. The calculation formula is as follows:

[0133]

[0134] In the formula, θ are model parameters, l is the loss function, yi For the i The true labels of samples, is the label predicted by the model, For the i The feature vector of the samples, is the best solution found among all possible parameter values, n is the total number of samples in the training set, i is an index variable, For the model i The predicted value of samples;

[0135] Evaluation and optimization unit: used to evaluate the first prediction model using the test set during the training process, and adjust the parameters and learning rate of the first prediction model according to the evaluation results. After multiple iterations of training and optimization, the final gradient boosting machine model is obtained. The adjustment process includes pruning decision trees and adjusting the number and depth of trees. The final gradient boosting machine model can predict the user's health risks based on the sugar substitute intake data and other physiological indicators.

[0136] Specifically, the prediction association module 704 includes:

[0137] The third processing unit is used to calculate the support and confidence between the sugar substitute intake characteristics and health risks based on the final gradient boosting machine model after prior algorithm and association rule confidence processing;

[0138] The fourth processing unit is used to obtain a comprehensive model based on the support and confidence between the sugar intake characteristics and the health risks through risk scoring model construction processing, wherein the comprehensive model is a model for calculating the risk score;

[0139] The fifth processing unit is used to perform weighted fusion processing on the comprehensive model to obtain the risk score of individual sugar substitute intake, wherein the confidence level and the model prediction value are combined, and the final risk score is calculated using weighted average, wherein the weight is determined based on the confidence level.

[0140] It should be noted that, regarding the system in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0141] Embodiment 3:

[0142] Corresponding to the above method embodiment, this embodiment also provides a risk assessment device for human health status. The risk assessment device for human health status described below and the risk assessment method for human health status described above can refer to each other.

[0143] Figure 3 FIG. 8 is a block diagram of a risk assessment device 800 for a human health condition according to an exemplary embodiment. Figure 3 As shown, the risk assessment device 800 for human health condition includes: a processor 801 and a memory 802. The risk assessment device 800 for human health condition also includes one or more of a multimedia component 803, an I / O interface 804, and a communication component 805.

[0144] The processor 801 is used to control the overall operation of the risk assessment device 800 for human health status to complete all or part of the steps in the above-mentioned risk assessment method for human health status. The memory 802 is used to store various types of data to support the operation of the risk assessment device 800 for human health status, and these data may include, for example, instructions for any application or method used to operate on the risk assessment device 800 for human health status, and application-related data, such as contact data, sent and received messages, pictures, audio, video, etc. The memory 802 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, disk or optical disk. The multimedia component 803 may include a screen and an audio component. The screen may be, for example, a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone, which is used to receive external audio signals. The received audio signal may be further stored in the memory 802 or sent via the communication component 805. The audio component also includes at least one speaker for outputting audio signals. The I / O interface 804 provides an interface between the processor 801 and other interface modules, and the above-mentioned other interface modules can be a keyboard, a mouse or a button, etc. These buttons can be virtual buttons or physical buttons. The communication component 805 is used for wired or wireless communication between the risk assessment device 800 for human health status and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G or 4G, or a combination of one or more of them, so the corresponding communication component 805 can include: a Wi-Fi module, a Bluetooth module or an NFC module.

[0145] In an exemplary embodiment, the risk assessment device 800 for human health status can be implemented by one or more application specific integrated circuits (ASIC), digital signal processors (DSP), digital signal processing devices (DSPD), programmable logic devices (PLD), field programmable gate arrays (FPGA), controllers, microcontrollers, microprocessors or other electronic components to execute the above-mentioned risk assessment method for human health status.

[0146] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, and when the program instructions are executed by a processor, the steps of the above-mentioned human health risk assessment method are implemented. For example, the computer-readable storage medium can be the above-mentioned memory 802 including program instructions, and the above-mentioned program instructions can be executed by the processor 801 of the human health risk assessment device 800 to complete the above-mentioned human health risk assessment method.

[0147] Embodiment 4:

[0148] Corresponding to the above method embodiment, a readable storage medium is also provided in this embodiment. The readable storage medium described below and the risk assessment method for human health status described above can refer to each other.

[0149] A computer program is stored on the readable storage medium, and when the computer program is executed by the processor, the steps of the risk assessment method for human health status of the above method embodiment are implemented.

[0150] The readable storage medium may specifically be a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, or other readable storage medium that can store program codes.

[0151] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0152] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for risk assessment of human health status, characterized in that: include: Step S100, obtaining text data information of all sugar substitute products, and cleaning them, using natural language processing technology to perform key entity recognition, preprocessing and feature extraction on the cleaned text data information, wherein feature extraction includes word frequency, phrases, semantic roles and dependencies, and extracting key information about sugar substitute products through data analysis algorithms, and converting the key information into a structured format to form a first sugar substitute product data set, wherein the text data information includes chemical structure, sweetness, calories and safety research, wherein the key information includes dosage, effect description and side effect report; Step S200: Based on the first sugar substitute product data set, a time series analysis algorithm is used to monitor and identify the sugar substitute intake data of the user, wherein the sugar substitute intake data includes a sugar substitute intake pattern and trend, thereby obtaining the user's sugar substitute intake behavior pattern and change trend data; Step S200 includes: Based on the first sugar substitute product data set, the autoregressive integrated moving average model is used to construct a time series, and the residuals are autoregressive and moving average calculated to capture the statistical characteristics of sugar substitute intake data, and the time series data model of user sugar substitute intake is obtained, which contains the user's sugar substitute intake records; Based on the time series data model of users' sugar substitute intake, Fourier transform is applied to analyze the periodic components of the intake data to identify the periodic trend in the intake pattern, and the main periodic components of the intake pattern are determined by analyzing the peak value in the frequency domain, so as to obtain the user's sugar substitute intake behavior pattern over a period of time, where the periodic trend includes daily periodicity and weekly periodicity. The calculation formula of Fourier transform is as follows: In the formula, Y ( k ) is the first k Quantity, yt is the original data point in the time domain, N is the total number of data points, k is the frequency index, t is the time point, It is the complex exponential term in Fourier transform, which is used to convert the time domain signal to the frequency domain; The behavior pattern of sugar substitute intake is processed by trend analysis algorithm, including smoothing the time series data by local weighted scatter plot smoothing method to obtain the behavior trend data of sugar substitute intake by users; Step S300: combining the behavior pattern and change trend data of the user's sugar substitute intake, as well as the obtained eating habits and physiological data of the user, and analyzing them through a machine learning algorithm to obtain a health risk prediction result for the user; Step S300 includes: Integrate the user's sugar substitute intake behavior patterns, change trend data, user's eating habits and physiological data to build a comprehensive feature data set; Divide the comprehensive feature dataset into feature sets X and label sets y , select a machine learning model for training, and during the training process, use an optimization algorithm to adjust the model parameters θ , iteratively update the parameters θ Until the performance of the model on the training data reaches the predetermined number of iterations, the model training is completed and a model that can map features to health risk is obtained. The model training process is: In the formula, θ are model parameters, is the loss function, y Labels for known health risks, is the best solution found among all possible parameter values, X is the dataset of input features; The new data features of the user are input into the health risk model to predict the health risk and obtain the health risk prediction result of the user. The calculation formula is as follows: In the formula, h For a trained machine learning model, is the health risk predicted by the model, X is the input feature, θ are model parameters; Step S400: performing data integration processing based on the health risk prediction results and sugar substitute intake data of the user to form a second sugar substitute product data set, using the gradient boosting algorithm to train the model to predict the relationship between sugar substitute intake and health risk, and using the association analysis method to analyze the association rules between sugar substitute intake and health risk indicators, and finally calculating the risk score of individual sugar substitute intake, wherein the risk score includes short-term and long-term health effects; Step S500, performing data integration processing on the risk score of individual intake of sugar substitutes, constructing a data set including the frequency of individual sugar substitute intake, intake and preferred sugar substitute types, and applying a knowledge graph construction algorithm based on the user's sugar substitute intake behavior pattern to generate a user behavior knowledge graph, wherein the user behavior knowledge graph includes the relationship between user preferences, living habits and health status; Step S600: Combine risk scores and user behavior knowledge graphs, analyze through machine learning models, and combine with real-time user feedback to provide users with personalized health advice and alerts.

2. The method for risk assessment of human health status according to claim 1, characterized in that: The model is trained using the gradient boosting algorithm to predict the relationship between sugar substitute intake and health risks, including: The second-generation sugar product data set was divided into a training set and a test set. After feature selection, the most important features for predicting health risks in the training set were determined. The training set was used for model training, and the test set was used for evaluating model performance. The most important features included sugar substitute intake, intake frequency, user age, gender, and weight physiological indicators. Select the gradient boosting machine as the prediction model, initialize the model parameters, obtain the first prediction model, and input the most important features into the first prediction model, gradually build a decision tree and optimize the model parameters. In each step, the gradient boosting machine will add a new decision tree to correct the errors of the previous model. The calculation formula is as follows: In the formula, θ are model parameters, l is the loss function, yi For the i The true labels of samples, is the label predicted by the model, For the i The feature vector of the samples, is the best solution found among all possible parameter values, n is the total number of samples in the training set, i is an index variable, For the model i The predicted value of samples; During the training process, the test set is used to evaluate the first prediction model, and the parameters and learning rate of the first prediction model are adjusted according to the evaluation results. After multiple iterations of training and optimization, the final gradient boosting machine model is obtained. The adjustment process includes pruning decision trees and adjusting the number and depth of trees. The final gradient boosting machine model can predict the user's health risks based on their sugar substitute intake data and other physiological indicators.

3. The method for risk assessment of human health status according to claim 2, characterized in that: The association analysis method is used to analyze the association rules between sugar substitute intake and health risk indicators, and finally calculate the risk score of individual sugar substitute intake, including: Based on the final gradient boosting machine model, after prior algorithm and association rule confidence processing, the support and confidence between sugar substitute intake characteristics and health risks were calculated; According to the support and confidence between sugar intake characteristics and health risks, a comprehensive model is obtained through risk scoring model construction, wherein the comprehensive model is a model used to calculate the risk score; The comprehensive model is weighted and fused to obtain an individual risk score for sugar substitute intake, in which the confidence level and model prediction value are combined and a weighted average is used to calculate the final risk score, in which the weight is determined based on the confidence level.

4. A human health risk assessment system, based on the human health risk assessment method according to claim 1, characterized in that: include: Extraction module: used to obtain text data information of all sugar substitute products, clean it, use natural language processing technology to identify key entities, preprocess and extract features of the cleaned text data information, where feature extraction includes word frequency, phrases, semantic roles and dependencies, and extract key information about sugar substitute products through data analysis algorithms, and convert the key information into a structured format to form the first sugar substitute product data set, where the text data information includes chemical structure, sweetness, calories and safety research, and the key information includes dosage, effect description and side effect report; Monitoring and identification module: used to monitor and identify the user's sugar substitute intake data based on the first sugar substitute product data set using a time series analysis algorithm, where the sugar substitute intake data includes sugar substitute intake patterns and trends, thereby obtaining the user's sugar substitute intake behavior patterns and change trend data; The first analysis module is used to combine the behavior pattern and change trend data of the user's sugar substitute intake, as well as the obtained eating habits and physiological data of the user, and obtain the user's health risk prediction results through machine learning algorithm analysis; Prediction association module: used to integrate data based on the user's health risk prediction results and sugar substitute intake data to form a second sugar substitute product data set, use the gradient boosting algorithm to train the model to predict the relationship between sugar substitute intake and health risks, and use association analysis to analyze the association rules between sugar substitute intake and health risk indicators, and finally calculate the risk score of individual sugar substitute intake, where the risk score includes short-term and long-term health effects; Generation module: used to integrate data for risk scores of individual sugar substitute intake, build a data set containing individual sugar substitute intake frequency, intake and preferred sugar substitute types, and apply knowledge graph construction algorithm based on the user's sugar substitute intake behavior pattern to generate a user behavior knowledge graph, where the user behavior knowledge graph includes the relationship between user preferences, living habits and health status; The second analysis module is used to combine the risk score and the user behavior knowledge graph, analyze the machine learning model, and combine the user's real-time feedback to provide users with personalized health advice and alerts; wherein the monitoring and identification module includes: The first calculation unit is used to construct a time series using an autoregressive integrated moving average model based on the first sugar substitute product data set, perform autoregressive and moving average calculations on the residuals, thereby capturing the statistical characteristics of the sugar substitute intake data, and obtaining a time series data model of the user's sugar substitute intake, which includes the user's sugar substitute intake records; Determination unit: used to analyze the periodic components of the intake data based on the time series data model of the user's sugar substitute intake by applying Fourier transform to identify the periodic trend in the intake pattern, and determine the main periodic components of the intake pattern by analyzing the peak value in the frequency domain, so as to obtain the user's sugar substitute intake behavior pattern in a period of time, wherein the periodic trend includes daily periodicity and weekly periodicity, and the calculation formula of its Fourier transform is as follows: In the formula, Y ( k ) is the first k Quantity, yt is the original data point in the time domain, N is the total number of data points, k is the frequency index, t is the time point, It is the complex exponential term in Fourier transform, which is used to convert the time domain signal to the frequency domain; The first processing unit is used to process the behavior pattern of sugar substitute intake through a trend analysis algorithm, including smoothing the time series data using a local weighted scatter plot smoothing method to obtain the user's sugar substitute intake behavior trend data.

5. The human health risk assessment system according to claim 4, characterized in that: The first analysis module includes: Integration unit: used to integrate the behavior patterns, change trend data, dietary habits and physiological data of users’ sugar substitute intake to construct a comprehensive feature data set; Training unit: used to divide the comprehensive feature dataset into feature sets X and label sets y , select a machine learning model for training, and during the training process, use an optimization algorithm to adjust the model parameters θ , iteratively update the parameters θ Until the performance of the model on the training data reaches the predetermined number of iterations, the model training is completed and a model that can map features to health risk is obtained. The model training process is: In the formula, θ are model parameters, is the loss function, y Labels for known health risks, is the best solution found among all possible parameter values, X is the dataset of input features; Prediction unit: used to input the new data features of the user into the health risk model to predict the health risk and obtain the health risk prediction result of the user. The calculation formula is as follows: In the formula, h For a trained machine learning model, is the health risk predicted by the model, X is the input feature, θ is the model parameter.

6. The human health risk assessment system according to claim 4, characterized in that: The prediction association module includes: The second processing unit is used to divide the second sugar substitute product data set into a training set and a test set, and determine the most important features in the training set for predicting health risks through feature selection processing. The training set is used for model training, and the test set is used for evaluating model performance. The most important features include sugar substitute intake, intake frequency, user's age, gender, and weight physiological indicators; Input unit: used to select the gradient boosting machine as the prediction model, initialize the model parameters, obtain the first prediction model, input the most important features into the first prediction model, gradually build a decision tree and optimize the model parameters. In each step, the gradient boosting machine adds a new decision tree to correct the errors of the previous model. The calculation formula is as follows: In the formula, θ are model parameters, l is the loss function, yi For the i The true labels of samples, is the label predicted by the model, For the i The feature vector of the samples, is the best solution found among all possible parameter values, n is the total number of samples in the training set, i is an index variable, For the model i The predicted value of samples; Evaluation and optimization unit: used to evaluate the first prediction model using the test set during the training process, and adjust the parameters and learning rate of the first prediction model according to the evaluation results. After multiple iterations of training and optimization, the final gradient boosting machine model is obtained. The adjustment process includes pruning decision trees and adjusting the number and depth of trees. The final gradient boosting machine model can predict the user's health risks based on the sugar substitute intake data and other physiological indicators.

7. The human health risk assessment system according to claim 5, characterized in that: The prediction association module includes: The third processing unit is used to calculate the support and confidence between the sugar substitute intake characteristics and health risks based on the final gradient boosting machine model after prior algorithm and association rule confidence processing; The fourth processing unit is used to obtain a comprehensive model based on the support and confidence between the sugar intake characteristics and the health risks through risk scoring model construction processing, wherein the comprehensive model is a model for calculating the risk score; The fifth processing unit is used to perform weighted fusion processing on the comprehensive model to obtain the risk score of individual sugar substitute intake, wherein the confidence level and the model prediction value are combined, and the final risk score is calculated using weighted average, wherein the weight is determined based on the confidence level.

Images