Automatic data processing method based on artificial intelligence

Through the automated data processing method of artificial intelligence, API interface, sparse coding, deep learning, fuzzy logic and causal graph model are used to solve the efficiency and accuracy of health data acquisition and processing in traditional methods, and the accurate identification of health risks is achieved. The provision of personalized health management suggestions has improved the effectiveness of health management.

CN119008000AInactive Publication Date: 2024-11-22MEDISHARE

Patent Information

Application Number
CN202411062091.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-05
Publication Date
2024-11-22
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Traditional manual or semi-automatic methods make it difficult to efficiently collect and process large-scale multi-source health data, resulting in poor real-time and accuracy of data collection, and the inability to fully explore deep-level patterns and relationships, resulting in inaccurate identification of health risks and lack of targeted health management. sex.

Method used

Adopting an automated data processing method based on artificial intelligence, collecting raw data through the API interface, using sparse encoding and deep learning algorithms for intelligent denoising, feature extraction and data augmentation, combining fuzzy logic and causal graph model, generate health assessment reports and personalization Health management advice and iterative optimization through feedback.

Benefits of technology

Improve the real-time and accuracy of data collection and processing, realize the accurate identification of health risks and chronic disease trends, provide personalized health management suggestions, improve the scientificity and effectiveness of health management, and continuously improve health management plan.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119008000A_ABST
    Figure CN119008000A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of data processing, in particular to an automatic data processing method based on artificial intelligence. Comprising the following steps: collecting original data such as health archives, chronic disease archives and contract signing information through an AP I interface; performing intelligent denoising, feature extraction and data enhancement on the original data by using a sparse coding algorithm; analyzing data by using a deep learning algorithm, and identifying health risks, chronic disease trends and follow-up visit requirements; a health assessment report and an intelligent follow-up visit plan are generated through a fuzzy logic algorithm, and accurate health management suggestions are provided; and displaying a decision result by using a causal graph model, collecting feedback, performing iterative optimization, and continuously improving a health management scheme. According to the invention, intelligent acquisition, processing, analysis and management of health data are realized, the system has the advantages of high efficiency, accuracy and intelligence, and the health management effect and user experience are significantly improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to an automatic data processing method based on artificial intelligence. Background Art

[0002] With the rapid development of information technology and artificial intelligence (AI) technology, the demand for data processing in the medical field is growing. Medical institutions generate a large amount of data every day, including patient medical records, diagnostic reports, imaging data, and treatment records. These data are not only large in volume, but also diverse in variety. Traditional manual processing methods are difficult to cope with, inefficient, and prone to errors. However, when facing large-scale, multi-source data, the existing system still has certain technical bottlenecks. The following problems still exist: traditional methods rely on manual or semi-automatic methods to collect health data, with limited coverage, poor real-time and accuracy of data collection, and it is difficult to meet the needs of large-scale health management; existing methods rely on simple statistics and rule algorithms in health risk identification, which cannot fully explore the deep patterns and relationships in the data, resulting in inaccurate health risk identification, insufficient personalization and precision; traditional health management programs often lack pertinence and cannot provide accurate health management suggestions based on individual specific circumstances, resulting in unsatisfactory health management results. Summary of the invention

[0003] To solve the above problems, the present invention provides an automated data processing method based on artificial intelligence, which solves the technical problem of how to achieve efficient collection, in-depth analysis and precise health management of large-scale, multi-source health data, thereby improving the real-time and accuracy of data processing and the health management effect.

[0004] To achieve the above object, the technical solution adopted by the present invention is:

[0005] An automated data processing method based on artificial intelligence comprises the following steps:

[0006] S1: Collecting raw data through the API interface, the raw data includes health records, chronic disease records and contract information;

[0007] S2: Using a sparse coding algorithm to intelligently denoise, extract features, and enhance data on the original data;

[0008] S3: Apply deep learning algorithms to analyze data to identify health risks, chronic disease trends, and follow-up needs;

[0009] S4: Optimize through fuzzy logic algorithm, generate health assessment reports and intelligent follow-up plans, and provide accurate health management suggestions;

[0010] S5: Use causal graph models to display decision results and collect feedback, perform iterative optimization, and continuously improve health management solutions.

[0011] Furthermore, the API interface includes a digital health record collection interface, a chronic disease file creation interface, a contract information interface and a data transmission interface.

[0012] Furthermore, the step S2 includes the following steps:

[0013] The original data is denoised by using a sparse coding algorithm to remove noise and outliers in the data;

[0014] Use Bayesian algorithm to extract features and extract health features;

[0015] The diffusion probability model is used for data enhancement, and diverse data samples are generated by simulating the diffusion and anti-diffusion processes in the data generation process.

[0016] Furthermore, the formula of the diffusion probability model is as follows:

[0017]

[0018] Among them, p(x t |x t-1 ) represents the data x at time step t given time step t-1 t-1 The generated data x t The probability distribution of x t represents the data sample at time step t, that is, the data of a time segment of the processed health records, chronic disease records and contract information; x t-1 represents the data sample at time step t-1, that is, the data of the previous time segment in the health records, chronic disease files and contract information before processing; α t It represents the degree of change of health records, chronic disease records and contract information data from time step t-1 to t; I represents the independence between the dimensions of each health record, chronic disease records and contract information data.

[0019] Furthermore, the step S3 includes the following steps:

[0020] Construct a hybrid deep learning model based on convolutional neural network and variational autoencoder to conduct comprehensive analysis of health records, chronic disease files and contract information;

[0021] Through model training, large-scale health data sets are used to train hybrid deep learning models and optimize model parameters;

[0022] Based on the trained hybrid deep learning model, health risk identification is performed on the input data;

[0023] Analyze health records and follow-up records, identify individual health trends and follow-up needs, and generate personalized health management plan recommendations.

[0024] Furthermore, the formula of the hybrid deep learning model is as follows:

[0025]

[0026] Where L represents the loss function of the hybrid deep learning model, which is used to measure the error of the model when reconstructing the input data; N represents the total number of samples of training data, that is, the number of samples of health records, chronic disease records, and contract information; x (i) represents the input data, the i-th health record, chronic disease file and contract information data sample; z (i) The latent variable corresponding to the i-th data sample is generated by the variational autoencoder model to capture the latent features of the input data; p(x (i) |z (i) ) represents a given latent variable z (i) Under the condition that the data sample x (i) The likelihood function of , generated by the convolutional neural network, is used to reconstruct the input data.

[0027] Furthermore, the fuzzy logic algorithm specifically defines fuzzy sets and rules, utilizes fuzzy reasoning mechanism to process input data, generates health assessment results, and converts them into specific health management recommendations and follow-up plans through defuzzification processing, while optimizing fuzzy rules and weight parameters according to the patient's personalized health status.

[0028] Furthermore, the step S5 includes the following steps:

[0029] Construct a causal graph model to display the health management decision-making process through node and edge representation, where the nodes represent key health-related variables and the edges represent the causal relationship between the variables;

[0030] Apply causal reasoning algorithms to generate health management decision recommendations and visualize the decision-making process;

[0031] Collect user feedback on health assessment reports and follow-up plans, and use the feedback information to optimize the causal graph model;

[0032] Update health assessment reports and follow-up plans in real time, and notify users through a smart push mechanism.

[0033] Furthermore, the formula of the causal graph model is as follows:

[0034]

[0035] Where P(Y|X) represents the conditional probability of the target variable Y given all health-related variables X; X represents the set of health-related variables, including health records, chronic disease records, and contract information; Y represents the target variable for health management decisions, including health assessment results or follow-up plans; Z k Represents potential intermediate or implicit variables; k represents the total number of intermediate variables; P(Y|X, Z k ) means that given the health-related variable X and the intermediate variable Z k The conditional probability of the target variable Y is P(Z k |X) means that given the health-related variable X, the intermediate variable Z k The conditional probability of .

[0036] The beneficial effects of the present invention are:

[0037] The present invention collects raw data through an API interface, including health records, chronic disease file creation and contract information, realizes the automation and efficiency of data collection, reduces manual intervention, and improves the real-time and accuracy of data. The sparse coding algorithm is used to perform intelligent denoising, feature extraction and data enhancement on the raw data, which improves the quality and availability of the data and provides a reliable data basis for subsequent analysis. By analyzing the data through a deep learning algorithm, it is possible to accurately identify health risks, chronic disease trends and follow-up needs, realize a comprehensive assessment and prediction of individual health conditions, and improve the scientificity and accuracy of health management. The fuzzy logic algorithm is used to optimize health assessment reports and intelligent follow-up plans, provide personalized health management recommendations, realize precise health management, and effectively improve the effect of health management. The causal graph model is used to display decision results, and feedback information is collected for iterative optimization, which realizes the continuous improvement of health management solutions and continuously improves the intelligence level of health management and user satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a flow chart of an automatic data processing method based on artificial intelligence of the present invention.

[0039] Figure 2 It is a flowchart of step S3 provided in an embodiment of the present invention.

[0040] Figure 3 It is a flowchart of step S5 provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0041] See also Figure 1-3 As shown, the present invention relates to an automatic data processing method based on artificial intelligence.

[0042] Example

[0043] An automated data processing method based on artificial intelligence comprises the following steps:

[0044] S1: Collect original data through the API interface, the original data includes health records, chronic disease file creation and contract information; the API interface includes a health record collection interface, a chronic disease file creation interface, a contract information interface and a data transmission interface.

[0045] It should be noted that the raw data is collected through multiple API interfaces, including the following:

[0046] The health record collection interface is used to obtain patients' health records from hospitals, community health service centers and other medical institutions, including past medical history, family medical history, drug allergy history, physical examination data, etc.

[0047] The chronic disease file creation interface is used to specifically collect patients' chronic disease file information, such as diagnosis and treatment records of chronic diseases such as diabetes, hypertension, and heart disease.

[0048] The contract information interface is used to obtain the contract information between patients and medical service providers, covering contract time, service content, follow-up plan, etc.

[0049] The data transmission interface is responsible for securely transmitting the collected data to the data processing platform to ensure data integrity and security.

[0050] S2: Using a sparse coding algorithm to intelligently denoise, extract features, and enhance data on the original data;

[0051] Wherein, the step S2 comprises the following steps:

[0052] The original data is denoised by using a sparse coding algorithm to remove noise and outliers in the data;

[0053] Specifically, a sparse representation model is constructed, and a sparse base dictionary D is obtained through dictionary learning, so that the original data X can be represented by a sparse coefficient α, that is, X≈Dα. The dictionary learning method can use the K-SVD algorithm to iteratively update the dictionary D and the sparse coefficient α by minimizing the reconstruction error.

[0054] Use Bayesian algorithm to extract features and extract health features;

[0055] Specifically, the Bayesian algorithm is used to extract features from the denoised data: calculate the conditional probability distribution of each health indicator. Extract key features from health records (such as weight, blood pressure, blood sugar, etc.). Extract important features from chronic disease records (such as chronic disease type, drug use, etc.). Extract features from contract information (such as contract time, service content, etc.).

[0056] The diffusion probability model is used for data enhancement, and diverse data samples are generated by simulating the diffusion and anti-diffusion processes in the data generation process.

[0057] Furthermore, the formula of the diffusion probability model is as follows:

[0058]

[0059] Among them, p(x t |x t-1 ) represents the data x at time step t given time step t-1 t-1 The generated data x t The probability distribution of x t represents the data sample at time step t, that is, the data of a time segment of the processed health records, chronic disease records and contract information; x t-1 represents the data sample at time step t-1, that is, the data of the previous time segment in the health records, chronic disease files and contract information before processing; α t It represents the degree of change of health records, chronic disease records and contract information data from time step t-1 to t; I represents the independence between the dimensions of each health record, chronic disease records and contract information data.

[0060] S3: Apply deep learning algorithms to analyze data to identify health risks, chronic disease trends, and follow-up needs;

[0061] Wherein, the step S3 comprises the following steps:

[0062] Construct a hybrid deep learning model based on convolutional neural network and variational autoencoder to conduct comprehensive analysis of health records, chronic disease files and contract information;

[0063] Specifically, the hybrid deep learning model combines convolutional neural networks (CNNs) and variational autoencoders (VAEs) to take advantage of the strengths of both. VAEs are used to learn the latent feature distribution of input data, while CNNs are used to process high-dimensional data and extract important features.

[0064] The construction process of the variational autoencoder is as follows:

[0065] The input data x (including health records, chronic disease records and contract information) first passes through the encoder network to generate the distribution parameters (mean and variance) of the latent variable z. The encoder network structure: multiple convolutional layers and pooling layers are used to extract high-level features of the data. The formula is as follows:

[0066]

[0067] Among them, μ(x) and σ 2(x) are the mean and variance of the latent variable z, respectively.

[0068] The construction process of the convolutional neural network is as follows:

[0069] The decoder network of VAE uses a CNN structure to reconstruct the input data x. Decoder network structure: deconvolution layer and upsampling layer, which are used to restore the latent variable z to the input data.

[0070] Through model training, large-scale health data sets are used to train hybrid deep learning models and optimize model parameters;

[0071] Specifically, collect and organize large-scale health data sets, including health records, chronic disease records, and contract information. Clean, standardize, and normalize the data so that it can be input into the model for training. Divide the data into training sets, validation sets, and test sets to ensure that the model can be effectively verified and evaluated on different data sets. Use the training set to train the hybrid deep learning model. Use optimization algorithms such as stochastic gradient descent (SGD) or Adam to iteratively update the model parameters and minimize the loss function. During the training process, use the validation set to monitor the performance of the model to avoid overfitting. Adjust hyperparameters (such as learning rate, batch size, etc.) to optimize model performance. Calculate gradients through the backpropagation algorithm and update model parameters. Use the features in the large-scale health data set to train the model to capture the potential patterns and features of the data. Regularly evaluate the performance of the model on the validation set, perform early stopping or model adjustment, and ensure the generalization ability of the model.

[0072] Based on the trained hybrid deep learning model, health risk identification is performed on the input data;

[0073] Specifically, new health records, chronic disease records, and contract information data are input into the trained hybrid deep learning model. Necessary data preprocessing is performed to make it meet the model input requirements. The trained hybrid deep learning model is used to predict the input data and identify individual health risks. The model outputs health risk probabilities or risk levels by analyzing the characteristics of the input data. Based on the comprehensive analysis capabilities of CNN and VAE, the model can accurately identify different health risk factors, including chronic disease risks, potential disease trends, etc.

[0074] Analyze health records and follow-up records, identify individual health trends and follow-up needs, and generate personalized health management plan recommendations.

[0075] Furthermore, the formula of the hybrid deep learning model is as follows:

[0076]

[0077] Where L represents the loss function of the hybrid deep learning model, which is used to measure the error of the model when reconstructing the input data; N represents the total number of samples of training data, that is, the number of samples of health records, chronic disease records, and contract information; x (i) represents the input data, the i-th health record, chronic disease file and contract information data sample; z (i) The latent variable corresponding to the i-th data sample is generated by the variational autoencoder model to capture the latent features of the input data; p(x (i) |z (i) ) represents a given latent variable z (i) Under the condition that the data sample x (i) The likelihood function of , generated by the convolutional neural network, is used to reconstruct the input data.

[0078] S4: Optimize through fuzzy logic algorithm, generate health assessment reports and intelligent follow-up plans, and provide accurate health management suggestions; the fuzzy logic algorithm specifically defines fuzzy sets and rules, uses fuzzy reasoning mechanism to process input data, generates health assessment results, and converts them into specific health management suggestions and follow-up plans through defuzzification processing, while optimizing fuzzy rules and weight parameters according to the patient's personalized health status.

[0079] It should be noted that the core of fuzzy logic algorithm lies in the definition of fuzzy sets and rules, which are designed according to health management needs: input variables include physiological indicators such as blood pressure, blood sugar, cholesterol, etc., and each indicator is defined as a fuzzy set. For example, blood pressure can be defined as three fuzzy sets: low, normal, and high. Output variables include health risk assessment results and follow-up recommendations, such as fuzzy sets such as healthy, sub-healthy, and need for follow-up. Rules are formulated based on expert knowledge and historical data. For example: If blood pressure is high and blood sugar is high, the health risk is high. If cholesterol is normal and weight is normal, the health risk is low. Each rule consists of an "if-then" statement that maps input variables to output variables.

[0080] Convert precise input data (such as specific blood pressure values) into corresponding fuzzy sets. For example, a patient's blood pressure is 150 / 95 mmHg, and this value is mapped to the fuzzy set "hypertension". Infer the fuzzified data according to the defined fuzzy rules. For example, the patient's health risk level is inferred through fuzzy rules based on their blood pressure and blood sugar conditions. Calculate the output of all rules and aggregate them according to the importance (weight) of the rules. For example, if multiple rules are applicable at the same time, the output and weight of each rule are considered comprehensively to derive the comprehensive health risk.

[0081] Defuzzify the fuzzy reasoning results and convert them into specific health management recommendations and follow-up plans: Common defuzzification methods include the maximum membership method, weighted average method, etc. The specific method to be used depends on the system design and requirements. For example, the health risk assessment result is a high risk of 0.7, which is interpreted as "immediate follow-up is required" through the maximum membership method. Based on the defuzzification results, specific health management recommendations and follow-up plans are generated. For example: For high-risk patients, it is recommended that they undergo detailed physical examinations and regular follow-up. For medium-risk patients, it is recommended that they pay attention to diet and exercise, and regularly monitor relevant indicators.

[0082] Dynamically adjust fuzzy rules and weight parameters based on the patient's personalized health status to optimize the accuracy and applicability of health management recommendations and follow-up plans: Adjust the conditions and outputs of fuzzy rules based on the patient's health history and follow-up feedback. For example, if a patient has a special reaction to a certain drug, adjust the relevant rules to reflect this. Use machine learning algorithms to optimize the weight parameters of fuzzy rules to ensure that the generated recommendations are more accurate and personalized. For example, by analyzing historical data and adjusting the weights of certain rules, the system makes follow-up recommendations for patients with hypertension more reasonable. Generate a health assessment report with personalized recommendations, which details the patient's health risks, management recommendations, and follow-up plans. Reports can be sent to patients and doctors via email or mobile applications to ensure timely communication and follow-up.

[0083] S5: Use causal graph models to display decision results and collect feedback, perform iterative optimization, and continuously improve health management solutions.

[0084] Wherein, the step S5 comprises the following steps:

[0085] Construct a causal graph model to display the health management decision-making process through node and edge representation, where the nodes represent key health-related variables and the edges represent the causal relationship between the variables;

[0086] It should be noted that the nodes represent key health-related variables, such as:

[0087] Variables in health records: weight, height, blood pressure (systolic, diastolic), blood sugar, cholesterol, BMI, heart rate, etc.

[0088] Variables in chronic disease file creation: chronic disease type (such as diabetes, hypertension, coronary heart disease, etc.), medication use, family medical history, disease course, etc.

[0089] Variables in contract information: contract time, service content, follow-up plan, doctor's advice, etc.

[0090] Edges represent causal relationships between variables, such as weight affects BMI, BMI affects blood pressure, and blood pressure affects cardiovascular disease risk.

[0091] Based on the knowledge of medical experts and historical data, an initial causal graph model is constructed.

[0092] Use structural learning algorithms (such as PC algorithm, greedy algorithm) to further optimize the structure of the causal graph. The specific steps are as follows:

[0093] Initialize graph structure: Initialize the nodes and edges of the causal graph based on known medical knowledge.

[0094] Data-driven optimization: Use historical data to optimize the graph structure through structural learning algorithms, adjust the direction and weight of edges, and make the model more consistent with the actual data distribution.

[0095] Apply causal reasoning algorithms to generate health management decision recommendations and visualize the decision-making process;

[0096] Specifically, the Bayesian network inference algorithm is used to infer the causal graph model and generate health management decision recommendations. The specific steps are as follows: Input health records, chronic disease files and contract information into the causal graph model. Based on the causal graph model and input data, calculate the conditional probability distribution of the target variable Y. Based on the inference results, generate health management decision recommendations, for example: If a patient's blood pressure and blood sugar are both high, it is recommended that they conduct more frequent blood pressure and blood sugar monitoring, and adjust their diet and exercise plans. If a patient's contract information shows that there has been no follow-up for a long time, it is recommended to reschedule the follow-up time.

[0097] Use charts, cause-and-effect diagrams, decision trees, etc. to visualize the health management decision-making process and results to users. Specific methods include: displaying the cause-and-effect diagram between health-related variables, highlighting key nodes and edges. Displaying the generation process of health management recommendations in the form of a decision tree to help users understand the basis for decision-making. Use bar charts, line charts, etc. to display the changing trends and current status of key health indicators.

[0098] Collect user feedback on health assessment reports and follow-up plans, and use the feedback information to optimize the causal graph model;

[0099] Specifically, user feedback on health assessment reports and follow-up plans is collected through the system interface, mobile applications or emails. The specific content includes: user evaluation of health management suggestions, such as whether the suggestions are feasible and helpful to health. User implementation of the follow-up plan, such as whether follow-up is carried out on time and problems encountered during the follow-up. User suggestions for improving the health management plan, such as which health indicators they would like to increase monitoring of and adjust the follow-up frequency.

[0100] Store the collected feedback data in a database to ensure data integrity and security. Analyze feedback data regularly to identify common problems and improvement opportunities. Specific methods include: statistical analysis of feedback data to calculate the frequency and type of feedback. Perform natural language processing (NLP) on user text feedback to extract key opinions and suggestions.

[0101] Based on user feedback, the structure and parameters of the causal graph model are optimized. The specific steps are as follows: According to the feedback information and new data, adjust the nodes and edges in the causal graph, add or delete certain causal relationships. Use machine learning algorithms (such as the EM algorithm) to re-estimate the conditional probability distribution in the model to ensure the accuracy and adaptability of the model.

[0102] Regularly iterate the model and continuously optimize it based on the latest feedback and data. Specific methods include: comprehensively updating and optimizing the causal graph model at regular intervals (such as monthly or quarterly). After collecting new feedback data, immediately fine-tune the model to ensure that the model always reflects the latest user needs and health status.

[0103] Update health assessment reports and follow-up plans in real time, and notify users through a smart push mechanism.

[0104] Specifically, according to the optimized causal graph model, the health assessment report and follow-up plan are updated in real time. Specific steps include: continuously monitoring the user's health data and follow-up records to ensure the real-time nature of the data. When the user's health data or feedback information changes, the system automatically updates the health assessment report and follow-up plan and generates new health management suggestions. Using push notification technology, the updated health assessment report and follow-up plan are pushed to users in a timely manner through mobile applications, emails and other channels. Specific methods include: formulating personalized push strategies based on user preferences and urgency. For example, important health warnings are pushed immediately, and routine health advice is pushed regularly. Ensure the accuracy and readability of the pushed content, including updated health assessment results, follow-up plans and health management suggestions. Provide a user-friendly interactive interface, where users can view detailed health assessment reports and follow-up plans by clicking on push notifications, and provide instant feedback.

[0105] Furthermore, the formula of the causal graph model is as follows:

[0106]

[0107] Where P(Y|X) represents the conditional probability of the target variable Y given all health-related variables X; X represents the set of health-related variables, including health records, chronic disease records, and contract information; Y represents the target variable for health management decisions, including health assessment results or follow-up plans; Z kRepresents potential intermediate or implicit variables; k represents the total number of intermediate variables; P(Y|X, Z k ) means that given the health-related variable X and the intermediate variable Z k The conditional probability of the target variable Y is P(Z k |X) means that given the health-related variable X, the intermediate variable Z k The conditional probability of .

[0108] In summary, the present invention collects raw data through multiple API interfaces, covering health records, chronic disease files and contract information. This multi-dimensional data collection method ensures the comprehensiveness and diversity of the data, making subsequent data processing and analysis more comprehensive and accurate. The sparse coding algorithm is used to perform intelligent denoising, feature extraction and data enhancement on the raw data to effectively remove noise and outliers in the data and retain important health features. By constructing a sparse representation model and using the Bayesian algorithm for feature extraction, the key features in health records, chronic disease files and contract information can be extracted to enhance the data analysis effect.

[0109] The present invention uses a diffusion probability model for data enhancement. By simulating the diffusion and reverse diffusion processes in the data generation process, it generates diverse data samples, enhances the diversity and robustness of the data, and helps to improve the generalization ability and prediction accuracy of the model. A hybrid deep learning model based on convolutional neural network (CNN) and variational autoencoder (VAE) is constructed to make full use of the advantages of both to conduct a comprehensive analysis of health records, chronic disease files and contract information. Through the training and optimization of the deep learning model, it can accurately identify individual health risks, chronic disease trends and follow-up needs, and generate personalized health management plan recommendations.

[0110] The present invention generates health assessment reports and intelligent follow-up plans through fuzzy logic algorithms, providing accurate health management suggestions. The algorithm uses fuzzy sets and rules to process input data, and converts it into specific health management suggestions and follow-up plans through defuzzification. At the same time, it dynamically optimizes fuzzy rules and weight parameters according to the patient's personalized health status to ensure the accuracy and applicability of the suggestions.

[0111] The present invention constructs a causal graph model, displays the health management decision-making process through the representation of nodes and edges, and intuitively displays the causal relationship between health-related variables. The causal reasoning algorithm is used to generate health management decision suggestions, and the decision-making process is visualized to help users understand the basis for the decision. By collecting user feedback information, optimizing the causal graph model, continuously improving the health management plan, and improving user satisfaction and health management effects. The causal graph model is used to display decision results and collect feedback, perform iterative optimization, and continuously improve the health management plan. By updating health assessment reports and follow-up plans in real time, and notifying users in a timely manner using an intelligent push mechanism, it is ensured that users can obtain the latest health management suggestions and follow-up plans in a timely manner, thereby improving the timeliness and effectiveness of health management.

[0112] The above implementation modes are merely descriptions of the preferred implementation modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary engineering and technical personnel in the field shall fall within the protection scope determined by the claims of the present invention.

Claims

1. An automated data processing method based on artificial intelligence, characterized in that: The following steps are involved: S1: Collecting raw data through the API interface, the raw data includes health records, chronic disease records and contract information; S2: Using a sparse coding algorithm to intelligently denoise, extract features, and enhance data on the original data; S3: Apply deep learning algorithms to analyze data to identify health risks, chronic disease trends, and follow-up needs; S4: Optimize through fuzzy logic algorithm, generate health assessment reports and intelligent follow-up plans, and provide accurate health management suggestions; S5: Use causal graph models to display decision results and collect feedback, perform iterative optimization, and continuously improve health management solutions.

2. The method of automatic data processing based on artificial intelligence according to claim 1, characterized in that: The API interface includes a digital health record collection interface, a chronic disease file creation interface, a contract information interface and a data transmission interface.

3. The method of automatic data processing based on artificial intelligence according to claim 1, characterized in that: The step S2 comprises the following steps: The original data is denoised by using a sparse coding algorithm to remove noise and outliers in the data; Use Bayesian algorithm to extract features and extract health features; The diffusion probability model is used for data enhancement, and diverse data samples are generated by simulating the diffusion and anti-diffusion processes in the data generation process.

4. The method of automatic data processing based on artificial intelligence according to claim 3, characterized in that: The formula of the diffusion probability model is as follows: Among them, p(x t |x t-1 ) represents the data x at time step t given time step t-1 t-1 The generated data x t The probability distribution of x t represents the data sample at time step t, that is, the data of a time segment of the processed health records, chronic disease records and contract information; x t-1 represents the data sample at time step t-1, that is, the data of the previous time segment in the health records, chronic disease files and contract information before processing; α t It represents the degree of change of health records, chronic disease records and contract information data from time step t-1 to t; I represents the independence between the dimensions of each health record, chronic disease records and contract information data.

5. The method of automatic data processing based on artificial intelligence according to claim 1, characterized in that: The step S3 comprises the following steps: Construct a hybrid deep learning model based on convolutional neural network and variational autoencoder to conduct comprehensive analysis of health records, chronic disease files and contract information; Through model training, large-scale health data sets are used to train hybrid deep learning models and optimize model parameters; Based on the trained hybrid deep learning model, health risk identification is performed on the input data; Analyze health records and follow-up records, identify individual health trends and follow-up needs, and generate personalized health management plan recommendations.

6. The method of automatic data processing based on artificial intelligence according to claim 5, characterized in that: The formula of the hybrid deep learning model is as follows: Where L represents the loss function of the hybrid deep learning model, which is used to measure the error of the model when reconstructing the input data; N represents the total number of samples of training data, that is, the number of samples of health records, chronic disease records, and contract information; x (i) represents the input data, the i-th health record, chronic disease file and contract information data sample; z (i) The latent variable corresponding to the i-th data sample is generated by the variational autoencoder model to capture the latent features of the input data; p(x (i) |z (i) ) represents a given latent variable z (i) Under the condition that the data sample x (i) The likelihood function of , generated by the convolutional neural network, is used to reconstruct the input data.

7. The method of automatic data processing based on artificial intelligence according to claim 1, characterized in that: The fuzzy logic algorithm specifically defines fuzzy sets and rules, uses fuzzy reasoning mechanism to process input data, generates health assessment results, and converts them into specific health management recommendations and follow-up plans through defuzzification processing, while optimizing fuzzy rules and weight parameters according to the patient's personalized health status.

8. The method of automatic data processing based on artificial intelligence according to claim 1, characterized in that: The step S5 comprises the following steps: Construct a causal graph model to display the health management decision-making process through node and edge representation, where the nodes represent key health-related variables and the edges represent the causal relationship between the variables; Apply causal reasoning algorithms to generate health management decision recommendations and visualize the decision-making process; Collect user feedback on health assessment reports and follow-up plans, and use the feedback information to optimize the causal graph model; Update health assessment reports and follow-up plans in real time, and notify users through a smart push mechanism.

9. The method of automatic data processing based on artificial intelligence according to claim 8, characterized in that: The formula of the causal graph model is as follows: Where P(Y|X) represents the conditional probability of the target variable Y given all health-related variables X; X represents the set of health-related variables, including health records, chronic disease records, and contract information; Y represents the target variable for health management decisions, including health assessment results or follow-up plans; Z k represents potential intermediate variables or implicit variables; k represents the total number of intermediate variables; P(Y|X, Z k ) means that given the health-related variable X and the intermediate variable Z k The conditional probability of the target variable Y under the condition of k |X) means that given the health-related variable X, the intermediate variable Z k The conditional probability of .

Citation Information

Patent Citations

  • Medical data mining method and system fused with deep learning

    CN117292840A

  • Disease prediction and risk assessment method based on large medical model

    CN118280570A

  • Patient data visualization method and system for assisting decision making in chronic diseases

    US20220157468A1

Cited By

  • Regional carbon emission accounting system

    CN119443534A

  • A regional carbon emissions accounting system

    CN119443534B

  • Multifunctional vital sign monitoring method and system

    CN119791621A

  • A multi-functional vital sign monitoring method and system

    CN119791621B

  • Health data automatic processing and precise health management method based on artificial intelligence

    CN121439195A