Design And Implementation Of Double-high Disease Risk Prediction System Based On LightGBM Algorithm

Chronic diseases are formed by long-term accumulation,lack of exact etiology diagnosis,once the prevention and treatment is not enough,it will bring great economic burden to people,and even endanger life.In China,the number of patients with chronic diseases has reached hundreds of millions,seriously affecting the development of China's medical and health undertakings.Because of the incubation period of chronic diseases is long,the management of chronic diseases mainly starts from prevention,and the research of disease prediction has important significance in the aspect of disease prevention.With the development of the Internet and big data,electronic medical records are getting better.People have begun to use mathematical models in disease research.Quantitative analysis is used to study the characteristics and principles of disease incidence.Through the analysis of the research status at home and abroad,it is found that machine learning method can achieve better accuracy in the prediction of disease research.In order to further improve the medical level,predicting,preventing and diagnosing the disease,this paper applies LightGBM algorithm to the prediction of hypertension and hyperlipidemia(double-high diseases)by analyzing the common indicators of these two diseases.This method of predicting diseases through artificial intelligence can effectively improve the accuracy of disease diagnosis,predict the incidence of double-high diseases in advance,realize early warning and intervention of diseases,and thus effectively manage double-high diseases.The main research contents of this paper include the following aspects.The core part of the risk prediction system for double-high diseases is the algorithm module.This paper designs and implements the algorithm engine module according to the requirements of the medical system.First of all,combining the current research status of medical prediction using machine learning algorithms at home and abroad with the characteristics of electronic medical record data,various indicators in the obtained medical report data set are collected and cleaned.Next,double-high disease risk prediction models based on the four algorithms LightGBM,XGBoost,KNN and LinearRegressor respectively are constructed,and the parameters are continuously adjusted during the training process to make the error prediction smaller.Finally,the four models are evaluated in the paper.The results show that the loss degree is least and the training time is shortest of LightGBM model.This paper uses object-oriented method to analyze and design the double-high disease risk prediction system,applying Java language to complete the development of the main functional modules of the system.This system can be divided into four modules: personal information management,medical examination,index prediction and health guidance.The personal information management module manages users' personal information,provide partial characteristic data for index prediction,and provide personal information of medical examination personnel for doctors' diagnosis.The medical examination module manages patients' medical examination report data and provides data sources for the index prediction module.The index prediction module can call LightGBM model constructed in the paper to predict the five indicators of systolic blood pressure,diastolic blood pressure,serum triglyceride,serum high-density lipoprotein cholesterol and serum low-density lipoprotein cholesterol based on patients' medical examination data and return the prediction result to provide data reference for the health guidance module.In the health guidance module,medical personnel can view the prediction results of predicted patients and provide targeted measures for disease intervention.