Research On Combination Forecasting Of Disease Models Based On Data Mining

With the development of informatization in the medical industry,more and more medical information data is generated,and the combination of the medical industry and big data has become an inevitable trend.Data mining technology is used to discover potential knowledge,prevent diseases,and have medical research and disease diagnosis and it is very important.Health is increasingly valued by people,and the improvement of living standards has caused more and more people to suffer from health diseases.Hypertension and hyperlipidemia are common chronic diseases that are prone to stroke,heart failure,kidney failure,cardiovascular and cerebrovascular diseases,etc.The occurrence of these diseases is related to a variety of factors.How to give the probability of occurrence of each disease and prevent it in advance is a further study.From the perspective of preventive medicine,this paper aims to build a combined disease prediction model,explore tacit knowledge in massive physical examination data,and predict disease risk based on tacit knowledge.The main work of this paper is as follows:(1)A prediction algorithm based on the combined model of Choquet integral is proposed,and a single model screening process is implemented based on the combined prediction algorithm.There are many prediction models,and each model will have different prediction results for the same data.Considering the different measurement angles of the model to the prediction results and the influence of the interaction between the models on the prediction results,the fuzzy integrals are used to measure the fuzzy measures of different models and the interaction between the models.Through experimental analysis,the predictions are greatly improved.In this paper,five models are combined,and the screening process of the model is established.The optimal combination object is determined and the accuracy of model prediction is improved.(2)A combined feature filtering algorithm based on Choquet integral is proposed.Model-based feature filtering is the degree of contribution of feature interpretation to prediction results from the perspective of data and model.The measurement angles of different models will be different.This paper constructs an algorithm based on feature selection of combined models to achieve the perspective of model and model interaction.The degree of contribution of the importance of the feature.At the same time,considering the interaction of features,the interaction between features and tags also analyzes the features.(3)This paper uses the physical examination data in the actual scene to predict the hypertension of a certain type of disease,and builds a prediction system based on model fusion through data integration,data cleaning,data filling and dimensionality reduction,and develops a disease risk prediction based on big data,model.The results obtained by predicting the model can provide some decision-making assistance for the medical staff.