Design And Implementation Of A Prediction Platform For Heart Disease Based On Random Forest

With the advent of mobile Internet medical era,increase the accumulation of big data,and the development of big data has also promoted the medical progress,how to combine big data with mobile Internet medical is the most important development in the field of intelligent medical treatment in the future.This paper combines data mining with mobile medical APP,designs and implements a prediction platform for heart disease based on random forest,the platform is focused on the improvement of the random forest algorithm,the design and implementation of the platform architecture and the performance evaluation of the platform,The main work and research contents are as follows:(1)For the random forest algorithm data imbalance,individual classifier single,large-scale data poor scalability and other shortcomings,proposed three improvements-1?Multilevel extraction of data sets using Bootstrap to ensure the balance of data;2?The clustering method is used to cluster individual classifiers,and a representative classifier is selected from each cluster to form a diversity forest.3?Parallelization of Random Forest with Hadoop Distributed Computing Capabilities.BDRF is designed according to the proposed scheme.(2)The requirement analysis of the heart disease prediction platform is carried out,and the specific design plan of the platform is given.The APP display platform completes the prototype interface design that corresponds to the business function.The background server completes the SSH framework logic design,the database scheme design,and the HDFS and MySQL data integration design in the heart disease prediction model.The background logic design adopts MVC mode to reduce the coupling between layers,so as to facilitate the expansion of the platform's business functions.HDFS and MySQL through the sqoop data mutual guidance,data stored in MySQL unified,convenient data query and management.(3)The various business modules of the platform are realized,and the achievement effect is demonstrated.The decision tree algorithm,stochastic forest algorithm and BDRF algorithm are respectively modeled and analyzed according to the UCI heart disease dataset to verify the effectiveness of the heart disease prediction platform And performance improvement.