Research On The Heart Disease Risk Models Based On LDBN

With the development of technology,deep learning and recurrent neural network technology are widely used in various fields and solve many difficult problems plaguing many areas for many years successfully.At the same time,along with the rapid development of society,the people’s material life level is improved continually,the average lifespan is prolonged.Heart disease has increasingly become the center of interest to people.Many heart disease risk prediction models are proposed by many researchers,but some prediction models can’t make more accurate predictions of heart disease because of the training algorithm itself that used has the limitation that gathering speed slow and easy to fall into local optimum.Therefore,it is of great significance to study the heart disease risk model.Under such a background,we build a heart disease risk prediction model based on LDBN(Long Deep are Network)for more accurate prediction of heart disease combing with DBN(Deep Belief Network)and the improved recurrent neural Network(Long Short Term Memory),and using heart disease diagnosis relevant medical data.In this paper,firstly we elaborate the background and significance of the heart disease,and introduce the research status of the domestic and international of heart disease,deep learning and recursive neural network.Secondly,the networks associated with construct LDBN are introduced in details from the network structure and training algorithm.Then,according to the data preprocessing phase,the sample simulation stage,the stage of similarity calculation,the simulation sample processing phase,the data set expansion and processing stage,the stage of DBN network training and the testing phase of LDBN introduces the build process of LDBN network,and introduces in detail the data preprocessing phase,the sample simulation stage,the stage of similarity calculation and the DBN network training stage is presented in detail.Finally,the experiment is performed using LDBN model.The experiment includes two stages: First,we simulate target samples after that process to expand thedata set by using the LSTM algorithm and BPTT algorithm.The risk prediction of heart disease is conducted in the second stage.And we evaluate the prediction results of different heart disease risk prediction models using classification accuracy evaluation criterion.The experimental results show that the heart disease risk prediction model based on LDBN can get higher classification accuracy.