Data Mining for Heart Disease Prediction Based on Echocardiogram and Electrocardiogram Data

Abstrak

Prediction of heart disease is a complicated thing when done with traditional medical analysis. Sometimes the prediction results are not as expected and even fail. The development of computational science offers a number of alternatives to help predict the type of heart disease based on the availability of patient medical record data. One alternative that can be used is data mining. The goal of this study is to implementing data mining to predict the type of heart disease based on medical record data from echocardiograms and electrocardiograms. Multilayer perceptron backpropagation (MLP) is used in this study. The results show that MLP with the Tanh activation function is a better predictive model than logistic and Relu. The classification accuracy level (CA) for MLP with Tanh is 0.788 for a data sharing scenario using k-fold cross validation and 0.672 for a data sharing scenario using Bootstrap. From both the CA level and the AUC value, MLP with the Tanh activation function is the best model with a value of 0.832 (k-fold cross validation) and 0.857 (bootstrap).