Enhanced Heart Disease Prediction Using HCR-PSO Based Data Analytical Model

Abstract

Data Mining is an important aspect of diagnosing and predicting diseases in automatic manner. It involves developing appropriate techniques and algorithms to analyze data sets in medical field. At present, heart disease has excessively increased and heart diseases are becoming the most fatal diseases in several countries. In this paper, heart patient datasets are investigate for building classification models to predict the heart disease. This paper implements feature extraction technique construction and comparative study for improving the accuracy of predicting the heart disease. By the use of HCR-PSO (Highly Co-Related PSO) feature selection technique; a subset from whole normalized heart patient datasets is acquired which have only significant attributes. The study emphasized on finding the effective heart disease prediction construction by using various machine learning algorithms that are KNN(K-Nearest Neighbor), Random forest, SVM(Support Vector Machine), Bayesian network and MLP(Multilayer Perceptron). The research work central point is on finding the efficient classification algorithm for the prediction of heart disease in the early stage based on the accuracy using validation metrics that are Mean Absolute Error(MAE), Relative Squared Error(RSE) and Root Mean Square Error(RMSE).