Heart Disease Early Prediction Using Machine Learning Approaches

Abstract

According to the recent WHO (World Health Organization) report, heart diseases are becoming more prevalent. This causes the results in death of 17.9 million individuals every year. It becomes more difficult to detect and begin therapy at an early stage as the population grows. As a result of recent technological advancements with machine learning approaches have speed up the health sector by several researches. The goal of this particular approach is used to develop the machine learning model for early prediction of heart disease using the relevant parameters which this work has taken. The Cleveland dataset has been taken for UCI (University of California Irvine machine learning repository) heart disease prediction, which includes 14 different major parameters for analysis. The development of the model has made use of machine learning methods includes, Random Forest (RF), Support Vector Machine (SVM), K-Nearest Neighbour (KNN), and Decision Tree (DT). With the use of these conventional machine learning techniques, this work has attempted to identify correlations between the various features present in the dataset with the purpose of effectively predicting early heart disease with maximum accuracy. The suggested approach depicts the outcomes of the three algorithms namely, SVM, Random Forest, and Decision Tree. This present work has integrated all three methods to determine whether the patient is having possibility of heart disease or not. Through that this approach has given 89% accuracy at the end. Keywords: SVM; Naive Bayes; Decision Tree; Random Forest; Logistic Regression; Adaboost; XG-boost; python programming; confusion matrix; correlation matrix