Prediction of Cardiac Diseases Using Machine Learning Algorithms

Abstract

Coronary illness or heart strokes are most common diseases across the world. To reduce this disease, prediction of heart strokes is needed to be done. Many machine learning are already existing in the medical field. Machine learning algorithms can be used to predict coronary illness or heart strokes. In this paper, various machine learning algorithms are applied on the dataset to predict the heart strokes or coronary illness. Dataset used is Cleveland’s dataset, where it contains 14 attributes which are the medical parameters of the patients. Few attributes are the results obtained from medical tests. Algorithms like decision tree, logistic regression, random forest, MLP, bagging are applied on the dataset to predict heart strokes. Implementation is done by dividing dataset into training and testing datasets, and algorithms are applied to find out the accuracies in predicting the heart strokes. Implementation is done using RStudio and WEKA. Even, we applied the same implementation by eliminating few attributes in dataset. The idea behind reducing (eliminating) attributes is that few attributes in the dataset are results of medical tests done to the patients. Sometimes, medical tests can be costly. So, if we can get the same accuracy with reduced attributes, we can conclude that prediction can be done with a smaller number of medical tests. From the implementation and analysis done, it shows that logistic regression provides the highest accuracy followed by decision tree.