Heart Disease Prediction Using Machine Learning Models

Heart disease is a condition that poses a serious and potentially deadly challenge encompassing a spectrum of heart-related issues ranging from irregular heartbeats and congenital defects to conditions such as atherosclerosis and heart failure. Accurately predicting heart disease is critical for enabling early detection and allowing for prompt treatment thereby preventing more severe complications and possibly saving multiple lives. The aim of this study is to identify the most accurate classifier for predicting heart disease using a dataset of 918 records and 11 features that was derived from five different sources: Stalog, Cleveland, Switzerland, Hungarian and Long Beach VA. In the study, the machine learning algorithms that are employed include the k-nearest neighbours, Decision Tree, Random Forest, Support Vector Machine and Artificial Neural Network. Techniques such as feature selection as well as hyperparameter tuning were applied to improve model performance which were assessed using different performance metrics. The most accurate model was Random Forest that demonstrated an accuracy of 88.59%. A web-based interface was also developed using the Flask library to allow users to input patient attributes and to obtain predictions from the pre-trained models.