Cardiac Arrhythmia Classification Using Machine Learning Techniques

Cardiac arrhythmia refers to the medical condition during which the heart beats irregularly. Effective monitoring of cardiac patients can save enormous amount of lives. During the past few years, much importance has been gained by cardiac disease classification and prediction. This paper presents a model for diagnosis of cardiac arrhythmias. It works by selecting best features with the help of three filter-based feature selection methods on three different machine learning methods applied over cardiac arrhythmia dataset. Feature selection is a crucial preprocessing step in determining factors responsible for patients suffering from arrhythmia. In particular, we want to examine the underlying health factors of patients that could potentially be a powerful predictor for deaths that are related to heart. Three types of machine learning methods, namely, linear SVM, random forest, and JRip, were employed for analyzing the performance of the feature selection methods. Experimental analysis shows that highest accuracy of 85.58% was obtained with random forest classifier using gain ratio feature selection method with a subset of 30 features.