Classification of CAD dataset by using principal component analysis and machine learning approaches

Abstract

Machine-Learning (ML) methods are applied to diagnose diseases and to observe disease developments. We utilized several ML methods on Z-Alizadeh Sani dataset, which is about Coronary Artery Disease (CAD). We applied t-test for feature selection and then Principal Component Analysis (PCA) to reduce dimensionality because of small sample size. 10-fold Cross-Validation was applied to ML methods, which achieved higher than 80% average accuracy. Besides, sensitivity and specificity results are around 70% and 90%, respectively. The Artificial Neural Network reached 93% AUC, which is the best performance out of six methods. The overall results are quite promising compared to the previous study.