Publication

Abstract : Coronary Artery Disease (CAD) is a major contributor to severe cardiac events, entailing early and accurate diagnosis for prevention of such events as heart failure and cardiac arrest. This study has a novel approach to CAD prognosis, predicting severity levels - Low, Mild, and High through Machine Learning (ML) techniques. The proposed model not only detects the presence of CAD but also provides comprehensive insights which includes mini to micro feature level recognition for severity level checker and CAD depending on individual’s health condition, facilitating personalized healthcare different from traditional methods. The work uses UCI Heart Disease Data dataset from Kaggle with 607 samples of data including 14 features. The proposed work involves the first phase of investigation of all features, applying the ML classifiers and obtained the accuracy results as follows: Linear Regression (LR)-0.8442, Support Classifier (SVC)-0.8688, Decision Tree (DT) - 0.9836, Random Forest (RF)-0.9846. Further, identified the best classifier among the four classifiers utilized. RF algorithm achieved the highest accuracy rate. The second phase involves applying the best ML classifier to the selected prominent compact set of features. The third phase involves predicting the severity level of CAD for the identified CAD samples. The result obtained after experimentation is identifying concise features of 8 collected from the 14 features in the dataset, further implementing ML algorithms to find the best classifier for predicting CAD – RF algorithm. The presence of CAD was predicted using the ML algorithm, if present the severity level was identified long with healthcare feedback on how to improve the patient’s situation overall.