A Novel Approach for Heart Disease Prediction Using Genetic Algorithm and Ensemble Classification

Abstract

Coronary Artery Disease (CAD) is one of the leading causes of death in humans across the world over the last few decades. Coronary artery disease also leads to disability, decreased quality of life and serious illness. CAD can be controlled by identifying the risk factors and on timely diagnosis can also help to reduce the cause of heart failure (death). The conventional method of going through medical history proved that they were less effective in early identification of the disease. So, modern and emerging methods like AI, Machine Learning are more reliable and effective in identifying people with heart disease and can help in containing mortality rate. In the proposed work, three machine learning classification methods (Random Forest, XGBoost and Neural network) are auto-tuned using genetic algorithms to find the most prominent features for maximizing classification performance. These methods are applied to the Z-Alizadeh Sani dataset having demographic examination, ECG, Laboratory and echo data of 303 patients. The computational results of the above application show that three approaches need further ensemble by giving equal importance to all three models to increase the overall performance in assessing the risk and forecasting the presence of disease in the 303 patients.