CLASSIFICATION OF CORONARY ARTERY DISEASE USING HYBRID APPROACH

Abstract

Cardiovascular diseases (CVDs) are known globally to be among the major cause of sudden death, hence the prompt identification of CVDs could help reduce the casualties recorded via them. Diagnosis is a medical term used to describe the “process” involved that lead to the identification of a specific illness. When it comes to Coronary Artery Disease (CAD), however, this is achieved by following sophisticated and costly medical procedures in well-equipped hospitals and healthcare facilities. Furthermore, these procedures usually require only highly qualified medical experts to apply invasive methods. The number of patients who have access to this facility is limited. This research employs the use of Deep Neural Network (DNN) for the diagnosis of CAD for four (4) different datasets with Particle Swarm Optimization (PSO) assisted method for DNN. The aim of this research is to enhance the accuracy of diagnosing heart disease. Developed a conceptual framework to analyze CAD, also integrated PSO training algorithm to train DNN. Finally, evaluate and validate the performance of the proposed hybrid model with benchmark model. The research has shown that PSO is an effective evolutionary computing technique that improves the accuracy of classification. PSO selects the most optimum weight for DNN and increases the classification accuracy. The percentage improvement of the PSO hybridization to DNN are 8.8%, 11.4%, 3.3%, and 11.0% for Cleveland, Hungarian, Switzerland and ValongBeach respectively. The method put forward can improve patient diagnosis reliability and performance as it concerns CAD detection.