|International Journal of Applied Information Systems
|Foundation of Computer Science (FCS), NY, USA
|Volume 12 - Number 29
|Year of Publication: 2020
|Authors: Bakare K. Ayeni, Baroon I. Ahmad, Abdulsalam A. Jamilu
Bakare K. Ayeni, Baroon I. Ahmad, Abdulsalam A. Jamilu . An Improved Classification Method for Diagnosing Heart Disease using Particle Swarm Optimization. International Journal of Applied Information Systems. 12, 29 ( May 2020), 11-20. DOI=10.5120/ijais2020451857
Today, the diagnosis of some of the major cardiovascular diseases, for example Coronary Artery Diseases (CAD), heart rhythm problems, Ischemic, Atrial Fabrication and so on is generally accomplished by following modern and costly therapeutic strategies performed in well-equipped medical institutions. In addition, these procedures usually require the application of invasive methods by only highly qualified medical experts. Although this approach gives a high degree of accuracy regarding diagnosis, but the number of patients having access to this facility is limited. Hence, the development of an easily accessible method for cardiovascular disease diagnosis is highly desirable. In this research work, the past work which employs the use of Deep Neural Network (DNN) for the diagnosis of heart disease is extended, CAD for four (4) different datasets was used with Particle Swarm Optimization (PSO) assisted method for DNN to enhance the accuracy of diagnosing heart disease, which is very complex in the healthcare practices was proposed. The aim of this research is to enhance the accuracy of diagnosing heart disease. A conceptual framework to analyze CAD heart disease was developed with the end goal to improve human services partner for specialists with convenience in the advancement of treatment of disease, also integration of the PSO training algorithm to train the DNN and finally, evaluation and validation of the performance of the proposed hybrid model with benchmark model Neural Network Classifier was carried out to obtain a comparison of the proposed model to the existing classification models. The research datasets are obtained from data mining repository of the University of California, Irvine (UCI) Machine learning repository. Experimental results show that training DNN using PSO results 94%, 94.9%, 95.5%, 95.0% in accuracy for Cleveland, Hungarian, Switzerland, and VaLong beach respectively. The technique puts forth can be used in CAD detection.