Document Type : Original Article

Authors

1 MSc student, Computer Engineering, Department of Computer Engineering and Information Technology, Payame Noor University, IRAN

2 Associate Professor, Computer Engineering Department of Computer Engineering and Information Technology, Payame Noor University, IRAN

Abstract

Introduction: Selecting an appropriate method for modeling and analyzing health data based on available data is very crucial. This study was conducted according to Probabilistic Neural Network (PNN) to detect if the coronary artery is closed or not.Methods: This study was diagnostic and it was implemented on patients of Kowsar Hospital in Shiraz, Iran who were exposed to Coronary artery angiography in September 2013. The number of population was calculated based on related formulation and the division of neurons in hidden layer by error rate of 0.1. Therefore, 152 patients were randomly selected for this research. In these implementations, 85% of data was used for training phase of network and 15% for the test phase. In this study, Probabilistic Neural Network (PNN) was used for prediction of coronary artery disease. The proposed neural network was implemented through facilities and functions of MATLAB software (7.12.0version) and simulated by a system of core i5, 2.4 GHz processor and 4GB memory and windows7 as operating system.Results: Performance indicators of this system were sensitivity and specificity. The presented system performance on the basis of these indicators was achieved1 and 0.94, respectively. Ultimately, the designed and implemented system could confirm its superiority for diagnosis of patients of coronary artery according to similar studies.Conclusion: The results of this research indicated that in the studied population, probabilistic neural networks could achieve more accurate diagnosis for coronary heart disease comparing other studied neural networks. Due to high specificity and sensitivity of the system, it can prevent the possible side effects and injuries of angiography for the patients who don’t need it. And also, it can distinguish the patients who really need diagnostic actions in the least time and the most accuracy.

Keywords