Heart Disease Prediction by Using Artificial Neural Networks Saeideh Kabirirad a , Hossein Kardanmoghaddam b , Vahidreza Afshin c a Department of computer engeneering, Birjand University of Technology,Birjand, Iran, kabiri@birjandut.ac.ir b Department of computer engeneering, Birjand University of Technology, ,Birjand, Iran, h.kardanmoghaddam@birjandut.ac.ir c Ministry of Education,Iran, afshin.vahidreza@gmail.com Abstract One of the effective methods to predict heart disease is to use artificial neural network. In this paper, we are going to predict heart disease with the help of information from instances of va, Cleveland, Hungarian and Switzerland databases through a variety of neural network. Thirteen inputs such as age, gender, cp, etc. are considered as effective factors to make predictive model and the final output would be a parameter (the closure of the coronary arteries of the heart). Experimental results show that our prediction system has high accuracy. Keywords: Heart Disease Prediction, Neural Network, MLP, Generalized FF, SVM, RBF, PCA. 1. Introduction Today, heart disease and vascular are the leading causes of death among human beings. One of the most common cardiovascular diseases is coronary artery disease. Many factors will raise the risk of heart disease such as cholesterol, blood pressure, lack of exercise, smoking, and so on. The World Health Organization (WHO) estimates that by 2030, nearly 23.6 million people die due to heart disease. To minimize the risk of heart diseases, such diseases should be detected based on physical signs and checkup, and delay of detection is caused negative effects, even to the point of death. One of the effective ways of diagnosing heart diseases is artificial neural networks (ANNs). ANNs have important characteristics including adaptivity, capacity to generalize input information and to give correct answers for unfamiliar data. Todays, ANNs have been widely used in the medical field to diagnose a variety of diseases such as cancer [1], diabetes, hepatitis, heart disease and so on. Many studies have been carried out to diagnose heart disease with the help of neural networks. Some of them are briefly reviewed in this article. Paper [2] predicts heart disease by using SVM and cascade neural network, the population included 13 characteristics of 270 patients. In the article [3], heart disease detection is done by utilizing information in cleveland database. MLP network with BP algorithm has been used with 13 numbers, three hidden layer, 2, 5 and 8 neurons and one output indicative of cardiac diagnosis. In the article [4], a decision support system based on MLP with BP algorithm is used to predict heart disease. In the article [5], a bunch of neural networks have been used to diagnose heart diseases. Methods based on classification build new models with the help of combination of previous possibilities or predicted values from several previous models. Paper [6] predicts heart disease by using SVM network and classifies patient’s history using cascade neural network. In the article [7], a prediction system for heart attack is suggested using MLP neural network and decision tree. In the article [8], an automated system to detect heart disease is invented by combination of MLP neural network and fuzzy-neural system. In the article [9] and [10], MLP neural network and BP algorithm is used to diagnose heart disease. Paper [11] has compared performances of classification techniques including logistic regression, decision trees and artificial neural networks, and has concluded that MLP is the best technique to predict presence of coronary artery disease in studied data set. Khemphila and Boonjing [12] have compared performances of the classification techniques and have resulted artificial neural networks have the least of error rate and the highest International Journal of Computer Science and Information Security (IJCSIS), Vol. 14, No. 1, January 2016 181 https://sites.google.com/site/ijcsis/ ISSN 1947-5500