Intelligent System based on Genetic Algorithm and Support Vector Machine for Detection of Myocardial Infarction from ECG signals

Abstract

Myocardial Infarction (MI) is one of the well-known heart attacks. This cardiac abnormality occurs when the artery connecting the heart is blocked. The main aim of this paper is to identify electrocardiogram (ECG) signals using morphological, time-domain and discrete wavelet transform (DWT) features in order to distinguish MI samples from normal. To this end, a model based on support vector machine (SVM) and Genetic algorithm (GA) is proposed. The whole experimental study was conducted on an open database called PTBDB. According to experimental study, 9 features were determined by GA as most relevant. Also, the dimension of feature set was decreased from 23 to 9. Lastly, the sensitivity and specificity were achieved as 87.80% and 85.97%, respectively.