Analysis of the sustained ventricular arrhythmias from SAECG using artificial neural network and fuzzy clustering algorithm

Patients with sustained ventricular tachycardia and ventricular fibrillation have a potential for sudden death. After myocardial infarction the chance to get sustained ventricular tachycardia or ventricular fibrillation increases, thus reduction in number of sudden death requires advanced predictive procedures. In this study, frequency domain feature extraction, clustering and classification models are combined for providing an integrated system for the sustained ventricular arrythmias. The radial basis function network and the fuzzy c-means algorithm for training clustering and classification were investigated. These techniques do not have limitation of the previous classical procedures. The value of sensitivity and specificity, on the data was used here, for the RBFN were found to be 92.3% and 71.4%, respectively, also the value of sensitivity and specificity for the PCM were found to be 84.6% and 71.4%, respectively.