Classification of ECGs using Intensity Weighted Fire-Fly Optimization and Multimodal Decision Learning algorithm

Abstract

This paper proposes a method of feature reduction process i.e., selection of the best features for accurately identifying an electrocardiogram (ECG) signal as basically normal or abnormal. This is done by Intensity Weighted Fire-Fly Optimization (IWFFO) Algorithm. After extracting the best features, a kernel model for machine learning is proposed to classify the category of the signal as normal or abnormal. This is done by Multimodal Decision Learning (MDL) classification method. The proposed work is compared with traditional ECG signal classification method i.e., support vector machine (SVM) by considering the parameters like true positive (TP), true negative (TN), False positive (FP), False negative (FN), False rejection ratio (FRR), false acceptance ratio (FAR), global acceptance ratio (GAR), confusion matrix (CM), Kappa coefficient (KC), Sensitivity, Specificity and Accuracy. Records for MIT-BIH database are used for performance evaluation. Simulation results indicate the ECG signal as normal or abnormal with respect to the above defined parameters.