Automatic Detection of Arrhythmia Using Optimized Feature Selection

Abstract

In present life scenario, it is common to witness various types of heart related diseases irrespective of age, whether young or old. Enhanced classification had been done this research work to automatically identify the test Electrocardiogram (ECG) is a ‘normal’ case or ‘Arrhythmia’ case. The novelty in this classification had been attained by adopting efficient pre-processing and feature selection methods. The frequency domain features extracted from the ECG signals were carefully selected using Memtic redundancy (MR) approach, which could ideally filter the redundant features obtained from wavelet lifting schemes. The proposed method could yield an accuracy of 99.75% when 65,448 ECG signals from MIT-BIH Arrhythmia database were tested.