GA based KELM Optimization for ECG Classification

Abstract

Electrocardiogram (ECG) is the detection of the motion of heart rate. Cardiac diseases are very common in today’s routine and should be detected within time so that appropriate treatment can be given to the subject. This is identified by doctors manually. But sometimes the problems are so sensitive and unidentifiable that detection becomes late. In such cases, a system is needed which is accurate in doing classification between various forms of arrhythmias in ECG. Therefore, a novel method is proposed in which kernel extreme learning machine (KELM) is optimized with the help of genetic algorithm (GA). This experimentation is performed over UCI repository arrhythmia and PTBDB databases. Cumulants are utilized on UCI repository arrhythmia database for replacing the missing values. Using the proposed method, 100% accurate results are computed on PTBDB database and very promising results are achieved on the former database. Comparison is also performed with other available state-of-art approaches to highlight the efficacy of the proposed approach.