A Novel Approach for Detection of ECG Beats Events

In this paper, normal and abnormal ECG beats classification methodology, in which amplitude modulation and frequency modulation bandwidth features, are obtained from the analytical representation of intrinsic mode functions (IMFs) of ECG beats. This paper proposes an automatic, normal and abnormal ECG beats classification methodology, which uses amplitude and frequency modulation bandwidth features computed from IMFs of ECG beats as input to the proposed Genetic algorithm (GA) tuned Least squares support vector machines (LSSVM) classifier. These features are tested with the Kruskal-Wallis statistical test to check their class discrimination capability. Then, these features were classified with a Genetic Algorithm-tuned LSSVM classifier to identify normal and abnormal ECG beats with 92.3$ accuracy.