Multi-kernel SVM Approach for Arrhythmias Classification

Heart diseases are a critical issue all over the world; WHO estimates that approx 17.9 million people died in 2017 due to heart disease; our proposed model detects cardiac arrhythmias and helps healthcare professional to early and easy diagnostic of different heart diseases, for classification of arrhythmias. We use different features of ECG signal which are extracted by peak finder algorithm and apply higher-order statistics, and MIT-BIH database is used for ECG signals. The proposed method uses different kernels for SVM, and different kernel performances are evaluated and compared; then, we know the most suitable kernel for classification of arrhythmias.