Ventricular Arrhythmia Classification Based on High-Order Statistical Features of ECG Signals

One class SVM classification model based on high-order statistical features of ECG signals is proposed. This utilizes distinct features of variance, skewness and kurtosis between normal signals and ventricular arrhythmia ECG signals. The model based on a few simple features motivates immediate treatment for sudden cardiac event and wearable technology in practice. The classification algorithm shows significantly improved performance of 98.9% accuracy in correct classification in the experiment using the MIT-BIH Malignant Ventricular Arrhythmia Database (VFDB). It is expected to be used in real-time electrocardiogram monitoring system in conjunction with ECG measurement part and application part.