Abstract:

Ectopic beats are a common cause of cardiac arrhythmia. In order to automatically detect and classify ectopic beats in the ECG a new method based on a Support Vector Machine (SVM) was developed. The numerical patterns needed for this classification task were obtained from rhythmical and morphological characteristics of the QRS complexes. The SVM was trained and tested using the MIT-BIH Arrythmia Database and the MIT-BIH Supraventricular Arrythmia Database. A sensitivity of 92.46% was achieved.