Classification of paroxysmal atrial fibrillation using second order system

Abstract

In this paper, we monitored and analyzed the characteristics of atrial fibrillation in patient using second order approach. Atrial fibrillation is a type of atria arrhythmias, disturbing the normal heart rhythm between the atria and lower ventricles of the heart. Heart disease and hypertension increase risk of stroke from atrial fibrillation. This study used electrocardiogram (ECG) signals from Physiobank, namely MIT-BIH Atrial Fibrillation Dataset and MIT-BIH Normal Sinus Rhythm Dataset. In total, 865 episodes for each type of ECG signal were classified, specifically normal sinus rhythm (NSR) of human without arrhythmia, normal sinus rhythm of atrial fibrillation patient (N) and atrial fibrillation (AF). Extracted parameters (forcing input, natural frequency and damping coefficient) from second order system were characterized and analyzed. Their ratios, time derivatives, and differential derivatives were also observed. Altogether, 12 parameters were extracted and analysed from the approach. The results show significant difference between the three ECGs of forcing input, and derivative of forcing input. Overall system performance gives specificity and sensitivity of 84.9 % and 85.5 %, respectively