Detection of A Fib and its Classification using SVM

Atrial Fibrillation (AFib) is an irregular rhythm of the heart that initiates in the atria. In AFib, the rate of the rhythm of the heart is asynchronous compared to the rate of the rhythm of a healthy heart. The electrical impulses directed from the atria cause a very fast rhythm due to which the atria cannot contract or squeeze blood effectively into the ventricles. Patients suffering from AFib are prone to heart failure, stroke and eventually death. The presence of AFib can be detected by tracking the changes in the P wave of ECG. In this paper, an algorithm is developed that can detect the AFib by using morphological features of atrial activity acquired from ECG. Classification of AFib is done using (SVM) Support Vector Machines. This approach is analyzed for AFib on 44 records including Normal Sinus Rhythm (NSR) and AFib data that are extracted from various sources such as MIT-BIH NSR Database, Intracardiac AFib and AF Termination Challenge. Satisfactory result is obtained for AFib as 91.67% Specificity, 87.5% Sensitivity, and 90.91% Accuracy.