Atrial fibrillation detection using support vector machine

This article introduces a new method for detection of atrial fibrillation (AFib) using a support vector machine (SVM). AFib could lead to heart failure and stroke and thus an AFib early detection is very important. In this article, an SVM and variabilities of electrocardiographic heart rate are employed to detect AFib. Radial basis functions (RBF) is utilized for SVM. Different SVM constructions are tested to find the best one. Furthermore, two features of electrocardiogram are examined as the inputs of SVM. Using clinical electrocardiogram, the proposed method find the performance of 95.81 %, 98.44% and 97.50% in terms of sensitivity, specificity and accuracy.