Prediction of paroxysmal atrial fibrillation onset by using ECG

The advances in medical technologies increased the importance of a reliable arrhythmia predictor to predict the arrhythmia onset. The goal of this study is to predict paroxysmal atrial fibrillation (PAF) onset by using electrocardiograph (ECG) signals. 53 pairs of 30 minute ECG signals were used for this purpose. Complex correlation measures (CCM) were calculated for 5 minute ECG windows. It was observed that the standard deviation of CCM values obtained for records just before the PAF onset were significantly greater than the ones obtained away from PAF onset. Based on this observation, the records just before to PAF onset were correctly predicted with 72% success rate.