An electrocardiogram marker to detect paroxysmal atrial fibrillation

Abstract

An electrocardiogram marker to detect patients who have paroxysmal atrial fibrillation (PAF) is reported. The data set of ECG records made available by PhysioNet for Cardiology Challenge 2001 was used. The method uses a filtered time series with a frequency range between 3 and 9 Hz obtained from the electrocardiogram record. Typically, frequencies observed when a patient is in atrial fibrillation and atrial flutter is within this frequency range. Filtering was done using wavelets. Thereafter, the temporal properties of this filtered time series are studied. The temporal properties studied are the standard deviation, standard deviation of successive differences, and the length of the ellipse in the Poincare plot. The results indicate that these temporal properties of patients with PAF are depressed compared to the healthy group. A marker based on these temporal properties shows promise in detecting PAF when patients are in normal sinus rhythm. Results also show that the values for these temporal properties for a patient with PAF show little variation with time, and its measure is not dependent on the time of occurrence of a PAF episode.