The lagged central tendency measure applied to assess P-wave duration variability improves paroxysmal atrial fibrillation onset prediction

The prediction of paroxysmal atrial fibrillation (PAF) onset is an interesting clinical challenge, because the chronification of this highly prevalent arrhythmia could be avoided. Recently, the quantification of the P-wave duration variability over time has revealed a promising ability to detect accurately the onset of PAF. However, the possible scale-dependent variations in this P-wave variability have not been studied yet. In the present work that variations have been analyzed by using a m-lagged central tendency measure (CTM). Thus, once P-waves were delineated, their time course variability was quantified by computing CTM for lags m = 1;2;...;10. Statistically significant differences between ECG segments one-hour far from the onset of PAF and those immediately before the onset were obtained for every lag. Although no great differences were observed among the CTM values obtained for the studied lags, a predictive ability increase of about 3.5% was observed for m = 2 compared with m = 1. This result suggests the existence of scale-dependent dynamics within the transition process from sinus rhythm to PAF.