Study on the P-wave feature time course as early predictors of paroxysmal atrial fibrillation

Abstract

Atrial fibrillation (AF) is the most common cardiac arrhythmia in clinical practice, increasing the risk of stroke and all-cause mortality. Its mechanisms are poorly understood, thus leading to different theories and controversial interpretation of its behavior. In this respect, it is unknown why AF is self-terminating in certain individuals, which is called paroxysmal AF (PAF), and not in others. Within the context of biomedical signal analysis, predicting the onset of PAF with a reasonable advance has been a clinical challenge in recent years. By predicting arrhythmia onset, the loss of normal sinus rhythm could be addressed by means of preventive treatments, thus minimizing risks for the patients and improving their quality of life. Traditionally, the study of PAF onset has been undertaken through a variety of features characterizing P-wave spatial diversity from the standard 12-lead electrocardiogram (ECG) or from signal-averaged ECGs. However, the variability of features from the P-wave time course before PAF onset has not been exploited yet. This work introduces a new alternative to assess time diversity of the P-wave features from single-lead ECG recordings. Furthermore, the method is able to assess the risk of arrhythmia 1 h before its onset, which is a relevant advance in order to provide clinically useful PAF risk predictors. Results were in agreement with the electrophysiological changes taking place in the atria. Hence, P-wave features presented an increasing variability as PAF onset approximates, thus suggesting intermittently disturbed conduction in the atrial tissue. In addition, high PAF risk prediction accuracy, greater than 90%, has been reached in the two considered scenarios, i.e. discrimination between healthy individuals and PAF patients and between patients far from PAF and close to PAF onset. Nonetheless, more long-term studies have to be analyzed and validated in future works.

Alteration of the P-wave non-linear dynamics near the onset of paroxysmal atrial fibrillation

“…In fact, atrial electrophysiological evolution preceding the onset of PAF can be considered as an inherently chaotic and non-stationary process [5,10], which could be modelled by non-linear equations [11,12]. The present study hypothesises that the P-wave variability estimation over time through non-linear methods may help to anticipate recurrent PAF episodes which could be misdiagnosed by linear methods previously proposed [8,9]. To this respect, non-linear methods have previously demonstrated their interesting capabilities dealing with the envision of events related to AF [13,14], as well as in tracking the behaviour of this arrhythmia [15].…”

“…Thus, the rectified P-wave length was computed as its arc length (P al ) and the P-wave amplitude was characterised by its normalised root mean square (nrms) value (P nrms ) and its area (P area ). On the other hand, different temporal features related to the P-wave fiducial points have also proven to be indicative of an increased risk of AF development [4,8]. Hence, the P-wave duration (P dur ) together with the duration of its initial (P ini ) and terminal (P ter ) portions were computed.…”

“…Hence, the P-wave duration (P dur ) together with the duration of its initial (P ini ) and terminal (P ter ) portions were computed. The rhythm variability between successive P-waves was also estimated (PP) [8].…”

“…Recently, the analysis of P-wave variability over time has shown a certain ability to quantify electrophysiological alterations during the 2 h preceding the onset of PAF [8,9]. In these studies, every P-wave was delineated and then characterised by computing temporal distances between its fiducial points [8] or quantifying its morphology [9].…”

“…In these studies, every P-wave was delineated and then characterised by computing temporal distances between its fiducial points [8] or quantifying its morphology [9]. Next, the variability evolution of each analysed parameter was estimated by means of a linear fitting [8,9].…”