Prediction of atrial fibrillation using the recurrence complex network of body surface potential mapping signals

OBJECTIVE: Atrial fibrillation (AF) is the most common type of persistent arrhythmia. Early diagnosis and intervention of AF is essential to avert the further fatality. The technique of noninvasive electrical mapping, especially the body surface potential mapping (BSPM), has a more practical application in the study of predicting AF, when compared with the invasive electrical mapping methods such as the epicardial mapping and interventional catheter mapping. However, the prediction of AF with noninvasive signals has been inadequately studied. Thus, the aim of this paper was to analyze the properties of atrial dynamic system based on the noninvasive BSPM signals (BSPMs), using the recurrence complex network, and consequently to evaluate its role in predicting the recurrence of AF in clinical aspect. METHOD: Twelve patients with persistent AF were included in this study. Their preoperative and postoperative BSPMs were recorded. Initially, the preoperative BSPMs were transformed into the recurrence complex network to characterize the complexity property of the atria. Subsequently, the parameters of recurrence ratio (REC), determinism (DET), entropy of the diagonal structure distribution (ENTR), and laminarity (LAM) were calculated. Furthermore, the difference in the parameters in the four regions of the body and the difference obtained from the dominant frequency (DF) method were compared. Finally, the results obtained for the atrial dynamic system complexity from a 12-lead electrocardiogram (ECG) from the BSPMs were discussed. RESULTS: Our study revealed that the patients whose REC is greater than an average threshold, and with a lower LAM presented a much higher possibility of AF recurrence, after the AF surgery. CONCLUSIONS: The recurrence complex network is a useful and convenient way to evaluate the nonlinear properties of the BSPMs in patients with AF. It has good immunity to the lead position and has a potential role in the understanding of predicting the recurrence of AF.