Causality in Atrial Fibrillation determined by transfer entropy

High level of complexity makes characterization of wave conduction during Atrial Fibrillation (AF) very difficult. Here we aim to use statistical approach characterizing AF as a system with determined information flow using a concept of transfer entropy. Left and right atrial 60 s electrograms were recorded at high right atrium (HRA), coronary sinus (CS) and Left Atrial Appendage (LAA) in 42 patients undergoing catheter ablation of AF. Transfer entropy (TE) was used to asses causality calculating direction and extent of information flow between neighboring sites in the atria. TE was calculated between electrograms recorded along each catheter. Additionally, numerical analysis were performed on a set of unidirectionally coupled stochastic signals modeling electrical activity during AF. We found an asymmetry in information flow along the catheters. In HRA catheter, in general, information flows from proximal to distal portion of the catheter and in CS from the distal towards the proximal portion. The dominant flow of information from the base into the LAA was the most pronounced and in agreement with believed passive role of LAA in maintenance of AF. Information flow in the atria during AF is asymmetric and it is possible to determine the direction of the flow using concept of entropy transfer.