Transfer Entropy Provides Insight into Wave Propagation Dynamics: A Preliminary Study

Objective: The highly complex wave propagation dynamics of the heart are of profound clinical significance, particularly during episodes of atrial fibrillation (AF) where mechanisms remain unknown. We hypothesise that measuring information flow using the statistical approach transfer entropy will provide insights into wave-propagation dynamics.

Methods: Simulations were carried out on two-dimensional, isotropic, square grids (80 × 80) in a custom C++ environment. Multiple AF wave-propagation scenarios demonstrating (1) a train of planar and (2) stable spiral waves were developed using the Tusscher–Panfilov phenomenological model. Extracellular bipolar electrograms were constructed, and transfer entropy between pairwise electrograms computed to measure the direction and degree of information flow.

Results: During planar wave propagation, adjacency matrices showed strongest connectivity and information flow from top to bottom nodes, reflecting the unidirectional wave propagation simulated in the planar model. Connectivity was also found strongest at neighbouring nodes. In comparison, adjacency matrices demonstrated information flow in both horizontal and vertical directions from the pivot point of the rotor in the spiral model, with a region of low information flow to nodes located at the edge of the grid.

Conclusion: This study demonstrates the capacity for transfer entropy to assess wave propagation dynamics and information flow. Future investigations are required to understand the relationship between transfer entropy and wave propagation during human AF to provide potential fundamental and clinically applicable insights into the AF mechanism.