Cellular Automata for Fast Simulations of Arrhythmogenic Atrial Substrate

Abstract

Atrial biophysical simulations can improve therapies by simulating pharmacological and ablative strategies, however their computational times are not compatible with the diagnostic ones. Discrete models such as cellular autom ata (CA) lower computational times by considering a finite number of states evaluated through restitution properties, although it is necessary to question whether this approach is sufficient to reproduce pathological simulations. The analysis of biophysical atrial simulations, under both healthy conditions and different degrees of electrical remodeling, shows an expected increase of Action Potential Duration (APD) with the previous Diastolic Interval (DI) interval. Short-term memory of atrial cardiomyocytes was observed as the dependency of the predicted APD+1 with the previous activation (APD0): shorter APD0 provoked shorter APD+1, and this effect was comparable to the effect of previous DI. Independent prediction based on both APD0 and DI allowed better estimation of APD+1 values, compared to using DI alone (p << 0.01). Finally, the comparison of re-entrant activity in CA simulations showed a closeness between biophysical and CA results, where the CA automata reproduced reentrant patterns and cycle lengths of different states of atrial remodeling. Atrial automata considering short-term memory allows to accurately reproduce the arrhythmic behavior of pathological tissue with computational times for clinical use.