Reinforcement learning discovers new mechanisms of reentry in excitable media
The transition from transient excitation to sustained reentry is a fundamental problem in the physics of excitable media. In cardiac tissue, reentry underlies many life-threatening cardiac arrhythmias, yet the pathway to initiation of reentry remains incompletely understood. Here, we formulate reentry initiation as a reinforcement-learning problem in which an agent applies sequences of spatial…
Reinforcement learning has uncovered novel pathways for generating reentry in excitable media, such as cardiac tissue. This approach models reentry initiation as an agent learning to apply stimulation patterns, balancing rewards for lasting activity and penalties for excessive stimuli. In one-, two-, and three-dimensional cellular automata simulations, the agent identified multiple mechanisms for unidirectional propagation and reentry, including previously known methods involving superthreshold and subthreshold stimulation, as well as two previously unknown methods relying solely on subthreshold stimuli.
One mechanism involves sequentially applying stimuli at different locations and times, while the other consists of a spatial mechanism where multiple subthreshold sites work together to initiate reentry. In geometries with boundaries or branching structures, the learned protocols also utilized structural source-sink asymmetries.
Confirmed by optogenetic experiments in cardiac monolayers, the discovered subthreshold patterns successfully induced unidirectional propagation, with preliminary evidence suggesting similar patterns could shape early propagation in whole-heart experiments. Overall, reinforcement learning offers a versatile framework for discovering reentry mechanisms in various geometries and formulating testable hypotheses about reentry initiation in excitable systems.
Written by urgent.news from bioRxiv's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
