Leveraging agent-based models and deep reinforcement learning to predict taxis in cell migration: Insights from barotaxis

Abstract

We present a novel computational framework that combines Agent-Based Modeling (ABM) with Reinforcement Learning (RL) using the Double Deep Q-Network (DDQN) algorithm to determine cellular behavior in response to environmental signals. We showcase its potential by modeling how pressure gradients direct cell migration in confined environments, a phenomenon known as barotaxis. By integrating RL, the model allows cells to learn and adapt their migration behavior based on sensed pressure gradients, capturing the dynamic, environment-dependent nature of cell behavior. We validate the framework using real microfluidic devices and experimental data, demonstrating the models ability to predict barotactic migration. Thus, this approach introduces a novel direction for modeling how cells sense and transduce environmental cues into biological behaviors.