Data-driven modeling of zebrafish behavioral response to acute caffeine administration

Over the last thirty years, we have witnessed a dramatic rise in the use of zebrafish in preclinical research. Every year, more than 5000 technical papers are published about zebrafish, many of them seeking to explain the underpinnings of anxiety through animal testing. In-silico experiments could significantly contribute to zebrafish research and welfare, by offering new means to support the 3Rs principles of replacement, reduction, and refinement. Here, we propose a data-driven modeling framework to predict the anxiety-related behavioral response of zebrafish to acute caffeine administration. The modeling framework unfolds along a two-time-scale dichotomy to capture freezing behavior along a slow temporal scale and burst-and-coast locomotion at a fast time-scale. Anchored in the theory of Markov chains and stochastic differential equations, we demonstrate a parsimonious, yet robust, modeling framework to accurately simulate experimental observations of zebrafish treated at different caffeine concentrations. Our results complement recent modeling efforts, laying the foundations for conducting in-silico experiments in zebrafish behavioral pharmacology.