Speakers
Description
Understanding biological systems requires models that are both grounded in empirical observations and mechanistically interpretable. Recent advances in equation learning provide powerful new tools to infer governing equations directly from biological data, bridging modern machine learning with classical mathematical biology. This minisymposium brings together leading and emerging researchers developing state-of-the-art approaches for learning biological dynamics from data. The session will feature neural-network–based methods, including universal differential equations and biologically-informed or physics-informed neural networks, which enable learning of full dynamical models or unknown components within mechanistic frameworks. Speakers will present recent advances demonstrating how these techniques can uncover interpretable biological structure while maintaining predictive power. Complementing these approaches, regression-based and probabilistic methods – such as SINDy variants and Bayesian equation learning – will be highlighted through work that has delivered pioneering results across cell biology applications. By bringing together researchers at different career stages and showcasing complementary methodologies, this session aims to provide a broad and balanced perspective on current equation-learning techniques in mathematical biology, fostering discussion and future collaborations.
