A modeller's guide for biomedical discovery

Mathematical and computational modelling can do far more than reproduce experimental data or make predictions. When used with intent, models become instruments of discovery: they translate qualitative biological ideas into quantitative, testable hypotheses; they connect microscopic questions to macroscopic data; and, crucially, they help falsify plausible but incorrect mechanistic narratives. This Perspective explores how modelling achieves these goals. I begin by outlining why classical experimental strategies and conventional statistics sometimes fall short of addressing the mechanisms we care about. I then present a model-centred workflow for scientific discovery, using clonal lineage tracing as a running example. The second half focuses on a phenomenon that both limits and empowers model inference - universality - and explains how to turn it from a curse into an opportunity. I conclude with a concise, practical guide that distils these ideas into steps for day-to-day biomedical research.