Morphodynamics, the characterization and change of cell shape in time, appears complex and random, but there are underlying structure and simplicity to be discovered. Discovered features may in turn inform us about the strategies cells implement, or any underlying molecular and physical mechanisms. In this seminar I will discuss two things informally. In the first part, I will review previous work from my group on migrating amoeba, with focus on the relation between morphodynamics and the physical limits of sensing. In the second part, I will discuss ongoing work on classifying phenotypes of soybean rust, an agriculturally detrimental fungus, when exposed to pesticides. Taken together, my talk will aim to link mechanistic modelling and machine learning, with challenges arising from the vast range of spatiotemporal scales.