Dynamical systems modeling of longitudinal data

How can dynamical system parameters be better inferred from longitudinal data and clinical measurements over time? We apply maximum likelihood and Bayesian approaches to train and validate dynamical systems models. Based on a model’s formulation, we also perform structural identifiability and practical identifiability analyses. These approaches allow us to select models, evaluate model complexity, and test theories of biological processes regarding their utility in explaining real-world experimental or clinical data.