Alejandro F. Villaverde

This paper evaluates the predictive capabilities of existing mathematical models of T-cell activation by assessing parameter identifiability and sensitivity.

The paper develops a general method to derive expressions for Emax and EC50 in T-cell activation models and applies it to models lacking these expressions.

The paper develops a method to automatically discover structurally identifiable and interpretable dynamical models from biological data using SINDy-PI and additional procedures.

The paper warns that the Scaling Invariance Method can miss non-identifiability and non-observability caused by symmetries other than scaling, and suggests using symbolic computation tools instead.

The paper benchmarks optimization methods for parameter estimation in large kinetic models and finds that a hybrid metaheuristic outperforms multi-start local searches.