This text responds to the question "how can we use computational methods to improve the success rate in drug development?" It aims to show how modelling makes it possible to extract the maximum amount of information and predictive value from preclinical data.