Andrew Hooker, Mats Karlsson, Sebastian Ueckert
It can be a challenge to evaluate how well pharmacometric models fit to experimental data. To make this assessment, evaluations can be based on model predictions, residuals, simulations from the model, simulations followed by evaluation, and simulations followed by full re-estimation. We develop diagnostic tools based on these principles and for both continuous and categorical type data. Further, knowledge about model and parameter uncertainty is often crucial to understanding model fit to data and for model-informed decision-making. To that end, we develop diagnostic tools to assess existing methods of uncertainty estimation. We also develop new methods for estimating model and parameter uncertainty and how these methods can be used in decision making.