Project T: Model selection for non-identifiable dose response models
Project leaders
Holger Dette
Abstract
This project extends methods from previous funding periods for model selection and design of experiments to render them suitable for application by our partner Novartis in the context of dose-response studies. A major goal is the minimization of the statistical uncertainty by an adequate modeling of the dose response relation and by optimal design of experiments. In particular, the "MCP-Mod" developed by the cooperation partner, which has been certified by the European Medicines Agency (EMA) and classified as a "fit-for-purpose" method by the U.S. Food and Drug Administration (FDA), will be improved.