zoom link, opens an external URL in a new window

Meeting-ID: 280 519 2121
Passwort: 584190

Abstract:

In clinical trials, there are generally two conflicting objectives: one is to gain as much information from the experiment as possible, while the other is not to give doses that are either ineffective and/or have serious adverse effects. In Phase I/II, safety and efficacy outcomes are investigated jointly within the same trial, which leads to improved precision of estimating both efficacy and toxicity compared to separate phases. As the observations are multivariate, it is not straightforward to compute optimal designs for the models used in these cases. In the talk, we will address this problem and show how to adapt algorithms designed for the more conventional models with univariate observations to effectively tackle the complexities introduced by multivariate data. Specifically, we will show how to compute locally optimal and adaptive designs in continuation ratio model that has been used in phase I/II trials. To this end, we will use the algorithms in the R library OptimalDesign that enable the user to compute both approximate and exact designs with respect to several frequently used optimality criteria.