Applied statistics workshop on "Bayesian Optimal Interval Design (BOIN) in phase I dose escalation clinical trials" by Alexandre Lambert (Qubes)

Abstract:
In most phase I clinical trials, one of the objectives is to find the maximum tolerated dose  of a new drug. A dose escalation approach is typically followed with dose escalation decisions based on observed toxicities. Many algorithms have been proposed in the literature  including the well-known  traditional “3+3” dose escalation design. Given it simplicity, this design is widely used in oncology but has poor operational characteristics.

In this presentation, we will present the BOIN design (Liu and Yuan, 2015) which uses a model assisted dose escalation scheme. It has been shown  to be more flexible and more efficient than the “3+3” design while maintaining simplicity in its implementation.