Presenter: Qirui Hou

Supervisory Committee: Clement Ma (Supervisor), Amy Liu, Peter Szatmari

Chair: Tony Panzarella

Date and Time: Thursday, August 22, 2024, 12:00-14:00pm EST

Location: 155 College Street, Health Sciences Building, Room 734

Abstract: Basket designs and Multi-Arm Multi-Stage (MAMS) trials enhance efficiency in evaluating multiple treatments across various indications. Chen et al. (2016) proposed a 2-arm 2-stage randomized bassket design, testing l=1 treatment vs. control. We expand this to evaluate multiple treatments (l>1) vs. a common control. This multi-arm, 2-stage design is broadly applicable and features rigorous scientific and statistical design. We extended Chen’s statistical calculations and used simulations to evaluate type I error rate, power and sample size. We applied the Dunnett test to prune inactive indications in the interim analysis and pool active indications in the final analysis. We compared Chen’s 2-arm design with our multi-arm design, validating the accuracy of our simulation approach and effective type I error control. The proposed design maintained stable power with smaller sample sizes compared to single treatment trials, providing a foundation for broader analytical applications. This design offers an efficient method to evaluate multiple therapeutic approaches within a single trial, enhancing the identification of effective treatments across various indications.