Title: Maximin optimal designs for cluster randomized trials with binary outcomes and related applications
Authors: Weng Kee Wong - UCLA (United States) [presenting]
Abstract: A nature-inspired metaheuristic algorithm is developed to find extended two-stage adaptive optimal designs for phase II trials with many parameters which is called discrete particle swarm optimization (DPSO). These designs include previous ones as special cases. We show that DPSO not only frequently outperforms greedy algorithms, which are currently used to find such designs when there are only a few parameters; it is also capable of effectively solving adaptive design problems with many parameters that greedy algorithms cannot. In particular, we consider situations where a treatment seems promising in stage 1 but there is great uncertainty in its efficacy rate, and both drug development cost and ethics dictate that there be three pre-determined user-specified efficiency rates for possible testing at stage 2 given testing error rate constraints. We provide a real application and demonstrate benefits of our proposed design strategy for a Phase II trial.