24 – Seasonal Adjustment at Statistical Agencies
Continual Reassessment Method with Bayesian Variable Selection in Phase I Clinical Trails
Zhenyu Zhao
Northwestern University
This study presents a variable selection procedure in Bayesian adaptive design of Phase I clinical trials. For the Phase I dose-finding problem, besides the dose level, the dose-limiting toxicity (DLT) may depend on other covariates. The variable selection procedure presented is designed to involve the proper covariates in the model and use the selected model to predict the maximum tolerated dose (MTD). This variable selection procedure can be applied in a large range of Bayesian adaptive designs.