Abstract
Clinical trials are typically run in order to understand the effects of a newtreatment on a given population of patients. However, patients in largepopulations rarely respond the same way to the same treatment. Thisheterogeneity in patient responses necessitates trials that investigate effectson multiple subpopulations - especially when a treatment has marginal or nobenefit for the overall population but might have significant benefit for aparticular subpopulation. Motivated by this need, we propose Syntax, anexploratory trial design that identifies subpopulations with positive treatmenteffect among many subpopulations. Syntax is sample efficient as it (i) recruitsand allocates patients adaptively and (ii) estimates treatment effects byforming synthetic controls for each subpopulation that combines control samplesfrom other subpopulations. We validate the performance of Syntax and provideinsights into when it might have an advantage over conventional trial designsthrough experiments.