Adaptive Designs for Precision Medicine in Clinical Trials: A Review and Some Innovative Designs

Precision medicine proposes the customization of health-care tailored to a patient based on the individual characteristics. Adaptive randomization designs are effective ways to optimize patients’ treatments by incorporating individual information. We illustrate the mathematical framework for adaptive randomization and provide a brief review of popular adaptive designs that incorporate covariates under this framework. We also provide a brief overview on covatiate-adaptive design and covariate-adjusted response-adaptive design. Currently no design has been proposed in literature to incorporate discrete and continuous prognostic covariates, and predictive covariates simultaneously. To fill the vacant, this chapter proposes a new class of covariate-adjusted response-adaptive procedures to optimize the treatments based on individual covariates. The proposed procedure can balance both discrete and continuous covariates to make valid and credible comparisons among different treatments. Theoretical properties for the new designs are investigated in some simplified scenarios. Extensive simulation studies have been conducted, which demonstrate the superiorities of new procedure over existing designs.