ABSTRACT
This chapter presents nonparametric optimal benchmark designs for adaptive dose-finding studies. Cheung applied this benchmark to other complex dose-finding problems, such as those accounting for different types and grades of toxicities. The performance of the optimal design can be evaluated over many simulation runs and can be used as a benchmark for efficiency. As an illustration of how the optimal can be used to assess design performance, the chapter presents simulation results in order to get an idea of how well the two-stage, likelihood-based, continual reassessment method (CRM-L) is performing when compared to the optimal design. The simulation results assess how well the CRM-L and the nonparametric optimal design are performing in two ways. One is by simply observing the distribution of dose selection as the maximum tolerated dose (MTD). The other is measured by the accuracy index. The accuracy index for the CRM-L designs yielded a value of 0.829, compared to 0.842 for the optimal method.
