Hoberman, Steven. Response Adaptive Designs For Highly Successful Treatments, Randomness And Relationship Detection In Clinical Trials. Chapel Hill, NC: University of North Carolina at Chapel Hill Graduate School, 2014. https://doi.org/10.17615/f3by-j736
Hoberman, S. (2014). RESPONSE ADAPTIVE DESIGNS FOR HIGHLY SUCCESSFUL TREATMENTS, RANDOMNESS AND RELATIONSHIP DETECTION IN CLINICAL TRIALS. Chapel Hill, NC: University of North Carolina at Chapel Hill Graduate School. https://doi.org/10.17615/f3by-j736
Hoberman, Steven. 2014. Response Adaptive Designs For Highly Successful Treatments, Randomness And Relationship Detection In Clinical Trials. Chapel Hill, NC: University of North Carolina at Chapel Hill Graduate School. https://doi.org/10.17615/f3by-j736
Affiliation: Gillings School of Global Public Health, Department of Biostatistics
In the first part of our research we consider a problem of reducing the expected number of treatment failures (a binary response indicator) in trials where the probability of response to treatment is close to 1 and treatments are compared based on log odds ratio. We propose a new class of urn designs for randomization of patients in a clinical trial. The new urn designs target a number of allocation proportions including the allocation proportion that yields the same power as equal allocation but significantly less expected treatment failures. The new designs are compared with the doubly adaptively biased coin design, the efficient randomized adaptive design and with equal allocation. The properties of the new class of designs are studied by embedding them into a family of continuous time stochastic processes. In the second part of our research we study entropy as a measure of randomness in a clinical trial. For any randomization design we define a sequence of probability distributions. We then use this sequence to formulate and prove a statement about conditions for the asymptotic mean entropy of a randomization design to achieve its maximum value. We compare randomization designs and response adaptive randomization designs with respect to the asymptotic mean entropy. We derive a relationship between the limiting variance of distributions in the sequence to the mean entropy under a normality condition and apply this result to the doubly adaptive biased coin design. We develop two new methods of imputation for survival data that allow for application of the Brownian Distance Covariance. We use these two methods with survival data that have different levels of censoring, and different relationships that are not all easily detected by the Cox model. These methods are also compared to a relationship detection technique developed by Kwak for censored data.