Xiaoqiong Joan Hu, Ph.D.
Department of Math Sciences
University of Memphis
Wednesday, March 12, 2003
3:30 PM
Moos 2-650 (Please note, this is a different room from our other candidate talks)
Minneapolis Campus
Abstract:
Formal interim analyses of clinical trials based on prespecified termination criteria (stopping rules) are commonly used to monitor patient safety and the evolving efficacy of treatment groups. In some circumstances, however, relatively little may be known about the nature and magnitude of the expected treatment effect or the clinical significance of certain types of differences in the outcome measure. When the response is a stochastic process, such as repeated measures overtime of a biological marker or a failure time process, there may be little confidence in the type of magnitude of treatment differences that will occur and hence no obvious metric for comparing groups. An approach motivated by these considerations will be presented in the talk. The approach is based on the use of repeated confidence bands (RCB) for the mean function of the response process. We will present the general principle underlying the construction of RCB and an algorithm for their computation. We will also describe how RCB can be adapted to cope with situations where domain of concern varies or a test of hypotheses is of interest. Examples involving HIV/AIDS clinical trials will be used for illustration.
A social tea will be held at 3:00 P.M. in A434 Mayo. All are Welcome.