|
Activity Number:
|
322
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #307313 |
|
Title:
|
Comparison of Bayesian and Frequentist Adaptive Designs for Combination Proof-of-Concept/Dose Response Modeling Trials in Phase II
|
|
Author(s):
|
Nitin Patel*+ and James Bolognese and Jerald Schindler and Scott Berry and Yannis Jemiai and Vipul Suru
|
|
Companies:
|
Cytel Inc. and Merck Research Laboratories and Cytel Inc. and Berry Consultants and Cytel Inc. and Cytel Inc.
|
|
Address:
|
675 Massachusetts Ave., Cambridge, MA, 02139,
|
|
Keywords:
|
|
|
Abstract:
|
Traditional designs for phase II proof-of-concept trials are inefficient in allocating doses to patients, as they often assign subjects to doses not of interest. Dose-adaptive designs use accumulating information on responses to determine sequentially which dose should be allocated next. Such adaptive designs efficiently provide information to assess efficacy and characterize dose response. Various frequentist and Bayesian adaptive designs have been proposed in the literature. Direct comparison between these methods is difficult, as no simple solutions to the power/sample size calculations are available. We will present results from a methodical comparison of the performance of these designs using computer simulation experiments. Our objective is to provide insights to guide the deployment of adaptive designs in practical settings.
|
