|
Activity Number:
|
381
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 11, 2004 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #301827 |
|
Title:
|
Applications of Predictive Probabilities in Phase II Cancer Clinical Trials
|
|
Author(s):
|
Diane D. Liu*+ and J. Jack Lee
|
|
Companies:
|
University of Texas M.D. Anderson Cancer Center and University of Texas M.D. Anderson Cancer Center
|
|
Address:
|
1515 Holcombe Blvd., Unit 447, Houston, TX, 77030,
|
|
Keywords:
|
Bayesian methods ; flexible monitoring scheme ; early termination ; efficiency ; robustness
|
|
Abstract:
|
A Bayesian predictive probability approach is proposed as an alternative to the commonly used two- or three-stage designs in phase II cancer clinical trials. The new design allows flexible and more frequent monitoring of the trial outcomes. Early termination of the trial can be reached when the interim data indicate that the experimental regimen is not promising. Even with more frequent monitoring, the predictive probability approach can still possess good frequentist's properties by controlling Type I and Type II errors. A bidimensional search algorithm is implemented to determine the design parameters. Exact computation and simulation studies demonstrate that the predictive probability approach is more efficient than the traditional multistage designs. The predictive probability design not only is more adaptable in evaluating the study outcome but also remains robust even when deviations occur in monitoring cohorts of patients as specified in the trial design. Examples will be given to illustrate the statistical properties of various design settings.
|
