|
Activity Number:
|
528
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 5, 2009 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #303755 |
|
Title:
|
A New Approach to Model Region Treatment Effects in Clinical Trials
|
|
Author(s):
|
Hua Guo*+
|
|
Companies:
|
Merck Research Laboratories
|
|
Address:
|
126 E. Lincoln Ave, Rahway, NJ, 07065,
|
|
Keywords:
|
Bayesian ; Topology ; Brownian Motion Process ; Treatment Effects
|
|
Abstract:
|
It is interesting to get overall treatment effect and individual region treatment effect simultaneously. Current subgroup analysis performed on region assumes homogeneous variance across different regions and ignores the relationship between regions. We use a tree structure with topology tau and branches T as prior information to pool strength from different regions. Topology tau accounts for correlation between different regions. By varying the lengths of branches, we account for different variation from different regions. Treatment effects follow a Brownian diffusion process along the tree describing the relationship among the regions. This model more accurately estimates subgroup effects for different regions; provides overall treatment effects and individual treatment effects simultaneously. The model can be extended to other subgroup factors besides regions.
|
