|
Activity Number:
|
73
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, August 2, 2009 : 4:00 PM to 5:50 PM
|
|
Sponsor:
|
Section on Risk Analysis
|
| Abstract - #304357 |
|
Title:
|
Bayesian Approaches for Simultaneous Estimation of BMD with Multiple Chemicals and Multiple Endpoints
|
|
Author(s):
|
Epiphanie Nyirabahizi*+ and Edward L. Boone and Chris Gennings
|
|
Companies:
|
Virginia Commonwealth University and Virginia Commonwealth University and Virginia Commonwealth University
|
|
Address:
|
730 East Broad Street, Theater Row, Room 3039C, Richmond, VA, 23298-0032,
|
|
Keywords:
|
Bayesian estimation of BMD ; Multiple endpoints data ; Dose response studies ; Bayesian model averaging ; MCMC ; Bayes' factors
|
|
Abstract:
|
Determining Benchmark Dosages (BMD) is of interest to toxicologists and risk assessors. Often data come from experiments with multiple chemicals and endpoints. Current methodology evaluates each chemical and endpoint separately resulting in multiple statistical tests consequently inflating Type I error rates. Methods to adjust for multiple comparisons are subject to reducing the power to detect effects of interest. Use of composite scores alleviate the concern of multiple testing for each endpoint but may not account for correlations that might exist between variables. In this presentation we introduce a Bayesian approach for both multiple endpoints and multiple chemical data into a single unified model. The model is estimated using MCMC in WinBUGS and R. Inferences on the endpoint model and chemical effects are done via Bayes' factors and Bayesian model averaging.
|
