|
Activity Number:
|
287
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #308248 |
|
Title:
|
Lipid Analysis for HIV Patients: Does Fasting Bias Matter?
|
|
Author(s):
|
Xiting Yang*+ and Joseph B. Kadane and Heidi M. Crane
|
|
Companies:
|
Carnegie Mellon University and Carnegie Mellon University and University of Washington
|
|
Address:
|
Department of Statistics, Pittsburgh, PA, 15213,
|
|
Keywords:
|
Bayesian ; Hierarchical ; Lipid measurement ; HIV
|
|
Abstract:
|
This study applies a Bayesian method for investigating the effects of uncertainty of fasting time on lipid measurement for HIV-infected patients and examines whether or not repeated measurement with eight hours fasting is necessary. A hierarchical structure is used to model the effects of fasting, with the posterior distributions calculated using Gibbs sampling and Metropolis-Hasting sampling. The necessity of repeated measurement with mandatory fasting is summarized by checking the distribution of the posterior for the hypothetical fasting lipid measurements for different observed measurement ranges.
|
