|
Activity Number:
|
132
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 3, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #304353 |
|
Title:
|
Bayesian Analysis and Classification of Two Quantitative Diagnostic Tests with False Negatives and No Gold Standard
|
|
Author(s):
|
Jingyang Zhang*+ and Kathryn Chaloner and Jack Stapleton
|
|
Companies:
|
The University of Iowa and The University of Iowa and The University of Iowa
|
|
Address:
|
#108 E177 General Hospital, Iowa City, IA, 52242,
|
|
Keywords:
|
Bayesian statistics ; Mixture Model ; Markov Chain Monte Carlo ; Metropolis-within-Gibbs Sampling ; Statistical Decision Theory
|
|
Abstract:
|
Two diagnostic tests are used to detect antibodies to the GB Virus type C (GBV-C) envelope glycoprotein E2. There is no reference test (gold standard). Each test looks for a characteristic that is typically present if the antibodies are present, but is sometimes absent. A model is developed reflecting the presence of false negatives due to the absence of either characteristic independently. A mixture of four bivariate normal distributions is used along with a prior distribution and a Bayesian analysis. Test results from 100 subjects are analyzed using the Metropolis-within-Gibbs sampler. Subjects are classified by a statistical decision rule, and the classification appears to separate the subjects well into those that are positive and negative for GBV-C antibodies.
|
