|
|
A Bayesian model is introduced for estimation of disease prevalence with imperfect diagnostic tests. A latent
variable approach leads to an easy-to-implement Gibbs sampling scheme for sampling from the joint posterior
distribution of prevalence and the sensitivity and specificity of the tests. These models have been widely
adopted for use with human medical data, and we apply them here for prevalence estimation for diseases in
fish from the northwest United States. In this setting where less information is available about sensitivity
and specificity of tests, some interesting differences occur in the behavior of the Gibbs sampler and in the
conclusions that result from the analyses.
|
|