|
Abstract:
|
Leishmaniasis is a potentially deadly parasitic disease that places hundreds of millions of people and domesticated dogs at risk worldwide. Since dogs serve as the main animal reservoir for this disease, correctly identifying leishmaniasis in individuals is critical to public health efforts. While diagnostic tests are helpful, they are not always accurate. Currently, no gold standard for diagnosing canine visceral leishmaniasis exists, which hampers efforts to correctly identify cases. We propose a Bayesian latent class model to distinguish between diseased and healthy individuals, which incorporates dichotomized or continuous diagnostic test results, sensitivity and specificity of the tests, and individual level data. We compare the performance of our model to that of models that employ more traditional means of identifying those with symptomatic disease.
|
