Bayesian Latent Class Models in Veterinary and Human Epidemiology

Latent class models have been widely used to evaluate the performanceof veterinary and human diagnostic tests, in the absence of a gold standard. In thiswork, we explore Bayesian latent class models, with and without restrictions, in different populations. We previously reported the Bayesian latent class analysis of themalaria dataset (n =3317) and now we apply a similar approach to a smaller datasetin the context of canine dirofilariasis (n =308). Although the last study presentsa small sample size, it was important to obtain the first estimates of the prevalence(and performance measures of diagnostic techniques) of canine dirofilariasis in threedistricts of Portugal, taking into account the relationship with human dirofilariasis.