Modellen met Latente Variabelen in Diagnostische Geneeskunde

In the thesis, we assess the use of latent vari able models applied to diagnostic accuracy and dis ease incidence studies in infectious diseases rese arch. The latent class models are used to correct¨ for bias in accuracy estimates (sensitivity an spe cificity) when imperfect reference standards are u sed. In particular we focus on the following theme s:(1) interpretation, identifiability and estimati on of latent class models incorporating conditiona l dependence between tests, (2) the use of mixture ¨modeling to estimate infectious disease incidence ¨and prevalence, (3) the meta-analysis of diagnost ic accuracy studies, allowing for the use of imper fect reference standards at the primary study leve l. Models are estimated in a Bayesian framework us ing Markov Chain Monte Carlo methods.status: publishe