ABSTRACT
In this chapter, we illustrate Bayesian models and methods for two commonly encountered tasks in medicine: (i) evaluating the performance of diagnostic tests and (ii) the use of diagnostic test data to estimate disease prevalence. We define a diagnostic test as any instrument that provides information about the presence or absence of some condition of interest, referred to as the “disease.” Testing accuracy is assessed by estimating the sensitivity and specificity. In this chapter, all methods are based on test data with dichotomous outcomes. The material presented is taken largely from work that is catalogued at the website www.epi.ucdavis/diagnostictests/, which contains papers and WinBUGS code that apply to this chapter and beyond.
