Know what you don't know: Embracing state uncertainty in disease‐structured multistate models

Hidden Markov models (HMMs) are broadly applicable hierarchical that derive their utility from separating state processes observation yielding the data. Multistate such as mark–recapture and dynamic multistate occupancy HMMs frequently used in ecology. In early formulations, states, pathogen infection status, were assumed to be perfectly observed without ambiguity. However, uncertainty is a pervasive feature of many ecological studies, multievent developed explicitly account for it. We novel extended model incorporates at multiple levels detection. Using disease-structured example, both false negative positive assignment errors modelled two assignment—the sampling process diagnostic samples subjected to. additionally describe methods jointly intensity integrate heterogeneity parameters, mortality dynamics, detection processes. provide code simulate analyse datasets with various underlying fit our dataset Mixophyes fleayi (Fleay's barred frog) infected amphibian chytrid fungus (Batrachochytrium dendrobatidis, Bd). case study, we found evidence errors: protocol performed poorly detecting Bd, was highly dependent on positives non-negligible. Incorporating yielded significantly higher estimates prevalence 4–5 times lower rates transitions compared those obtained traditional model. Our results highlight incorporating improves inference process, especially when sensitivity specificity low. The general structure can applied other HMMs, providing foundation modelling related models. For models, recommend conducting robust design surveys collecting during each capture event facilitate errors.