Weibull multi-state models with misclassification

ABSTRACTIt is often important to allow multi-state models (MSMs) to accommodate misclassification of states. We introduce Bayesian parametric MSMs with unknown misclassification of states and Weibull distributedwaiting times between states. This allows transitionsbetween statesto depend on the time spent in the current state, a feature lacking in commonly used exponential waiting times model. To fit the proposed model, a MCMC algorithm was employed. An example on the progression of bipolar disorderis presented along with simulation results. There was evidence that Weibull waiting times are an improvement over exponential in the study of bipolar disorder.