Bayesian inference for discretely observed continuous time multi-state models

TL;DR

This paper develops a Bayesian inference method for semi-Markov and inhomogeneous Markov multi-state models observed at discrete times, using a Metropolis-Hastings algorithm with nested Markov proposals, demonstrated through simulations and real data.

Contribution

It introduces a novel Bayesian inference approach for complex multi-state models with intractable likelihoods, utilizing trajectory reconstruction and nested Markov proposals.

Findings

Effective inference demonstrated via simulation studies.

Successful application to benchmark multi-state data.

Method improves inference accuracy for semi-Markov and inhomogeneous models.

Abstract

Multi-state models are frequently applied for representing processes evolving through a discrete set of state. Important classes of multi-state models arise when transitions between states may depend on the time since entry into the current state or on the time elapsed from the starting of the process. The former models are called semi-Markov while the latter are known as inhomogeneous Markov models. Inference for both the models presents computational difficulties when the process is only observed at discrete time points with no additional information about the state transitions. Indeed, in both the cases, the likelihood function is not available in closed form. In order to obtain Bayesian inference under these two classes of models we reconstruct the whole unobserved trajectories conditioned on the observed points via a Metropolis-Hastings algorithm. As proposal density we use that given by the nested Markov models whose conditioned trajectories can be easily drawn by the uniformization technique. The resulting inference is illustrated via simulation studies and the analysis of two benchmark data sets for multi state models.