Comment on 'Hidden Markov models for zero-inflated Poisson counts with an application to substance use' by S. M. DeSantis and D. Bandyopadhyay

摘要

DeSantis and Bandyopadhyay recently proposed an interesting variation of the standard hidden Markov model (HMM) in their paper 'Hidden Markov models for zero-inflated Poisson counts with an application to substance use' [1]. Their zero-inflated Poisson HMM (ZIP-HMM) model is in fact a three-state HMM with the conditional mean response associated with one of these states defined to be zero. However, by framing the model in the ZIP context, these authors provide a useful way of understanding processes where the response is likely to be a structural zero over certain time periods.