Porthole
Mixed Location Scale Hidden Markov Model with An Application to Ecological Momentary Assessment Data (307837)
Donald Hedeker, University of Chicago*Xiaolei Lin, Fudan University
Robin J Mermelstein, University of Illinois at Chicago
Keywords: Intensive longitudinal data; Hidden Markov Model; Mixed Effects; latent state classi cation.
Hidden Markov Models (HMM) present an attractive analytical framework for capturing the state-switching process for autocorrelated data. These models have been extended to longitudinal data setting where simultaneous multiple processes are observed by including subject specific effects. However, HMMs for intensive longitudinal data, where each subject gets intensively measured over relatively short period of time, has not been widely studied. In this paper, we extend the Mixed Hidden Markov Models to include both random effects in the mean and within subject variance of the outcome to account for heterogeneity in both perspectives. We focus on the application of this model to intensive longitudinal studies in psychological and behavioral research where individual's latent states and state-switching over time are of interest. Models are estimated using Bayesian sampling methods in Stan. Advantages over Mixed HMMs and simple HMM are illustrated using a series of simulation studies. Finally, models are applied to an adolescent mood study and results show that distinct states classi ed by the proposed HMM are well separated in terms of multiple psychological and behavioral outcomes.