A Multi-Stage Stochastic Model in the Analysis of Longitudinal Data

Abstract

Multi-stage transition model is playing an important role in dementia studies. Since death is a significant source of missing data in longitudinal epidemiological studies on elderly individuals, we consider four stages: normality, memory-impaired intermediate, dementia and death without dementia. To analyze longitudinal data, we develop the likelihood function based on a first order Markov chain model consisting of transitional probabilities between stages. Different from the typical illness-death model, we construct a reversible transition model between normality and memory-impaired intermediate. We use Kolmogorov’s backward equations to derive the probability of transition and ordinal logistic regression to investigate what covariates have significant influence on the transition.

Date
May 12, 2018 16:00
Event
Conference on Bayesian Modeling, Computation and Applications 2018
Location
Oak Hall, University of Connecticut
363-367 Fairfield Way, Storrs, CT, 06279, United States
Avatar
Qi Qi, Ph.D.
Statistical Scientist

My research interests include Survival Analysis, Bayesian Methods, Longitudinal Data Analysis, Joint Modeling, Stochastic Models, Data Visualization, Machine Learning, Data Mining, Statistical Computing.