SEMINAR DETAILS

Time:

2:00 PM - 3:00 PM

Place:

133 Rosenau Hall Auditorium

Speaker:

Sheng T. Luo

Affiliation

Department of Biostatistics
Johns Hopkins School of Public Health

Title & Abstract:

Smoking is the leading preventable cause of death in the U.S. A major problem when studying addiction behavior is that people typically make several quit attempts before they successfully quit. To describe the full stochastic nature of the smoking addiction pattern, I propose a discrete-time mixed effect model with three states: smoking, transient cessation (temporarily smoking-free, followed by a relapse), and permanent cessation (absorbent state), which is a latent state because of censoring. Rather than dichotomizing each individual as quitter or non-quitter, as is the common practice in epidemiology, I incorporate a “cure” component and estimate the cure probability, defined as the probability of permanent cessation given a quit attempt. Random subject-specific transition probabilities among these three states are used to account for the subject-to-subject heterogeneity.  

In the first part of the talk, I provide a computationally fast fitting algorithm using an innovative combination of geometric-like distributions of waiting times between addiction states and independent Beta distributions of subject-specific random effects. In the second part of the talk, I use a different modeling framework to address subject-specific predictions as well as potential correlation among random effects. The inference is conducted using a Bayesian framework via Markov Chain Monte Carlo (MCMC) simulation. Our methods are motivated by and applied to the Alpha-Tocopherol, Beta-Carotene (ATBC) Lung Cancer Prevention study, a large (29,133 participants) longitudinal cohort study of smokers from Finland. The modeling and inferential frameworks are applicable to other public health areas, e.g. chronic disease, behavioral disorders.

Contact :

Tania Osborn (919) 966-7268

Notes:

n/a