|
Activity Number:
|
359
|
|
Type:
|
Invited
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #307775 |
|
Title:
|
Generalized Linear Mixed Model Analysis of Multistate Sleep Transition Data: The Sleep Heart Health Study
|
|
Author(s):
|
Brian S. Caffo*+ and Bruce Swihart and Naresh Punjabi
|
|
Companies:
|
Johns Hopkins Bloomberg School of Public Health and Johns Hopkins University and Johns Hopkins Bayview Medical Center
|
|
Address:
|
Dept. of Biostatistics, Baltimore, MD, 21205,
|
|
Keywords:
|
GLMM ; sleep ; transition data ; Poisson log-linear models ; proportional hazards ; observational data
|
|
Abstract:
|
Sleep is a complex behavioral state that has been shown to be essential for maintaining physical and mental function. Recent research has shown that traditional epidemiologic summaries, such as the respiratory disturbance index and summaries of sleep architecture, fail to capture important aspects of sleep. In particular, the rate of transition between sleep states such as rapid eye movement (REM) sleep, non-REM sleep and wake have been shown to have predictive value that these other summaries lack; further refinements consider stages of non-REM sleep. A flexible Poisson generalized linear mixed model is proposed to analyze sleep transition data. This model is shown to synthesize two common complimentary methods for analyzing sleep transition data: multi-state Cox proportional hazards models and log-linear models on marginal sleep transition contingency tables.
|
