|
Activity Number:
|
104
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, July 30, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #310205 |
|
Title:
|
Analysis of Longitudinally Measured Multivariate Biomarkers and Left-Censored Data
|
|
Author(s):
|
Ghideon S. Ghebregiorgis*+ and Lisa Weissfeld
|
|
Companies:
|
University of Pittsburgh and University of Pittsburgh
|
|
Address:
|
2717 Cathedral of Learning, Pittsburgh, PA, 15260,
|
|
Keywords:
|
multivariate mixed model ; Biomarkers ; Longitudinal analysis ; Left Censoring
|
|
Abstract:
|
Our objective is to analyze longitudinally measured multivariate biomarkers obtained from a large cohort study of patients who enrolled in a study examining markers for the development of sepsis in a cohort of subjects with community acquired pneumonia. For continuous outcomes, estimation is based on a parametric multivariate linear mixed model for the longitudinally measured biomarkers (Thiebaut et. al. 2005).For bivariate left censored biomarkers, such as interleukin-6(IL-6)and interleukin-10(IL-10), an extension of this method based on the Tobit regression model will be used for the inference. The multivariate modeling approach is evaluated and compared with the univariate approaches through empirical findings and simulation studies. The simulation study will be used to measure the sensitivity of the performed analysis by considering various variance-covariances structures of the biomarkers
|
