|
Title
|
Room
|
|
Methods for Multivariate Longitudinal Data
|
H-Grand Salon C
|
|
Date / Time
|
Sponsor
|
Type
|
|
08/06/2001
10:30 AM
-
12:20 PM
|
Biometrics Section*, ENAR
|
Invited
|
|
Organizer:
|
Joel Dubin, Yale University
|
|
Chair:
|
Xihong Lin, University of Michigan School of Public Health
|
|
Discussant:
|
|
|
Floor Discussion
|
12:15 PM
|
Description
Longitudinal methods are prevelant throughout the biostatistical literature. Generally, the approaches focus on a single univariate longitudinal response, e.g., monitoring growth for children or systolic blood pressure for elder populations. Recently, however, several approaches have been suggested for tackling multivariate longitudinal data; that is, a multivariate response followed over time. Some examples of this data include following both systolic and diastolic blood pressure for individuals over time or tracking hemodialysis patients for several acute phase blood proteins over time. One needs to be careful to account for various sources of correlation, including dependence between the multivariate responses at a particular time as well as between time points for a particular response. The speakers in this session will discuss a variety of ways to handle multivariate longitudinal data problems, taking into consideration both discrete and continuous responses.
|
|
