|
Activity Number:
|
541
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 10, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #306488 |
|
Title:
|
Analysis of Repeated Measures Random Length Data
|
|
Author(s):
|
Ana-Maria Iosif*+ and Allan R. Sampson
|
|
Companies:
|
University of Pittsburgh and University of Pittsburgh
|
|
Address:
|
2717 Cathedral of Learning, Pittsburgh, PA, 15260,
|
|
Keywords:
|
random length data ; informative length ; longitudinal
|
|
Abstract:
|
Random length data occurs in experiments in which data on both the frequency of an event and its severity level are gathered and both are important. We model such data when subjects are measured repeatedly over time. In order to evaluate a treatment, one needs to jointly model the number of events and their correlated severity measures, as well as their relationship over time. For example, in a clinical trial of a new migraine drug, both the number of migraine headaches and their level of pain are recorded for each subject. If the drug is efficacious, the subjects in that treatment group are expected to improve: over time they will have a smaller number of headaches and lower pain levels. We model the vectors of severities with random lengths over time and provide methods to estimate the treatment effect.
|
