|
Activity Number:
|
272
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #309046 |
|
Title:
|
Effect of the Number of Top Genes on Survival Analysis in Microarray Data
|
|
Author(s):
|
Dung-Tsa Chen*+ and Michael Schell and Steven Eschrich and Alan Cantor and Timothy Yeatman
|
|
Companies:
|
Moffitt Cancer Center and Moffitt Cancer Center and Moffitt Cancer Center and Moffitt Cancer Center and Moffitt Cancer Center
|
|
Address:
|
University of South Florida, Tampa, FL, 33612,
|
|
Keywords:
|
number of top genes ; survival analysis ; microarray
|
|
Abstract:
|
Gene expression profiling in cancer prediction often involves determination of a set of top genes for analysis, it is important to evaluate how modeling varies with the number of top genes incorporated. Our preliminary analysis indicates there is considerable variation of prediction outcome when the number of top genes changes. The variation implies that use of a different number K of top genes is likely to yield unreliable results. We propose a predictive risk probability approach to attempt to accommodate for the variation. This approach identifies a range from K to L top genes. From each number of top genes, the analysis identifies each patient as having high risk (score = 1) or low risk (score = 0). The categorizations are then averaged (majority vote), giving a high risk score between 0 and 1, thus providing a ranking for the patient's need for further treatment.
|
