Activity Number:
|
22
|
Type:
|
Contributed
|
Date/Time:
|
Sunday, August 11, 2002 : 2:00 PM to 3:50 PM
|
Sponsor:
|
Biometrics Section*
|
Abstract - #301571 |
Title:
|
A Proportional Elasticity Model for a ROC Curve
|
Author(s):
|
Grant Izmirlian*+
|
Affiliation(s):
|
National Cancer Institute
|
Address:
|
EPNorth, STE 3131; 6130 Executive Plaza Blvd, Bethesda, Maryland, 20892-7354, USA
|
Keywords:
|
ROC curve ; semi-parametric ; optimal power
|
Abstract:
|
When evaluating a new continuous biomarker for cancer, a useful tool in measuring its capabilities at distinguishing cases from controls is the ROC curve. The ROC curve is a plot of the TPF versus the FPF. Recent literature on ROC curve models suggests that semi-parametric models should be used, and that the model should be based upon the ROC curve itself, as opposed to an ROC curve derived from biomarker models in the case and control populations. A new class of semiparametric models, based upon the percent gain in TPF at a given FPF, is proposed and demonstrated to have optimal power against alternatives of specific types. Applications to cancer screening will be presented.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2002 program |