|
Activity Number:
|
606
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 6, 2009 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #305421 |
|
Title:
|
Nonparametric Empirical Likelihood Estimation of AUC and Partial AUC for Test-Result-Dependent Sampling Study
|
|
Author(s):
|
Ma Junling*+ and Wang Xiaofei
|
|
Companies:
|
Duke University Medical Center and Duke University Medical Center
|
|
Address:
|
2424 Erwin Road, Box 2721, Department of Biostatistics and Bioinformatics, Durham, NC, 27710,
|
|
Keywords:
|
Area under Receiver Operating Characteristic Curve (AUC) ; Partial AUC ; Test-result-dependent sampling ; Nonparametric empirical likelihood
|
|
Abstract:
|
Area under ROC curve (AUC) and partial AUC are two important summary measures for assessing the prediction accuracy of a biomarker relative to true disease status. We consider nonparametric inference for AUC and partial AUC under a biased sampling scheme in which a random sample is combined with a test-result-dependent sample. The utility of such biased sampling schemes can be found in the situations where the measurement of the biomarker and the ascertainment of the true disease status are expensive/time-consuming or where investigators are interested in the performance of the biomarker in certain ranges. We develop nonparametric empirical likelihood estimators for AUC and partial AUC under the biased sampling scheme. We establish asymptotic properties for these estimators. Simulation shows that the proposed estimators have good finite sample properties and outperforms naïve methods.
|
