There exists an undeniable need for biologists and medical researchers to identify new biomarkers (biological markers) which are useful in determining exposure and/or for the purposes of disease detection. Consider- ing the rising cost of assessing new biomarkers, it is reasonable to believe that this area of research has the potential to be stunted due to funding concerns. My proposal aims at offering these researchers an alternate data collection mechanism and ?exible statistical tools to more cost-effectively identify meaningful biomarkers. Traditionally, biomarkers are evaluated by collecting specimens from individuals within target populations which are subsequently tested. Alternatively, the proposed data collection mechanism speci?es that several individual specimens are mixed to form a pool which is then tested. The salient feature of pooling is twofold: (1) one assay measures several individuals' biomarker levels, which reduces costs; (2) mixing several individual specimens provides a suf?cient amount of the biomarker to avoid the limit of detection. The statistical literature surrounding the use of pooling for biomarker assessment predominantly focuses on the estimation of the receiver operating characteristic (ROC) curve and the area under the curve. Regretfully, these works assume restrictive assumptions about the underlying distribution of the individual biomarker levels. If such assumptions are wrong, incorrect inference could be produced. To remedy this limitation, I propose ?exible methods that are robust to those assumptions. Speci?cally, I will propose a nonparametric estimator of the ROC curve based on pooled measurements, which will then be used to test whether the biomarker is a good discriminator (Speci?c Aim 1). Then I will extend the nonparametric method to to estimate a covariate- adjusted ROC curve from pooled measurements (Speci?c Aim 2). The third speci?c aim will ?nd the optimal linear combination of multiple biomarkers in order to improve the discriminatory ability. In addition, I will develop statistical software that easily implements the proposed methods. My proposal is innovative because it offers methods that could shift the current biomarker research paradigm into a more comprehensive and cost-effective regime. The successful completion of my proposal could signi?cantly impact biological and medical research practices, by allowing for the assessment of biomarkers that previously could not be appropriately evaluated due to cost considerations and the lack of statistically sound methodology.

Public Health Relevance

This proposal will promote the use of pooling as a cost-effective data collection mechanism in biomarker evalua- tion. Robust and ?exible statistical methodology for evaluating biomarkers, based on pooled measurements, will be developed along with statistical software that implements the same. This work will impact public health com- munity by allowing researchers to more ef?ciently and thoroughly investigate new biomarkers for the purposes of disease and exposure detection, when the budget and/or the number of specimens is limited.

National Institute of Health (NIH)
National Institute of Allergy and Infectious Diseases (NIAID)
Small Research Grants (R03)
Project #
Application #
Study Section
Biostatistical Methods and Research Design Study Section (BMRD)
Program Officer
Gezmu, Misrak
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
University of South Carolina at Columbia
Biostatistics & Other Math Sci
Schools of Arts and Sciences
United States
Zip Code