Activity Number:
|
300
- SPEED: Statistical Epidemiology
|
Type:
|
Contributed
|
Date/Time:
|
Tuesday, July 31, 2018 : 8:30 AM to 10:20 PM
|
Sponsor:
|
Section on Statistics in Epidemiology
|
Abstract #329884
|
Presentation
|
Title:
|
Comparison of Group Testing Algorithms for Clustered Data
|
Author(s):
|
Ana Best* and Paul S Albert and Yaakov Malinovsky
|
Companies:
|
NIH NCI DCEG Biostatistics Branch and National Cancer Institute and University of Maryland Baltimore County Dept. of Mathematics and Statistics
|
Keywords:
|
Group testing;
correlated data;
study design
|
Abstract:
|
Group testing study designs have been used since the 1930s for cost savings when estimating a small prevalence rate p or identifying individuals with prevalent disease under such a p. Recently, the group testing field has been revitalized by the development of new technologies and the discovery of new biomarkers. We examine the performance of finite-sample and dynamic Dorfman and Sterrett group-testing algorithms, as well as the optimal nested algorithm, in clustered-data settings under a Beta-Binomial model. We emphasize comparisons between algorithms when cluster-specific prevalences are unknown and must be estimated to determine group size. Additionally, we explore optimization of the number of cluster members used to estimate cluster prevalences. We illustrate the use of the best-performing algorithms with data from the National Health and Nutrition Examination Survey (NHANES).
|
Authors who are presenting talks have a * after their name.