Skip Internal Navigation
The Biostatistics & Bioinformatics Branch (BBB) conducts research in statistical theory and methodology relevant to problems under investigation in the areas of maternal and child health. The Branch also provides statistical support in the design and conduct of clinical trials and consults and collaborates with intramural and extramural scientists on statistical and mathematical problems. The Branch also develops quantitative procedures appropriate for application in biomedical and life sciences and supports the statistical and applied mathematics activities of DESPR.
Featured
- BBB investigators proposed a biologically valid discrete survival model that combines both the survival and hierarchical models allowing investigators to obtain the distribution of time-to-pregnancy and day-specific probabilities during the fertile window in a single model. The model allows for the consideration of covariate effects at both the cycle and daily level while accounting for daily variation in conception. (Sundaram et al., Biostatistics, 2012). PMID: 21697247
- BBB investigators derived optimality properties of group testing procedures for estimating prevalence of a rare disease whose status is classified with error. Exact ranges of disease prevalence are obtained for which group testing provides more efficient estimation when group size increases. (Liu et al., Biometrika, 2012).
- BBB investigators proposed a latent class model for associating high dimensional environmental exposure data with disease prevalence. This new innovative methodological approach is applied to an epidemiological study that examined the relationship between environmental polychlorinated biphenyl (PCB) exposure and the risk of endometriosis. (Zhang et al, Biostatistics, 2012). PMID: 21908867
- A BBB investigator examined a linear mixed modeling approach for predicting a binary event from longitudinal data under a simple to implement Gaussian random effects assumption. It was shown that overall measures of diagnostic accuracy (area under the curve) were insensitive to even large departures from this assumption. (Albert, Statistics in Medicine, 2012). PMID: 22081439
- Researchers developed a joint model for batched or pooled longitudinal data subject to informative dropout. Both a shared random parameter and a pattern mixture model formulation are considered for estimation with the pattern mixture model being shown to be more robust to modeling assumptions, while the shared random parameter model has a more direct interpretation. (Albert and Shih, in press, Statistical Methods in Medical Research, 2012). PMID: 21300625