12/9/2016

Post date: Jan 9, 2017 7:26:23 PM

Title: Modeling the temporal trend of between-biomarker association during HIV progression

Speaker: Zonghui Hu, Biostatistics Research Branch, NIAID, NIH

[Abstract]

In some HIV studies, researchers are interested in the temporal trends of the biomarkers that relate to disease progression. More importantly, they are interested in the association between two temporal trends, which may also change over time, reflecting the potentially different stages of disease. There are two challenges imposed by the biomarker data. First, the biological mechanisms of some markers are unclear and little information is known about their parametric forms. Second, there is big subject-to-subject variability and the data are sparse within each subject. Statistically, there has been little work on estimating the association between two temporal trends without specifying their parametric forms. In this work, we will adopt nonparametric mixed-effects modeling to estimate the temporal trend of association.