12/8/2017

Post date: Jan 16, 2018 4:56:14 PM

Title: Individualized prediction and subgroup identification from longitudinal biomarker data: applications to fetal growth

Speaker: Jared Foster, NCI, NIH

Abstract:

Longitudinal monitoring of biomarkers is often helpful for predicting disease or a poor clinical outcome. Typically, these longitudinal predictors are evaluated across an entire population; however, it is also possible that this prediction is only accurate in one or more subgroups of the population. For example, recent work suggests that accurate prediction of large-for-gestational-age birth (LGA) from ultrasounds taken late in pregnancy is possible, but that this prediction is poor when only early ultrasound measurements are used. To this end, I will discuss the identification of subgroups of women for whom early prediction is more accurate, should they exist. In particular, I will describe a tree-based approach, which extends the classification and regression tree (CART) methodology to a longitudinal classification setting, and simultaneously controls the risk of false discovery of subgroups. I will focus on the application of the proposed methods to data from the NICHD Scandinavian Fetal Growth Study, and will also briefly present some limited simulation results. Time permitting, I will also briefly discuss some potential extensions to oncology.