Stephanie Kliethermes
Department of Biostatistics, The University of Iowa
Growth patterns over time are routinely seen in medical studies. Being able to adequately model the growth requires following the development of individuals on one or more outcome variables over a period of time. Many of the current methods for modeling growth restrict the relationship between the outcome and time variables to a parametric relationship (e.g., linear, quadratic); however, these relationships don’t always accurately capture individual growth over time. Functional mixed effects (FME) models provide flexibility in handling longitudinal data with non-parametric temporal trends because they allow the data to determine the shape of the curve. Although these FME methods are well-developed for continuous outcome measures, which can often be assumed to be normally distributed, non-parametric methods for handling binomial outcomes are limited. We propose a Bayesian hierarchical functional mixed effects model to account for binomially distributed outcomes. Simulation studies show the usefulness of the binomial model particularly when outcomes occur on the boundary of the parameter space. Two relevant applications of this method involving (1) speech development and (2) word processing in cochlear implant patients will be discussed.
All statistics students are expected to attend.