|
Activity Number:
|
371
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 4, 2009 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #304495 |
|
Title:
|
A B-Spline--Based Semiparametric Nonlinear Mixed Effects Model
|
|
Author(s):
|
Angelo Elmi*+ and Sarah Ratcliffe and Wensheng Guo and Samuel Parry
|
|
Companies:
|
University of Pennsylvania and University of Pennsylvania and University of Pennsylvania and University of Pennsylvania
|
|
Address:
|
423 Guardian Drive, Philadelphia, PA, 19104,
|
|
Keywords:
|
Smoothing ; Adaptive Gaussian Quadrature ; Curve Registration ; Mixed Effects Models
|
|
Abstract:
|
Semiparametric nonlinear mixed effects models are used to model longitudinal data via an a-priori unspecified common shape function estimated from the data. I will present a new method for fitting this model, with a B-spline basis expansion used to estimate the common shape function. Existing approaches which use smoothing splines result in model parameters being estimated via a backfitting approach that iterates between two mixed effects models. Instead, our method allows for a unified likelihood requiring only one mixed effects model for estimation which guarantees convergence and allows for valid likelihood-based inferences and model selection as well as more accurate numerical integration based on Adaptive Gaussian Quadrature. The model is applied to a Women's Health data set to compare between the average labor curves of cases of uterine rupture to healthy controls.
|
