Conference Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 116 - Epidemiological Models for Longitudinal Studies, Time-to-Event Outcomes, and Functional Data
Type: Contributed
Date/Time: Monday, August 8, 2022 : 8:30 AM to 10:20 AM
Sponsor: Section on Statistics in Epidemiology
Abstract #322147
Title: Efficient Estimation of Breakpoints in Piecewise-Linear Mixed-Effects Models for Longitudinal Ophthalmic Studies
Author(s): TingFang Lee* and Joel S. Schuman and Gadi Wollstein and MarĂ­a de los Angeles Ramos Cadena and Jiyuan Hu
Companies: NYU Grossman School of Medicine and NYU Grossman School of Medicine and NYU Grossman School of Medicine and NYU Grossman School of Medicine and NYU Grossman School of Medici
Keywords: mixed effects model; bootstrap; broken stick analysis; longitudinal studies; chronic disease; segmented model
Abstract:

The current approach used in ophthalmology to determine the breakpoints in broken stick analysis is limited to cross sectional analysis which does not consider complex dependencies such as repeated measurements and correlation between two eyes and does not accommodate outliers. We propose a robust method that integrates segmented mixed model and the least trimmed square technique. This method estimates unknown breakpoints which accounts for dependencies and outliers in the data. We compare our methods with three current approaches of broken stick analysis in longitudinal studies that are utilized in other research fields. A simulation study was conducted to evaluate the performance of each method and compare with the conventional approach. The proposed method improves the prediction accuracy while the conventional way fails to capture the breakpoints. We further applied these methods to a clinical longitudinal study with 216 eyes (145 subjects) followed for ~3.7 years and evaluate the association between visual field mean deviation (MD) and retinal nerve fiber layer thickness and between MD and cup to disc ratio.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2022 program