Virtual seminar
Time: 12/5/2025, Friday 1:45-3:00pm
Zoom: Https://jh.zoom.us/j/95621579066
Speaker: Lu Xia, Assistant Professor, Department of Statistics and Probability, Michigan State University
Talk title: Estimating Equation-Based Approaches for Longitudinal Data: High-Dimensional Inference and Transfer Learning
[Abstract] Regression analysis of longitudinal data, where correlated responses from multiple time points are measured, is ubiquitous in many scientific areas such as biology, medicine and sociology. New challenges are emerging with the availability of complex longitudinal data in this big data era. Generalized estimating equations (GEE) are widely used to analyze longitudinal data, providing robust parameter estimates without requiring correct specification of the full likelihood. The first part of this talk will introduce a projected estimating equation approach to reliably drawing inference for linear functionals of regression parameters in GEE in high-dimensional settings, where the number of covariates may exceed the sample size. The second part of this talk will focus on a transfer learning approach for GEE that facilitates parameter estimation and prediction in the target population when external models from larger cohorts are readily available. We provide theoretical guarantees of the proposed methods to enhance the understanding of their behaviors, and demonstrate their superior performance with numerical studies.
Upcoming events
2/27/2026, 3/6/2026
Yanxun Xu, Associate Professor, Johns Hopkins University
Short Course: ``Adapting Generative AI for Task Automation in Biomedical Research''