Statistics Seminar - Spring 2026

Seminars are held on Thursdays from 1:00 - 2:00pm on Zoom unless otherwise noted. For access information, please contact the Math Department.

For questions about the seminar schedule, please contact Chenlu Shi and Yan Sun

*******************************************************************************************************

February 26

Dr. Chi-Kuang Yeh, Georgia State University

Single and multi-objective optimal designs for group testing experiments

Group testing, or pooled-sample testing, is widely used in large-scale screening and resource-constrained studies, yet principled design methodology for precise parameter estimation remains limited. This talk presents an optimal design framework for group testing that targets efficient estimation of key model parameters while accounting for cost constraints. Beyond classical criteria such as D-, D_s-, and A-optimality, a central novelty is the introduction of maximin design principles, including potentially non-differentiable criteria, into group testing procedures. These nondifferentiable criteria have not previously been explored in this context and yield designs with strong worst-case guarantees and improved robustness. The framework accommodates both large-sample settings through optimal approximate designs and small-sample studies through exact optimal designs, enabling systematic assessment of robustness to changes in criteria, statistical models, and cost structures. We demonstrate the practical impact of this approach through an application to Chlamydia screening using imperfect assays under budget constraints, and show that precise parameter estimation via optimal design is a foundational step that directly enables efficient and reliable sequential group-testing procedures.

This talk will be mainly based on the paper that is in revision, https://arxiv.org/abs/2508.08445, and the other two papers that are under review.

Homepage: https://chikuang.github.io/

*******************************************************************************************************