Lydia J. Sollis

Ph.D. student in Computer Science at the University of Hawaiʻi at Mānoa and graduate research assistant in the Department of Epidemiology at the University of Hawaiʻi Cancer Center, advised by Dr. John A. Shepherd.

I work on machine learning for medical screening, with a focus on whether these systems hold up in the populations they were not developed for. My current research examines where mammographic AI loses performance, for whom, and how to recover it, particularly across populations and institutions and among groups excluded from model development altogether.

Before graduate school I spent eight years in technical leadership roles in the U.S. Army and at UH Mānoa, managing IT operations, networks, and data infrastructure.

Ph.D. Portfolio

This site serves as my ICS Ph.D. portfolio, indexed in the categories specified by the program:

1. Statement of Purpose: research, teaching, service, and professional goals
2. Evidence of Core Competency: graduate degree and professional background
3. Evidence of Scholarly Ability: peer-reviewed publications with contribution statements

Also: Publications · Talks and posters · CV

Current work

Dissertation (in preparation): on model performance and adaptation in mammographic screening for underserved and underrepresented populations. Co-advised by John A. Shepherd and Huaijin Chen. Scope is being finalized; see my Statement of Purpose for the two research lines under consideration.

GMIC-FL: federated training and evaluation system for mammographic breast cancer detection across multiple clinical sites, built on GMIC, PyTorch, and NVIDIA FLARE.