Publications

You can also find my articles on my Google Scholar profile.

Journal Articles

RobustPDx: a structured web-based Parkinson’s assessment dataset and benchmark for AI robustness to device type and handedness

Published in Scientific Data (in press), 2026

A public dataset and robustness benchmark for device-type and handedness effects in remote Parkinson assessment - an acquisition-shift problem structurally parallel to the scanner and protocol heterogeneity studied in my dissertation.

Recommended citation: Tumpa ZN, Shahriar Zawad MR, Sollis L, Parab S, Azizian P, Honarmand M, Kargarandehkordi A, Tarrit K, Abbasi-Asl R, Bove R, Wall DP, Washington P. (2026). "RobustPDx: a structured web-based Parkinson's assessment dataset and benchmark for AI robustness to device type and handedness." Scientific Data. In press.
View Paper

Remote assessment of Parkinson disease using deep learning on structured mouse-trace data from suspected cases: machine-learning pilot feasibility study

Published in JMIR Formative Research, 2026

Deep learning applied to passively collected mouse-movement data as a low-burden remote screening signal for Parkinson disease.

Recommended citation: Shahriar Zawad MR, Tumpa ZN, Sollis L, Parab S, Washington P. (2026). "Remote assessment of Parkinson disease using deep learning on structured mouse-trace data from suspected cases." JMIR Formative Research. 10:e81368.
View Paper

Compact subsets of autism screening items predict clinical diagnoses with a machine learning analysis of the QCHAT-10

Published in Scientific Reports, 2025

Four-item subsets of the QCHAT-10, selected by recursive feature elimination and trained on New Zealand and Saudi Arabian cohorts, transferred to clinically diagnosed Polish cases (AUROC 85-87%). Eye contact, gaze following, and pretend play emerged as robust markers across all cohorts.

Recommended citation: Sollis LJ, Wall DP, Washington PY. (2025). "Compact subsets of autism screening items predict clinical diagnoses with a machine learning analysis of the QCHAT-10." Scientific Reports. 15:39091.
View Paper

Quantifying device type and handedness biases in a remote Parkinson’s disease AI-powered assessment

Published in npj Digital Medicine, 2025

Characterizes how input device and handedness introduce measurable bias into a remote AI-based Parkinson disease assessment, a device-fairness problem closely analogous to the acquisition-shift questions in my dissertation.

Recommended citation: Tumpa ZN, Shahriar Zawad MR, Sollis L, Parab S, Chen IY, Washington P. (2025). "Quantifying device type and handedness biases in a remote Parkinson's disease AI-powered assessment." npj Digital Medicine. 8(1):550.
View Paper

Conference Papers

Personalized federated learning for equitable breast cancer detection in underrepresented Pacific Islander populations

Published in 7th MICCAI Workshop on Distributed, Collaborative and Federated Learning (DeCaF 2026), Strasbourg, France, 2026

Adapted GMIC for federated fine-tuning with NVIDIA FLARE across a three-site federation including a diverse Pacific Island cohort. FedAvg, FedProx, and FedBN improved the largest site but left smaller sites at or below pretrained performance; Ditto improved the worst-performing site and was the most equitable strategy.

Recommended citation: Sollis LJ, Young PM, Bunnell A, Quon B, Hernandez BY, Wolfgruber TK, Shepherd JA. (2026). "Personalized federated learning for equitable breast cancer detection in underrepresented Pacific Islander populations." 7th MICCAI Workshop on Distributed, Collaborative and Federated Learning (DeCaF), Strasbourg, France.
View Paper

Do deep embedding-derived imaging phenotypes reveal breast cancer risk heterogeneity beyond scalar risk stratification?

Published in Proceedings of the 18th International Workshop on Breast Imaging (IWBI 2026), Thessaloniki, Greece, 2026

Tested whether unsupervised clustering of Mirai embeddings identifies stable, clinically meaningful phenotypes beyond the scalar risk score, in 26,573 HIPIMR women across 79,317 screening examinations. Clusters were highly stable (ARI > 0.98) but recovered groups corresponding largely to predicted risk rather than distinct clinical phenotypes.

Recommended citation: Sollis LJ, Shvetsov Y, Quon B, Wolfgruber TK, Bunnell A, Hernandez BY, Shepherd JA. (2026). "Do deep embedding-derived imaging phenotypes reveal breast cancer risk heterogeneity beyond scalar risk stratification?" In: Sechopoulos I, ed. Proceedings of the 18th International Workshop on Breast Imaging (IWBI 2026), Thessaloniki, Greece. Zenodo. pp. 423-432.
View Paper

Lydia J. Sollis