Interpretable Multimodal Depression Screening on Cam-CAN via fMRI Connectivity Graph Neural Networks and Clinical Measures

Ruiheng Yu, Martin Katzman, Alexandria Greifenberger, Elssa Toumeh, Sachinthya Lokuge, Tia Sternat, Xinyu Qin, Lu Wang
January 2026

Abstract

Population-based screening for Major Depressive Disorder requires balancing sensitivity against specificity to avoid overwhelming clinical resources. Existing screening pipelines often rely on self-report questionnaires and do not integrate neuroimaging with clinical phenotypes. We evaluate multimodal decision-level fusion that combines resting-state fMRI connectivity graphs with clinical measures (sleep quality, anxiety, age) for depression screening. Using the Cam-CAN cohort (N=652; ~15% reporting depression requiring treatment), we compare graph neural network (GNN) architectures and fusion policies under 5-fold cross-validation with nested validation. A Graph Isomorphism Network with Risk-Union Fusion (RUF) achieves AUC of 0.85 with 62% sensitivity at 90% specificity, supporting sensitivity-oriented first-stage screening by flagging risk detected by either modality. Integrated Gradients identifies connectivity biomarkers localizing predominantly to temporal and cingulate regions. These results support interpretable multimodal integration for population-based depression screening.

Type
Conference paper
Publication
Manuscript submitted to the Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC 2026)

R. Yu, M. Katzman, A. Greifenberger, E. Toumeh, S. Lokuge, T. Sternat, X. Qin, and L. Wang. Interpretable Multimodal Depression Screening on Cam-CAN via fMRI Connectivity Graph Neural Networks and Clinical Measures. Manuscript submitted to the Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC 2026).

Graph Neural Networks Depression Screening FMRI Multimodal Interpretability EMBC
Ruiheng Yu
Ruiheng Yu
PhD Student of Biomedical Engineering

My research interests include Computer vision, explainable AI, LLM in health area.

Xinyu Qin
Xinyu Qin
PhD Student of Biomedical Engineering

My research interests include Causal Inference, Survival Analysis, and Digital Twin.

Lu Wang
Lu Wang
Assistant Professor of Biomedical Engineering

My research interests are developing and applying Machine Learning, Data Mining and Statistical methods (e.g., Multi-task Learning, Survival Analysis, Clustering, Risk Factor Analysis and Causal Discovery) on various data including gene expression, electronic health/medical records, and DNA sequencing reads for both cognitive disorders (e.g., delirium, Alzheimer’s disease, dementia, major depressive disorder) and chronic diseases (e.g., cancer, obesity, hypertension). Inspired by the human factors approach, she also designs and develops Human-Centered Artificial Intelligence tools for users to integrate, visualize, analyze, and interpret health data in order to improve the interoperability and accessibility of AI-assisted healthcare decision support.