Federal Grant Award Summary

The University of Minnesota, through its Office of Sponsored Projects Administration, received a $764,310 Project Grant from the National Eye Institute (NEI) under the Vision Research program (CFDA 93.867), effective July 1, 2026 through December 31, 2030. The award funds development of robust, multimodal artificial intelligence (AI) models to improve prognosis and risk stratification for Primary Open-Angle Glaucoma (POAG), a leading cause of irreversible blindness affecting millions globally. The project delivers three primary products: (1) an unsupervised feature representation model that integrates multimodal clinical data—including fundus images, optical coherence tomography (OCT), visual fields (VF), and structured clinical variables—while handling missing data modalities; (2) a Multimodal Longitudinal Transformer for Survival Analysis (MLTSA) that estimates time-varying POAG risk probabilities with adaptive robustness loss to ensure consistent performance across patient populations; and (3) clinical evaluation and integration protocols developed through iterative feedback from glaucoma specialists to ensure usability and workflow alignment.

These deliverables address critical gaps in current POAG risk assessment by providing automated, data-driven tools to support clinical decision-making and identify high-risk individuals for timely intervention. The research responds to documented limitations in existing predictive models, including dependence on single-modality data, poor performance across diverse patient populations, and misalignment with clinical workflows. By combining longitudinal multimodal imaging and clinical data with advanced machine learning techniques, the project aims to improve diagnostic accuracy while reducing the burden on ophthalmology specialists and enabling more efficient risk stratification in clinical settings.