Distributed foundational models for multi-task learning in diabetic retinopathy
About This Grant
Abstract: This project aims to establish distributed federated learning (FL) approaches for multi-task training of foundational machine learning (ML) models for diabetic retinopathy (DR), using multi-modal, real-world optical coherence tomography (OCT) data (OCT cross-section, OCT angiography (OCTA), and OCT enface). DR is one of the leading causes of severe vision loss. Early detection, prompt intervention, and reliable assessment of treatment outcomes are essential to prevent irreversible vision loss from DR. However, there are major challenges towards developing clinically relevant holistic algorithms that can perform multi-tasks, i.e., multi-class classification of disease stages (diagnosis), prediction of onset and progression of disease stages (prognosis), and assessment of treatment outcomes. They require large amounts of well curated and labelled datasets from a diverse sub-population for robust performance. Moreover, efforts towards large, centralized datasets for ML research are hindered by significant barriers to data sharing and privacy concerns. In this project, we propose to develop foundational ML models that allow efficient learning of feature representations from a large corpus of ophthalmic imaging data for various downstream tasks – breaking the task-specific paradigm of current ML models. We also establish novel federated ML approaches, where the model training is distributed across institutions instead of sharing patient data. Our first aim is to establish and validate a domain adaptive FL framework for DR diagnosis across four independent institutions. We propose a novel ophthalmic adaptive personalized FL (optho-APFL) technique to tackle domain shift caused by heterogeneous data distribution at different institutions (due to different sub-population density and OCT devices/imaging protocols). We will conduct experiments on the FL deployment in a clinical setting and integrate a granular differential privacy (DP) algorithm into our FL framework to provide ‘patient-level’ data privacy. Key success criterion is to deploy the FL framework and validate FL-trained ML models against state-of-the-art models for DR diagnosis. The second aim is to develop foundational ML models with self-supervised learning (SSL) to learn multiple tasks within the same framework from label invariant OCT/OCTA data, where different institutions don’t need to have labeled data for each of the tasks. We will train these foundational models in a centralized and FL framework for comparative analysis. Key success criterion is to i) validate foundational model performance for multi-task learning (MTL) (DR staging, prediction of NPDR to PDR progression, and prediction of DME treatment evaluation) on new clinical data (centralized and FL approach, and ii) identify task-specific quantitative OCT/OCTA (mean and artery-vein specific) features. As an alternative approach, we propose diffusion probabilistic modeling (DPM) for SSL to learn holistic representations from multi-modal OCT data for MTL, and to explore dynamic federated averaging approaches. Success of this project will establish OCT/OCTA based distributed foundational models for objective MTL in DR using label-invariant data across multi-institutions and standardize OCT/OCTA features for MTL.
Grant Summary
Distributed foundational models for multi-task learning in diabetic retinopathy is a NEI - National Eye Institute grant providing up to $617K for university, nonprofit, healthcare org. Applications are due 2030-03-31 (open). Check eligibility and apply with FindGrants.
Not quite the right fit?
Search 9,000+ open grants, or get matches ranked for your organization — free.
Focus Areas
Eligibility
How to Apply
Up to $617K
2030-03-31
- 1Confirm your organization is eligible for Distributed foundational models for multi-task learning in diabetic retinopathy from NEI - National Eye Institute, checking organization type, location, and any population or project requirements.
- 2Gather the required documents and information, including your organization details, project plan, and budget figures.
- 3Draft your application narrative and budget addressing the funder's priorities and review criteria. FindGrants can draft each section for you to review and edit.
- 4Review every section against the requirements checklist, then export a submission-ready application pack and submit it to NEI - National Eye Institute before the deadline.
Don't want to draft it yourself?
We'll draft the complete application against NEI - National Eye Institute's requirements, run a quality review, and email you a submission-ready PDF plus an editable Word doc within 5 business days. Most orders deliver in 24-48 hours. Flat $399, any grant size.
AI Requirement Analysis
Detailed requirements not yet analyzed
Have the NOFO? Paste it below for AI-powered requirement analysis.
Distributed foundational models for multi-task learning in diabetic retinopathy: Frequently Asked Questions
Who is eligible for the Distributed foundational models for multi-task learning in diabetic retinopathy?
Distributed foundational models for multi-task learning in diabetic retinopathy is offered by NEI - National Eye Institute and is generally open to university, nonprofit, healthcare org. It is open to organizations nationwide unless the funder specifies otherwise. Review the specific eligibility terms before applying, since funders set their own requirements around organization type, location, and the population or project being served.
How much funding does the Distributed foundational models for multi-task learning in diabetic retinopathy provide?
Distributed foundational models for multi-task learning in diabetic retinopathy provides up to $617K per award from NEI - National Eye Institute. Actual award sizes depend on the scope of your project, available program funds, and the number of applicants, so build a budget that reflects realistic, allowable costs rather than the maximum figure.
When is the Distributed foundational models for multi-task learning in diabetic retinopathy deadline?
Applications for Distributed foundational models for multi-task learning in diabetic retinopathy are due 2030-03-31 (open). Because deadlines can change, verify the date with the funder, NEI - National Eye Institute, and give yourself enough time to prepare a complete, competitive application before the close date.
How do you apply for the Distributed foundational models for multi-task learning in diabetic retinopathy?
To apply for Distributed foundational models for multi-task learning in diabetic retinopathy, confirm your eligibility, gather the required documents, and prepare a narrative and budget that address the funder's priorities. FindGrants guides you step by step and can draft each section, then exports a submission-ready application pack for this grant from NEI - National Eye Institute.