RFA-DK-26-314
Single Source for Establishing Pilot/Opportunity program for AI Models to Accelerate Diabetes Research (U24- Clinical Trials not allowed)
Summary
Briefing: AI Foundation Models for Diabetes Heterogeneity
Diabetes presents a major public health challenge characterized by substantial disease heterogeneity—variation in disease mechanisms, progression, and treatment response across individuals—that demands personalized approaches to prevention, diagnosis, treatment, and prognosis. The diabetes research field has accumulated large, complex datasets and extensive prior knowledge, yet extracting actionable predictive signals remains difficult due to limited data science expertise and the absence of diabetes-specific artificial intelligence (AI) tools and foundation models. This pilot program seeks multidisciplinary teams combining diabetes researchers and data science experts to develop AI foundation models tailored to diabetes, validate them against key research questions in disease heterogeneity, and create practical use cases demonstrating their research acceleration potential. The initiative spans computational biology, clinical informatics, and machine learning, aiming to integrate AI expertise into the diabetes research workforce and deliver accessible models and tools for the broader community.
Who can apply: Multidisciplinary teams including both diabetes domain experts and data science/AI experts; merit-based peer review required.
Funding & project length: Not stated.
Award / mechanism: Pilot funding program.
Key dates: Not stated.
Best fit for: Researchers in endocrinology, metabolic disease, and biomedical data science using machine learning, predictive modeling, and foundation models to address diabetes heterogeneity and personalized medicine.
Insights (5)
Multidisciplinary team composition is non-negotiable; diabetes-only expertise insufficient
The initiative explicitly requires teams that include both diabetes experts AND data science/AI experts. A diabetes researcher without embedded data science collaborators will be at a structural disadvantage. This is not a preference but a core program design principle, making team assembly a critical success factor before submission.
Foundation model development positions applicants as field infrastructure builders
The program prioritizes creating reusable AI foundation models for diabetes heterogeneity that the broader community can adopt. Applicants with experience in model generalization, transfer learning, or building tools for non-specialist users will be more competitive than those proposing single-use predictive models. This positions successful applicants as workforce developers, not just researchers.
Pilot program structure favors established diabetes researchers with new AI partnerships
As a pilot initiative, this likely targets researchers with credibility in diabetes science who are adding AI capability, rather than early-stage investigators. The emphasis on 'integrating new AI experts into the diabetes research workforce' suggests the program values diabetes domain knowledge as the anchor, with AI as the new addition—advantaging mid-to-senior researchers over ESI.
Diabetes heterogeneity focus requires evidence of prior work on disease subtypes or phenotypes
The program centers on addressing heterogeneity through AI—not general diabetes research. Applicants with prior publications or preliminary data on diabetes subtypes, endotypes, or phenotypic stratification will demonstrate stronger fit than those proposing generic predictive models. This specificity narrows the competitive pool but rewards focused expertise.
Pilot program scope suggests limited awards; high bar for meritorious applications
The text emphasizes 'pilot funding program' and 'only meritorious applications will be considered'—language indicating selective, not broad, funding. Combined with the requirement for multidisciplinary teams and foundation model development (high technical bar), expect intense competition among well-resourced institutions with both diabetes and AI capacity.
Key Facts
Deadline
—
Posted
Thu, September 25, 2025
Keywords
Research Areas