Dean’s Awards for AI Research Development
Catalyzing a Sustainable School-wide AI Research Ecosystem for Public Health
Artificial intelligence (AI) is creating a once-in-a-generation opportunity to transform how we understand and improve health. By making it possible to learn from vast, complex, and diverse data, AI can uncover patterns, generate insights, accelerate workflows, and support decisions at a speed and scale not previously possible. For public health, its promise is especially powerful, as AI can connect many forces that shape health, from biology, behavior, lifestyle, environment, healthcare, to policy and social conditions, in ways that traditional tools could not. This creates new opportunities to have an integrated understanding of why health risks emerge, who is most vulnerable, which interventions are likely to work, and how to improve health across entire communities and populations. For a structured overview of major ways AI is advancing health research, with concrete examples, see Examples of AI Innovation in Health and Public Health Research.
The Dean’s Awards for AI Research Development are designed to catalyze this opportunity by supporting faculty-led AI-enabled research projects and building a sustainable School-wide AI research ecosystem for public health. The program welcomes a wide range of projects, from those that apply AI to important public health questions to those that advance new AI capabilities for health research.
Within this broad vision, the program especially encourages projects that explore emerging AI capabilities, including the development, adaptation, training, or evaluation of domain-specific foundation models, as well as agentic AI systems that can support research workflows, hypothesis generation, data integration, intervention design, or public health translation. These efforts should be driven by a clear public health or scientific question, with AI serving as a means to advance discovery, practice, or impact.
Program Concept: a Cohort-Based Model
From Faculty Ideas to a School-wide AI Research Ecosystem
The Dean’s Awards for AI Research Development are jointly coordinated by TRAIL4Health and the Dean’s Office of Research Strategy and Innovation. Within this jointly coordinated program, TRAIL4Health serves as the coordinating and scientific co-development hub for the program, helping awarded teams translate promising AI opportunities into well-scoped, technically feasible, and fundable research directions.
TRAIL4Health’s role is tailored to each project. Some teams may already have AI collaborators or technical capacity, while others may be exploring AI approaches for the first time. TRAIL will work with each awarded team to refine the project scope, assess data and infrastructure needs, identify gaps or opportunities in the AI strategy, and, where appropriate, co-develop AI-enabled solutions that are responsive to the project’s scientific or public health question.
Beyond individual project support, TRAIL4Health, together with the Dean’s Office of Research Strategy and Innovation, will coordinate the cohort as a shared learning and development environment. By bringing together public health investigators, AI faculty and experts, TRAIL AI staff, secure cloud infrastructure, and partners such as AWS and Google, the program will support peer learning, shared consultation, technical workshops, cross-project collaboration, and the development of practical examples, workflows, and collaborative capacity that strengthen Mailman’s broader AI research ecosystem.
Program Design: Two-Year Phased Design and Timeline
To operationalize the Dean’s Awards for AI Research Development, TRAIL4Health and the Dean’s Office of Research Strategy and Innovation will jointly coordinate a structured two-year support model that moves awarded projects from initial scoping and support matching, through AI-enabled project development, toward grant readiness, dissemination, and broader School-wide learning.
Phase 1. Launch, Scoping, and Support Matching
Months 0-2
After funding decisions are announced, awarded project teams will participate in an onboarding process coordinated by TRAIL4Health. Each team will meet with TRAIL to refine the project scope, clarify the AI opportunity, assess data readiness, identify technical or infrastructure needs, and develop an initial support plan.
The cohort will also participate in a joint launch meeting. Each team will briefly present its public health question, data resources, AI opportunity, anticipated challenges, and proposed next steps. This meeting will help identify common themes across projects, opportunities for collaboration, and shared needs for AI expertise, data support, cloud infrastructure, or platform resources.
Based on project needs, TRAIL will help match teams with the appropriate support. This may include TRAIL AI staff, faculty consultants, AWS or Google technical resources, or other internal and external experts. The level and type of support will be tailored to each project and will depend on project needs, existing team expertise, available program capacity, and opportunities for shared support across the cohort.
Phase 2. AI Scientific Consultation and Project Development
Months 3-12
During the first year, projects will begin implementation with tailored support from TRAIL4Health and relevant technical partners. For teams without established AI collaborators, TRAIL may help identify appropriate expertise and, where feasible, co-develop AI-enabled solutions tailored to the project’s scientific or public health question. For teams with existing AI capacity, TRAIL’s role may focus on complementary consultation, infrastructure support, partner engagement, or connection to broader cohort resources.
The program will convene an AI Scientific Consultation Council, composed of TRAIL core faculty and related external AI experts. The Council will provide scientific and technical consultation rather than direct project management. Council members may participate in cohort meetings, provide feedback on AI strategy and study design, advise on data and infrastructure challenges, and serve as consultants when specific expertise is needed.
The cohort will meet approximately every three to six months to share progress, discuss challenges, receive feedback, and identify opportunities for collaboration. External experts may be invited to selected meetings to provide guidance on AI methods, AI agents, data platforms, evaluation, implementation, or public health translation.
Phase 3. From Early-Phase Projects to Grant-Ready Programs
Months 13-24
In the second year, TRAIL4Health and the Dean’s Office of Research Strategy and Innovation will work with awarded teams to strengthen results, refine AI-enabled approaches, prepare demonstrations, and develop future funding plans. TRAIL4Health and the AI Scientific Consultation Council will focus on the AI strategy, technical approach, preliminary results, and scientific narrative. The Dean’s Office of Research Strategy and Innovation will support grant development, including identifying appropriate funding mechanisms, shaping proposal strategy, connecting teams with potential collaborators, and helping position projects for larger collaborative opportunities.
As projects mature, faculty experts who have provided consultation may become more formally involved as co-investigators, collaborators, or MPI partners for external grant applications, where appropriate.
Projects will be encouraged to move toward submission or near-submission of an external grant by the end of the two-year program. TRAIL AI staff who provide substantive support to project development are expected to be included in follow-on grant proposals, with appropriate roles and effort, where their continued expertise is needed. Potential pathways may include NIH R-series, U-series, center grants, coordination centers, PCORI, foundation grants, industry partnerships, philanthropic initiatives, or multi-institutional collaborations.]
Cross-Cutting Goal: Building a Living AI Research Ecosystem
Because AI methods and tools evolve rapidly, the program will not focus narrowly on preserving static tools. Instead, it will help build a living AI research ecosystem for the School. This ecosystem may include AI-ready data resources, secure cloud workflows, shared evaluation practices, technical expertise, AI-agent and co-scientist capabilities, training opportunities, and collaborative networks that can evolve as technology advances.
Where feasible, projects may contribute knowledge, data resources, workflows, documentation, prompts, evaluation benchmarks, dashboards, or other materials to the emerging Columbia Mailman AI Research Asset Commons. The Asset Commons should be viewed as a dynamic and evolving resource, not a fixed repository. Its purpose is to help future investigators discover useful data resources, methods, workflows, examples, and expertise generated through the program.]
Phase 4. School-wide Demo and Future Directions
Month 24 and Beyond
TRAIL4Health and the Dean’s Office of Research Strategy and Innovation will coordinate a School-wide demo or showcase session. Each project team will present its scientific or public health question, AI-enabled approach, progress and findings, lessons learned, future funding plan, and any resources or workflows that may be useful to the broader Columbia Mailman community.
By the end of the two-year period, each project is expected to have submitted, or be well positioned to submit, an external grant application or larger collaborative proposal. The expected outcome is not only a set of successful early-phase projects, but also a stronger School-wide AI research ecosystem: investigators with hands-on AI experience, an expert consultation network, stronger partnerships with AWS and Google, emerging AI co-scientist capabilities, shared research practices, and a pipeline of projects positioned for sustained public health impact.]