NLM Adds AI to Its Flagship Informatics Training Grant: $17.1 Million, Up to 27 Programs, One Shot Per Institution, Due February 11, 2027

In May 2022 the National Library of Medicine named the 18 universities that would train most of the country's federally funded biomedical informatics PhDs and postdocs for the next five years. The awards totaled close to $58 million over that period, about $11.5 million a year, and supported about 110 predoctoral and 60 postdoctoral trainees annually. The list included Columbia, Harvard Medical School, Johns Hopkins, Stanford, Vanderbilt, Yale, the University of Washington, the University of Utah, Oregon Health & Science, and nine others.

That cycle ends in 2027, and the next competition has been posted. RFA-LM-28-001, "NLM Institutional Grants for Research Training in Biomedical Informatics, Data Science, and AI," went up on Grants.gov on October 16, 2026. The title now includes "and AI." NLM and its partners plan to commit an estimated $17.1 million across 19 to 27 awards. The award count and the dollar figure are both higher than the 2022 cohort. Applications are due February 11, 2027.

The growth is significant, but it is tied to new requirements, and the rule allowing only one application per institution will make this an internal competition at many universities before it is a federal one.

The numbers

The parts of the NOFO that determine budgets and timelines:

• Mechanism: T15 (Continuing Education Training Grants), Independent Clinical Trial Not Allowed. Trainees cannot lead their own trials but may work on a trial led by a mentor, which NIH explicitly encourages.
• Funding: an estimated $17,100,000 for 19–27 awards.
• Budget cap: application budgets are "limited to $775K in direct costs," and must reflect actual need.
• Project period: up to five years.
• Indirect costs: 8% of modified total direct costs, the standard training-grant rate, not your negotiated rate. Tuition and fees, consortium costs above $25,000, and equipment are excluded from the base.
• Trainee costs: NRSA stipend, tuition, and training-related expense levels; $2,000 per trainee per year for travel, including the NLM Informatics Training Conference.
• Application types: New and Renewal.
• Limit: one application per institution, identified by UEI.
• Dates: the submission portal opens January 11, 2027. Applications are due February 11, 2027 at 5:00 p.m. local time, with no late applications. Peer review is July 2027, NLM Board of Regents review is October 2027, and the earliest start is December 2027.
• Eligibility: domestic only. Foreign institutions and foreign components of U.S. organizations cannot apply.

The NOFO does not say whether $17.1 million is a first-year or annual figure. Compared directly with the 2022 cohort's roughly $11.5 million a year, it is about 50% more money spread over up to nine more programs. At 25 awards, the count Grants.gov lists, the average is about $684,000, which leaves room under the $775,000 direct cap for programs that are not the largest.

The program grows by adding institutes

The money comes from more partners. NLM is joined by eight institutes: the National Cancer Institute, National Eye Institute, NIAID, NIDCD, NIDCR, NIDA, NIMH, and NINDS. Two offices, OBSSR and the Office of Data Science Strategy, may co-fund. The NIH Blueprint for Neuroscience Research also lists priorities.

That affects how a proposal should be written. A program requesting "additional training slots from participating disease-focused institutes or NIH offices must also be responsive to research priorities which are specific to the intended secondary institute or NIH office." Each partner publishes its own list in Section V. NCI wants informatics, data science, and AI/ML applied to cancer research, with mentorship from both cancer and computational faculty. NIMH wants training that uses the NIMH Data Archive and its common data elements and follows the NAMHC Workgroup on High Dimensional Data recommendations. NINDS and Blueprint want neuroscience across the full range from basic to clinical. ODSS lists partner sectors: government agencies and labs, industry (biotech, pharma, devices, diagnostics, digital health, health data analytics), public health organizations, and health systems.

These lists should be read carefully but not too literally. The NIDA entry repeats the NIAID text word for word ("infectious, allergic, immunologic, and transplant researchers"), and the NIDCD entry repeats NIDCR's oral-health language ("NIDCR's DOC research priorities"). Both look like copy-paste errors. NLM says it will hold an informational webinar before the receipt date where institute representatives "will explain research area priorities and responsiveness criteria in more detail," and will post details on the NLM Extramural Programs site. If you are designing slots around NIDA or NIDCD, attend that webinar or email NLMProgram@nih.gov before building a curriculum around text that may be corrected.

The strategy follows from this. The core T15 slots are NLM's. Additional slots come from showing a partner institute that its own research area is part of your program. A program with a real neuroscience or cancer informatics strength can ask for partner-funded slots, but only if that partner's priorities appear in the curriculum, the mentor list, and example trainee projects. A list of partner institutes that the application never returns to will not get those slots.

AI is now required in the core curriculum

The 2022 program covered health care and clinical informatics, translational bioinformatics, clinical research informatics, public health informatics, and consumer health informatics. The new NOFO adds AI explicitly. Every program "must include a required core curriculum" covering four areas: principles of informatics, data science, and AI; quantitative methods such as biostatistics or applied mathematics; computer science, engineering, or information sciences; and the design and conduct of rigorous, reproducible research. Advanced coursework and mentored research in the program's focus areas build on that core.

Three other requirements are easy to miss:

Course access documentation. Any course outside the PI's home unit requires letters showing trainees will have "routine access." Informatics programs commonly rely on courses in computer science, statistics, and public health, so these letters need to be requested early.

A specific career boundary. The program "is not intended to train individuals for careers primarily centered on the planning, deployment, maintenance, or administration of health care, public health, or research systems and data." It funds researchers, not health IT staff. A program whose graduates mostly go into hospital informatics operations will look misaligned. A program whose graduates do research in industry or health systems is acceptable, because reviewers are directed to count "the breadth of careers that sustain the biomedical research enterprise."

A more detailed evaluation plan. NLM wants a "comprehensive evaluation plan" with both qualitative and quantitative measures, alumni tracking, trainee feedback, and preferably a named framework: the NOFO cites CIPP, Kirkpatrick, and IHI Plan-Do-Study-Act. Renewing programs will be asked whether they recruited "trainees in sufficient numbers and of high quality" and how graduates have done in "academic, industry, or government careers." A 2022 renewal application should present those alumni outcomes as its central evidence.

Problems specific to renewing programs

The 18 current programs have a timing issue. Their awards were announced in May 2022 for up to five years, so most current project periods end during 2027. The new cycle's earliest start is December 2027. A program with a mid-2027 end date could have several months without funding unless NLM extends it or the institution covers the gap. Check your Notice of Award end date now and talk to your program officer before the February deadline. Trainee appointments run in 12-month increments, and stipends cannot be paid in advance.

Renewing programs also face more competition than in 2022. Expanding from 18 to as many as 27 awards gives new entrants a better chance, and the partner-institute structure helps universities with particular disease-area strengths. A current grantee that writes a renewal similar to its 2022 application, without the AI curriculum, partner-institute alignment, and the new evaluation framework, will be reviewed against new programs written specifically for this NOFO.

The internal deadline comes before the federal one

Because the limit is one application per institution by UEI, a university with informatics strengths in several units (a medical school department, a school of public health, an engineering college, a cancer center) can submit only one. The NOFO adds that "institutions with existing programs must explain what distinguishes this program from the others, how their programs will synergize," and demonstrate that the faculty pool and trainee pipeline can support another program.

Plan backward from February 11:

• By early November: the research office announces the limited submission, and units submit brief internal concepts. Many universities run limited-submission reviews four to eight weeks before the sponsor's deadline.
• By December: select one PD/PI team. NLM wants "an established investigator" with administrative and scientific leadership experience, and multiple-PI applications need a strong leadership justification. Gather course-access letters, compile the required training data tables, and write the advisory committee attachment (two pages maximum).
• By mid-January: the portal opens January 11. Submit early enough to fix eRA Commons errors before the 5:00 p.m. deadline on February 11.

Universities that do not hold a current T15 should look at the partner-institute route. A medical school with strong cancer or neuroscience data science, or a public health school with real-world longitudinal data that fits OBSSR's priorities, could build a new program around one partner institute and compete on that basis.

The 8% indirect rate means the T15 will not improve a university's finances. Its value is that it trains the computational researchers that many other NIH applications now depend on. For related NLM funding, see our analysis of the NLM clinical informatics R01 (PAR-26-042).

With four months to February 11, much of that time will go to internal selection. Granted can help a PD/PI team turn curriculum tables and partner-institute priorities into a coherent training narrative before the limited-submission review.