MIT HEALS Fellowships: AI Health Research Guide

MIT HEALS fellowships are not one program. There are separate postdoc fellowships funded by the Biswas Family Foundation and Novo Nordisk, a graduate fellowship round that drew 220 applications, and an undergraduate research track. Each has its own eligibility, funding source and application date, and this guide covers all four without merging them.

What Are MIT HEALS Fellowships?

MIT HEALS fellowships are named, funded training positions at the Massachusetts Institute of Technology that place early-career researchers where artificial intelligence meets health and life sciences. They are not a single program: the Biswas Fellows Program, the MIT Novo Nordisk AI Postdoctoral Fellows program, a HEALS graduate fellowship, and an undergraduate research program sit under the same initiative, each with its own funder and eligibility.

MIT HEALS itself is an institute-wide initiative at MIT that organizes research and training across health, AI and life sciences. The fellowship programs are how it recruits people into those labs. Funding comes from outside donors rather than from a single internal budget: the Biswas Family Foundation, started in 2023 by Samsara co-founder Sanjit Biswas and his wife Hope, backs the postdoc fellows named after the family, and Novo Nordisk backs the AI postdoc fellows.

The practical consequence for an applicant is that these programs behave differently. Postdoc fellowships carry a stipend and a cohort structure plus industry or foundation mentoring. The graduate fellowship covers one year of tuition and stipend. The undergraduate program is a research placement across departments rather than a funded fellowship in the same sense.

Because the programs were announced at different times and the cohorts rotate, treat any eligibility detail as provisional until you read the current program page. The figures below come from a 2025 institute presentation and describe those cohorts, not a standing quota.

The Biswas Fellows Program and Its First Cohort

The Biswas Fellows Program supports early-career postdoc fellows at MIT HEALS with funding from the Biswas Family Foundation, and its first cohort of five fellows was introduced publicly in 2025. Each fellow works under a named MIT faculty supervisor on a distinct problem, and the program was described as multiyear with more cohorts to come.

The first cohort's projects span surgical imaging, brain cancer, wearable brain-state monitoring, bacterial genomics and cell-signalling engineering. That range is the point of the program: the foundation said it was attracted by cross-pollinations between disciplines, so the selection favors ideas that combine fields rather than sitting inside one.

Novo Nordisk and the Biswas foundation are separate funders of separate programs, so do not read the Novo Nordisk cohort statistics below as Biswas program statistics. The Biswas program's published detail in 2025 covered its first cohort and its funder, not application counts or publication output.

MIT Novo Nordisk AI Postdoctoral Fellows: Numbers and Timeline

The MIT Novo Nordisk AI Postdoctoral Fellows program funds postdocs who work at the interface of AI and life science, and it has run three cohorts since launching in the School of Engineering in 2023. At the point the program was described publicly in 2025, applications were open for a fourth cohort with a priority closing date of April 15.

The program reported 26 fellows and 28 faculty advisors across its first three cohorts, with faculty drawn from 13 different departments, centers and labs. It also reported about 25 publications and 15 conference presentations from those cohorts. Those totals are program-reported figures from a single institute presentation, not an audited count, and they describe the state in 2025 rather than today.

Novo Nordisk's role goes past money. Fellows take part in monthly virtual journal clubs with Novo Nordisk scientists, and some have visited the company's Massachusetts site and international sites. Faculty describe Novo Nordisk as a mentoring partner rather than only a funder.

Recruiting topics listed for the program include AI for imaging, AI for drug discovery, large datasets and clinical analysis, AI and sensors for data collection with an emphasis on brain health, AI and robotics for human-robot interaction, and microbiomes, viromes and new pipeline molecules.

How Do the MIT HEALS Fellowship Tracks Compare?

The three main tracks differ most in career stage, funding source and what they cover. The table below lists what each program was reported to fund, based on the 2025 institute presentation; confirm current terms on each program's own page before applying.

What the First Cohorts Actually Research

The research topics below come from the fellows' own 2025 presentations, and each one names the faculty lab hosting the work. They are a snapshot of the first cohorts, not a list of open positions. All five Biswas fellows and both Novo Nordisk speakers presented work that either builds a measurement tool or applies one to a disease question.

Biswas Fellow projects

Tom Dylan, a recent MIT PhD, works with Brian Anthony on medical imaging and robotics for cardiovascular surgery. He described current procedures as guided mainly by 2D X-ray imaging, which forces surgeons to infer the 3D position of tools and implants, and he is building AI and virtual-reality guidance aimed at reducing intraoperative complications.

Janaina Macedo da Silva, who completed her doctorate in São Paulo, works in Forest White's lab on glioblastoma. She is testing a drug reported to cross the blood-brain barrier and using omics methods to map the phosphoproteome in primary and disseminated tumors, with the goal of understanding resistance mechanisms.

A third fellow works with Laura Lewis on wearable brain-state monitoring using fNIRS, a light-based technique that measures blood-flow changes in the brain, with simultaneous ultrasound stimulation explored as a way to shift arousal and sleep states. The stated aim is monitoring outside hospital settings.

Rachel Silverstein, joining from Harvard, works in a microbiology lab on what she calls microbial dark matter, using genome language models to predict the function of bacterial genes that have no known role. She cites CRISPR as an example of a major discovery that began with unexplained repeating sequences in bacterial genomes.

Constantine Tzouanas, a Harvard-MIT Program in Health Sciences and Technology graduate, works with Sangeeta Bhatia on cell-to-cell signalling. He cited a literature meta-analysis finding that about 90% of data covers roughly 10% of signalling proteins, and he plans to build protein engineering tools and AI/ML methods to test more signals against primary human cells and tissues, starting with viral hepatitis.

The HEALS Graduate Fellowship Round and the Undergraduate Track

The HEALS graduate fellowship awarded 32 fellowships from 220 applications in its first round, covering one year of tuition and stipend, and it drew students from nearly every MIT graduate program. The undergraduate research program reported more than 100 participants across 16 departments in its first year, spanning life sciences, economics and social sciences.

Both programs exist to force contact between disciplines. Faculty described graduate students as the connective tissue of MIT collaborations, noting that most lab partnerships start when a student in one lab knows a student in another, and that this happens less often across different graduate programs. Monthly meetings of the graduate fellows were described as the mechanism that produces new pairings.

The undergraduate program is adding a formal clinical innovators track, described in 2025 as launching that fall, which places students in hospitals and clinical settings during training rather than at the end of it. That is the clearest example in the initiative of translation being built into a training pathway rather than treated as a later step.

Graduate fellows' projects

Diego, a PhD candidate in the Sanchez Rivera lab, models cancer-associated genetic variants using base editing, a CRISPR-based tool that rewrites single DNA letters. His focus is disordered protein regions, which do not fold into stable structures, where the usual sequence-to-structure-to-function logic breaks down. Working with the Walter lab at Harvard, the project integrated functional readouts with a map of over 1 million pairwise protein interactions to find which interactions variants disrupt.

Elijah Pivo, a sixth-year PhD candidate at the Institute for Data, Systems, and Society, works on the US organ transplant system. He reported that one allocation-policy simulation took nearly seven hours and that evaluating a single policy required 10 simulations, so a written report on a handful of policies could take four to six months. His group built a replacement algorithm that runs the same simulations in about 15 seconds, which allows thousands of policies to be tested and an interactive site for policymakers.

Kasey Love, a graduate fellow in biological engineering working with Katie Galloway, builds gene circuits that hold expression steady across cells. The design adds microRNA-based regulatory elements that implement an incoherent feedforward loop on top of a therapeutic gene, addressing the problem that unregulated gene delivery produces some cells with too little protein and others with toxic levels.

A fourth graduate fellow works in the Computational Biophotonics Lab on label-free metabolic microscopy, imaging the natural fluorescence of cofactors such as NADH and FAD to read cell metabolism without staining. The work combines metabolic contrast with structural imaging to study breast cancer microenvironments and intact tissue.

Why Fellowships Matter More Than Grants in AI Health Research

Fellowship funding buys time and proximity, and both show up in what the cohorts reported producing. The graduate round put 32 students from nearly every MIT graduate program into monthly meetings together. The Novo Nordisk program reported about 25 publications and 15 conference presentations across three cohorts while running journal clubs with company scientists.

The proximity argument is measurable in one case. The organ-transplant project only became possible because a robotics-trained PhD student moved into a policy institute, then reduced a seven-hour simulation to 15 seconds using methods from his original field. Cross-department placement is what made that transfer happen.

Funders also shape the science through selection. The Biswas foundation said it looked for high-risk, cross-disciplinary ideas and chose postdocs specifically to attract talent to the institute. Novo Nordisk selected for depth in both AI and life science, and faculty described requiring fellows to be strong in both rather than strong in one with interest in the other.

Beware of reading these numbers as outcomes. The publication and presentation counts are program-reported, the transplant policy result is a single case, and the cohorts are small. Treat them as evidence that these programs are active and producing work, not as proof that the fellowship model outperforms other funding routes.

How to Apply and What to Expect

Applications run through each program separately, and the only application date stated publicly in the 2025 material was the Novo Nordisk postdoc priority closing date of April 15. Graduate and undergraduate routes follow MIT's own admissions and research-placement processes rather than a single HEALS application.

For the postdoc programs, prepare for a selection process that weighs cross-disciplinary fit as heavily as technical depth. The Biswas program said it screened for high-risk, high-reward ideas that combine fields, and the Novo Nordisk program stated a preference for fellows already deep in both AI and life science. Naming two labs or two departments you would work across is likely to matter.

Expect a cohort structure rather than a solo postdoc. Both postdoc programs place fellows in a group that meets regularly, connects them with mentors outside their lab, and in the Novo Nordisk case includes company scientists in journal clubs and site visits. If you want an isolated research post, these are the wrong programs.

One practical caveat: the fellowships are named after funders and rotate, so confirm the current cohort number, closing date and stipend on the program page before you rely on any figure here. The 2025 presentation describes the programs as they stood then.

FAQ

• What are MIT HEALS fellowships? MIT HEALS fellowships are funded research positions at MIT for postdocs, graduate students and undergraduates working where AI meets health and life sciences. They sit under the MIT HEALS initiative but are funded separately, mainly by the Biswas Family Foundation and Novo Nordisk, and each has its own eligibility and application route.

• How many fellows has the MIT Novo Nordisk program funded? The program reported 26 fellows and 28 faculty advisors across three cohorts launched since 2023, with faculty from 13 departments, centers and labs. It also reported about 25 publications and 15 conference presentations at the time of the 2025 presentation. These are program-reported totals for those cohorts, not current figures.

• What is the application deadline for MIT HEALS fellowships? The only deadline stated in the 2025 material was April 15, described as the priority closing date for the fourth Novo Nordisk postdoc cohort. The Biswas, graduate and undergraduate programs follow their own timelines, so check each program page for the current cycle.

• What kinds of research do HEALS fellows do? Fellows work on medical imaging and surgical navigation, glioblastoma drug response, wearable brain-state monitoring, bacterial genome language models, cell-signalling engineering, AI-assisted chemical synthesis, sleep slow-wave enhancement, cancer variant modelling, organ allocation policy, gene circuit design and label-free metabolic microscopy.

• Do HEALS fellowships require both AI and biology expertise? For the Novo Nordisk postdoc program, yes: faculty stated they recruit fellows who are deep in both AI and life science. The Biswas program weights cross-disciplinary creativity and high-risk ideas, which can include pairing fields that rarely meet, such as oceanography and cell biology.

Source video