Overview

The “Healthy ML” group at MIT, led by Dr. Marzyeh Ghassemi, focuses on creating and applying machine learning to understand and improve health in ways that are robust, private and fair. Health is important, and improvements in health improve lives. However, we still don’t fundamentally understand what it means to be healthy, and the same patient may receive different treatments across different hospitals or clinicians as new evidence is discovered, or individual illness is interpreted.

Unlike many problems in machine learning - games like Go, self-driving cars, object recognition - disease management does not have well-defined rewards that can be used to learn rules. Models must also be “healthy”, in that they should not learn biased rules or recommendations that harm minorities or minoritized populations. The Healthy ML group tackles the many novel technical opportunities for machine learning in health, and works to make important progress with careful application to this domain.

Read more about our Research Directions and Publications. The Healthy ML group is also affiliated with the AI for Society group at MIT.

News

  • April 2026 - Congratulations to Haoran, Hyewon, and Isha, whose work has been published in ICML 2026!
    • Position: Benchmarks Do Not Measure Deployment Readiness in Clinical AI
      Haoran Zhang, Hyewon Jeong, Olawale Salaudeen, Walter Gerych, Nigam Shah, Marzyeh Ghassemi
    • Escaping the Mode: Multi-Answer Reinforcement Learning in LMs
      Isha Puri, Mehul Damani, Idan Shenfeld, Marzyeh Ghassemi, Jacob Andreas, Yoon Kim
  • January 2026 - Congratulations to Cassie, Quinn, Yuxin, Sana, Vinith, and Kimia, whose work has been published in ICLR 2026!
    • WRING Out The Bias: A Rotation-Based Alternative To Projection Debiasing [Paper]
      Walter Gerych, Cassandra Parent, Quinn Perian, Rafiya Javed, Justin Solomon, Marzyeh Ghassemi
    • When Style Breaks Safety: Defending LLMs Against Superficial Style Alignment [Paper]
      Yuxin Xiao, Sana Tonekaboni, Walter Gerych, Vinith Menon Suriyakumar, Marzyeh Ghassemi
    • Complementing Self-Consistency with Cross-Model Disagreement for Uncertainty Quantification [Paper]
      Kimia Hamidieh, Veronika Thost, Walter Gerych, Mikhail Yurochkin, Marzyeh Ghassemi

Joining the Lab

If you are interested in doing an UROP, SUROP, or MEng, please review the lab’s research page, and talk to a graduate student or postdoc who might be closest to your research interest. It is important that all undergraduate students have a graduate or postdoctoral scholar who is close to their research area for advice.

If you are interested in doing a PhD with the Healthy ML group, Dr. Ghassemi admits PhD students from the MIT EECS, IDSS, and HST pools. Please plan to apply to any or all of these programs, and indicate in your application that you would be interested in working with her. Unfortunately, MIT does not allow for direct admission from an individual PI to the institution.

For postdoctoral fellows, please send your resume and a short statement of what work you would plan to do for your time with the lab (with references to relevant content from the lab). Katie O’Reilly (oreilly1@mit.edu) can schedule a 30-minute review meeting once an assessment of a good fit has been made.