Contact Us


David’s email: dsontag {@ | at}


How can I join the group as a Ph.D. student?

As of Fall 2018, we are accepting Ph.D. students into the lab. Graduate admissions are handled at the department level at MIT, so please first apply to the MIT EECS and/or HST (Health Sciences and Technology, joint between MIT and Harvard Medical School) Ph.D. program.

All students are expected to have a strong foundation in computer science or statistics, and many students will have previous research experience and publications at top machine learning conferences. Students primarily interested in machine learning should apply to EECS. For students interested in the intersection of machine learning and health care, either EECS or HST would be appropriate and there’s no difference in the Ph.D. thesis that you would write. Rather, the major difference is the emphasis of the training and the qualifying exam. HST Ph.D. students take additional classes at Harvard Medical School together with medical students, and gain a deeper appreciation of the impact that their research will have on clinical practice. As part of the qualifying exam, HST Ph.D. students write a short proposal that ties their theoretical work to its potential long-term impact in medicine.

Students may also consider applying to the Computational and Systems Biology, Institute for Data, Systems, & Society, and the Operations Research Center Ph.D. programs. For advice on CS Ph.D. admissions, Jean Yang and Philip Guo have written useful guides.

In your Ph.D. application, please explicitly mention your interest in working with Professor David Sontag. Unfortunately, we are unable to talk with students prior to their submitting their Ph.D. applications.

How can I join the group as a UROP/SuperUROP or MEng student?

We typically take one or two MEng students per year (slots may be available in both Fall and Spring). When looking for undergraduate students to join the lab, we look for students who have taken at least one machine learning class at MIT (e.g. 6.036 or 6.867) and received an A. For MEng students, we typically expect students to have taken at least one graduate-level machine learning class or to have had significant machine learning experience beforehand.

When contacting David, it is also helpful to have read over lab research projects beforehand.

I am an outside collaborator. How can I work with you?

The lab has had several successful collaborations with clinicians, insurance providers, and other parties. Please email David with a detailed description of the project, the data available, and the current challenges.

How can I learn more about machine learning and healthcare in general?

Clinical machine learning is a growing and important field, and we are excited that more people are interested in the topic. In addition to a general machine learning class, you may find useful the course materials for David’s Spring 2017 class 6.S897/HST.S53: Machine Learning for Healthcare.

Once you’ve developed a foundational base, research papers are an exciting way to learn more about the field. In addition to our recent publications, consider reading papers from relevant machine learning for healthcare conferences and workshops (e.g. MLHC and ML4H Workshop at NIPS) and general machine learning conferences (e.g. NIPS, ICML, AISTATS).