David Sontag

David Sontag joined the MIT faculty in 2017 as Hermann L. F. von Helmholtz Career Development Professor in the Institute for Medical Engineering and Science (IMES) and as Assistant Professor in the Department of Electrical Engineering and Computer Science (EECS). He is also a principal investigator in the Computer Science and Artificial Intelligence Laboratory (CSAIL). Professor Sontag’s research interests are in machine learning and artificial intelligence. As part of IMES, he leads a research group that aims to transform healthcare through the use of machine learning.

Prior to joining MIT, Dr. Sontag was an Assistant Professor in Computer Science and Data Science at New York University’s Courant Institute of Mathematical Sciences from 2011 to 2016, and postdoctoral researcher at Microsoft Research New England from 2010 to 2011. Dr. Sontag received the Sprowls award for outstanding doctoral thesis in Computer Science at MIT in 2010, best paper awards at the conferences Empirical Methods in Natural Language Processing (EMNLP), Uncertainty in Artificial Intelligence (UAI), and Neural Information Processing Systems (NIPS), faculty awards from Google, Facebook, and Adobe, and a NSF CAREER Award. Dr. Sontag received a B.A. from the University of California, Berkeley.

DEGREES

• Ph.D. in Electrical Engineering and Computer Science, Massachusetts
  
  Institute of Technology, 2010
• S.M. in Electrical Engineering and Computer Science, Massachusetts
  
  Institute of Technology, 2007
• B.A. in Computer Science, University of California, Berkeley, 2005

Professor Sontag’s research both aims to advance the field of machine learning and artificial intelligence, and to apply these to transform healthcare.

Precision Medicine

These are exciting times for the practice of medicine. The rapid adoption of electronic health records and has created a wealth of new data about patients, which is a goldmine for improving our understanding of human health. Our lab develops algorithms that use this data to better understand disease progression and to facilitate new, precise treatment strategies for a wide range of diseases and conditions such as Type 2 diabetes, which affects tens of millions of people worldwide every year, and multiple myeloma, a rare blood cancer. In pursuit of these aims, a major methodological focus has been on developing novel approaches to modeling high-dimensional time-series data, particularly approaches that bring together probabilistic modeling and deep learning, and causal inference from observational data.

Intelligent Electronic Health Records

Today’s electronic health records are predominately a place for recording a patient’s health data. We aim to develop the foundation for the next-generation of intelligent electronic health records, where machine learning and artificial intelligence is built-in to help with medical diagnosis, automatically trigger clinical decision support, personalize treatment suggestions, autonomously retrieve relevant past medical history, make documentation faster and higher quality, and predict adverse events before they happen. A major challenge is the need for robust machine learning algorithms that are safe, interpretable, can learn from little labeled training data, understand natural language, and generalize well across medical settings and institutions.