Research Interests

• Health
• Fairness
• Representation Learning
• Unsupervised Learning
• Latent Variable Models
• Neural Networks
• Creating and applying machine learning algorithms towards improved prediction and stratification of relevant human risks

Biography

Marzyeh completed her PhD at MIT where her research focused on machine learning in health care, exploring how to predict immediate and long-term patient needs to inform decisions in the intensive care unit and ambulatory care. Her current research interests include clinical risk prediction with semi-supervised learning, optimal treatment discovery using expert demonstrations, and non-invasive patient phenotyping for behavioral conditions. Prior to MIT, she received a B.S. degree in computer science and electrical engineering at New Mexico State University and Master’s degree in biomedical engineering from Oxford University. Marzyeh is on the Board of Women in Machine Learning (WiML), and co-organized the NIPS 2016/2017 Workshop on Machine Learning for Health, and MIT’s first Hacking Discrimination event.

Highlights

• Nominated for 2017 Best Student Paper at AMIA Summit on Clinical Research Informatics (CRI) for “Predicting Intervention Onset in the ICU with Switching State Space Models”
• First Place at 2014 MIT $100K Accelerate $10,000 Daniel M. Lewin Accelerate Prize, Kohana Student Team
• First Place at 2013 MIT Sloan-ILP Innovators Showcase, Sana AudioPulse Student Team
• American Marshall Scholar 2008

Research Activity and News

• Guest Editor for PLoS ONE call on Machine Learning in Health and Biomedicine http://blogs.plos.org/everyone/2018/03/09/call-for-papers-ml4health/
• Speaker at IACS SYMPOSIUM ON THE FUTURE OF COMPUTATION IN SCIENCE AND ENGINEERING “The Digital Doctor: Health Care in an Age of AI and Big Data” – https://www.youtube.com/watch?v=XKE8UY4_lWw
• Panelist at AMIA 2018 Informatics Summit Panel on “Deep Learning for Healthcare – Hype or the Real Thing?” https://www.amia.org/2018-Informatics-Summit/panels
• Semi-supervised Biomedical Translation with Cycle Wasserstein Regression GANs
  Matthew McDermott, Tom Yan, Tristan Naumann, Nathan Hunt, Harini Suresh, Peter Szolovits, and Marzyeh Ghassemi
  CWR-GAN code is released, at https://github.com/mmcdermott/CWR-GAN. The Model Usage jupyter notebook walks through how to use the code and shows a successful synthetic run.
• Clinical Intervention Prediction and Understanding using Deep Networks
  Harini Suresh, Nathan Hunt, Alistair Johnson, Leo Anthony Celi, Peter Szolovits, Marzyeh Ghassemi