MIT Research Positions in Machine Learning for Health

MIT Research Position in Machine Learning for Health

The Massachusetts Institute of Technology (MIT), Institute for Medical Engineering & Science (IMES) invites applications for a full-time research position in Machine Learning for Health. This is an immediate opening for a highly motivated researcher interested in developing machine learning methods for latent representation learning and generative modeling from complex, multimodal, time-varying clinical data, with the goal of informing sequential treatment decision-making and generating actionable insights with high potential impact in clinical medicine.

The project offers opportunities to develop and apply novel machine learning and statistical approaches to generate clinically meaningful insights from observational health data, including clinical time series and physiological signals, with extensions to multimodal learning from medical imaging and physiological waveforms. The successful candidate will join a multidisciplinary team working at the interface of computational methods and clinical medicine to develop approaches with high translational value that inform patient care and treatment decisions.

• Ph.D. in Computer Science, Machine Learning, Statistics, Biomedical Informatics, or a related field
• Strong publication record in top-tier machine learning and AI venues
• Demonstrated ability to conduct independent, high-quality research
• Familiarity with modern probabilistic inference methods
• Publications in top-tier AI/ML and/or ML for Health venues.

• Probabilistic machine learning, including latent variable models, approximate inference, variational autoencoders (VAEs), structured VAEs, and related deep latent variable models.
• Representation learning from time-varying, multimodal data, with an emphasis on interpretable latent structure discovery and latent state representation learning for forecasting and sequential decision-making.
• Dynamic systems and state-space modeling, including deep state-space models and switching dynamical systems [1,2,3,4].
• Incorporation of inductive biases, structural constraints, or domain knowledge into ML models to improve robustness and generalization (e.g. constrained inference [1,2]).

Applicants should send a CV to Li-wei Lehman (lilehman@mit.edu) - please specify your current affiliation, your expected timeline for starting the position, and list 2–3 representative publications and venues. Applications will be reviewed periodically, and candidates whose background and expertise are a strong fit will be contacted for next steps.

Research Associate: Candidates must have authorization to work in the U.S. This is a temporary, full-time research position ending August 31st, 2027, with the possibility of remote work within the US. Renewal maybe possible, subject to funding availability and performance. The position offers a compensation rate corresponding to approximately $68,000–$73,000 on an annualized full-time basis, with the specific rate determined based on qualifications and experience.

Postdoc: for candidates who currently have authorization to work in the US, a shorter-term postdoc (e.g. 11-months)may be possible upon requests. Postdoc appointments are renewable on a yearly basis, subject to funding availability and performance.

1. Knowledge Distillation via Constrained Variational Inference, Ardavan Saeedi, Yuria Utsumi, Li Sun, Kayhan Batmanghelich, Li-wei H. Lehman, Proceedings of the Thirty-Sixth AAAI Conference on Artificial Intelligence, February 2022.
2. Switching State Space Modeling via Constrained Inference for Clinical Outcome Prediction, Arnold Su, Anna Wong, Ardavan Saeedi, Li-wei H Lehman, MLHC 2025.
3. A Physiological Time Series Dynamics-Based Approach to Patient Monitoring and Outcome Prediction, Li-wei H. Lehman, Ryan P. Adams, Louis Mayaud, George B. Moody, Atul Malhotra, Roger G. Mark, Shamim Nemati, IEEE Journal of Biomedical and Health Informatics, 19(3):1068-1076, May 2015. doi:10.1109/JBHI.2014.2330827. Preprint.
4. Bayesian nonparametric learning of switching dynamics in cohort physiological time series: application in critical care patient monitoring, Li-wei H. Lehman, Matthew J. Johnson, Shamim Nemati, Ryan P. Adams, Roger G. Mark, Chapter in Advanced State Space Methods for Neural and Clinical Data, Cambridge University Press, 2015. Publisher's Version.

MIT Postdoc Associate in Machine Learning and Sequential Decision Making for Health

The Massachusetts Institute of Technology (MIT), Institute for Medical Engineering & Science (IMES) invites applications for a Postdoctoral Associate position in Machine Learning and Sequential Decision Making for Health. This is a position for a highly motivated researcher with strong background in machine learning and sequential decision making to work on developing approaches for time-varying, multimodal clinical data, supporting sequential treatment decision-making and generating actionable clinical insights.

The project provides opportunities to develop and apply state-of-the-art machine learning and statistical methods to large, multimodal, time-varying observational data from electronic health records. The successful candidate will join a multidisciplinary team working at the interface of computational methods and clinical medicine, contributing to projects with high translational impact on healthcare.

• Ph.D. or advanced degree in Computer Science, Machine Learning, Data Science, Statistics, or a related field
• Strong background in machine learning or statistical modeling
• Solid publication record in top machine learning venues

• Machine learning for health and sequential decision making
• Model-based offline reinforcement learning; off-policy policy evaluation
• Counterfactual reasoning from observational data
• Interpretable and trustworthy models for high-stakes decision-making
• Incorporation of inductive biases, structure, or domain knowledge for robust learning
• Experince in the following advantageous: Dynamical systems, foundation models for decision making, causal inference. Experience with real-world health datasets is helpful
• Publications in AI/ML venues preferred, e.g. NeurIPS, ICML, ICLR, AISTATS, UAI, AAAI, MLHC, ML4H, CHIL or equivalent

Applicants should email a CV, brief description of research interests, and expected timeline for starting the position to Li-wei Lehman (lilehman@mit.edu).

Please include current affiliation and 2–3 representative publications and venues. Applications will be reviewed periodically, and candidates with a strong match will be contacted for next steps.

Duties include conducting original research, developing and evaluating machine learning models on clinical and physiological data, publishing in top-tier conferences and journals, mentoring students, and contributing to research grant proposal writing. Annual salary: $73K.