Research

Our work focuses on developing and evaluating methods that use routinely collected health data to generate evidence. We combine methodological development with applied studies, with an emphasis on validity, fairness, and real-world performance.

Research Areas

• Clinical NLP and patient-generated data
  We study how language data (clinical notes, voice recordings, and patient-reported data) can be used for measurement and prediction.
• Real-world evidence from EHRs
  We develop approaches to use observational health data for study design and inference. This includes handling missing data, bias, and heterogeneity in large-scale EHR datasets.
• Decision and care support in serious illness and palliative care
  We develop and evaluate AI approaches to support complex clinical decision making in serious illness, with a focus on communication, uncertainty, and patient-centered care.
• Experimental AI ethics
  We examine how AI systems affect equity and decision making in healthcare through a critical lens, with a particular focus on palliative care.