Jeremy Weiss, M.D., Ph.D.

Dr. Weiss leads the Care Health and Reasoning Machines (CHARM) Lab, which develops machine learning methods for longitudinal clinical data to improve the characterization, forecasting, and understanding of disease trajectories. The lab's research focuses on temporal machine learning for multimodal electronic health record (EHR) data. We develop methods for clinical risk forecasting, temporal phenotyping and subgroup discovery, and causal inference, with particular emphasis on models that remain reliable as patient data, clinical practice, and data collection processes change over time.

The CHARM Lab's work spans several approaches to learning from longitudinal clinical data, including the development of robust prediction and survival models, methods for estimating heterogeneous treatment effects and reasoning about interventions from observational data, and approaches for identifying clinically meaningful subgroups and disease trajectories. By investigating how large language models and multimodal AI can improve the quality and temporal organization of EHR data through the integration of narrative text with structured records, we seek to reconstruct clinical timelines at scale. Across applications in sepsis, diabetes, post-acute illness, and other complex conditions, the lab aims to create methods and data resources that support more accurate forecasting, individualized risk characterization, and causal understanding of treatment and disease progression.