This study systematically examined what large language models memorize when adapted for medicine, how frequently memorization occurs, how much content is retained, and how it may affect downstream medical applications.
This study investigates high-risk medication prescribing for older adults discharged from emergency departments, showing 6.5% receive potentially inappropriate medications, highlighting opportunities for safer practices.
This study introduces MOSAIC, a computational framework that efficiently simulates complex systems with individual-level diversity and memory, enabling accurate modeling across biology, physics, and social networks.
This study investigates electronic health record (EHR) use patterns among physicians, showing that timely inbox management and reduced after-hours work improve chart completion efficiency and patient visit volume.