LLM Mechanics in Research Using Clinical Notes
Large Language Models are increasingly used to analyze clinical notes, but their behavior can be difficult to understand without examining how prompts, context, and outputs interact. In this seminar, Subhadeep Nag, MS, will introduce the mechanics of using LLMs for clinical note research, including how prompts are structured, how clinical text is prepared for LLM input, how outputs can be interpreted, and where caution is warranted when applying these methods in healthcare research. The session will focus on practical examples from EHR-based research and will highlight both the promise and limitations of LLMs in high-stakes clinical settings.
Speaker:
Subhadeep Nag, MS
Research Data Scientist
NC TraCS Institute
Seminars in the NC TraCS Data Science Lab Seminar Series cover a range of topics related to health care data science, clinical data, data engineering, and working in these areas at UNC-Chapel Hill. These hybrid seminars are usually held monthly on the third Tuesday of each month from 12:30-1:30 p.m. in the NC TraCS suite on the 2nd floor of Brinkhous-Bullitt or via Zoom.