Prompt Engineering for Portable Clinical Note Structure Awareness

Poster Number: P184
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Natural Language Processing, Large Language Models (LLMs), Interoperability and Health Information Exchange, Information Retrieval
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Clinical note structure and metadata elicitation are critical for many downstream clinical NLP applications, but portability, robustness, and cost of existing solutions continues to hamper the adaptability and adoption across healthcare systems. Prompt engineering strategies for large language models have demonstrated the potential for generalizable performance at a relatively low cost. Our study exhibits the promise of a dynamic few-shot prompting technique for section classification and proposes future approaches for document structure awareness.

Speaker(s):
Kurt Miller, M.S.
University of Minnesota

Author(s):
Steven Bedrick - Oregon Health & Science University; Hongfang Liu, PhD - University of Texas Health Science Center at Houston;