Comparison and Cost Strategies Using AI Chat vs API to Map CDEs Across Studies of Chronic Diseases and Disparities in the Health Equity Action Network (HEAN) Research Consortium

Poster Number: P18
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Large Language Models (LLMs), Data Sharing, Knowledge Representation and Information Modeling, Health Equity, Natural Language Processing, Population Health, Informatics Implementation, Racial Disparities
Primary Track: Foundations
Programmatic Theme: Clinical Informatics

HEAN is a national consortium for the prevention and management of multiple chronic conditions (46 R01-level studies). The Research Coordination Center's role is to harmonize common data elements (CDEs) (>8K question items) across studies. We piloted Versa (chat and API), UCSF’s secure generative AI platform as a mapping tool, on a subset of 5 studies (1,304 CDE items) and assessed mapping accuracy, time, and cost. We will discuss results, strategies, and recommendations.

Speaker(s):
Hyelee Kim, MD, MAS, MS
UCSF