JMIR Publications

ABSTRACT

With the emergence of large language models (LLMs) has come the opportunity to explore how they may be applied to facilitate current methods for concept extraction from large clinical databases. At Vanderbilt University Medical Center (VUMC), the in-house developed Word Cloud natural language processing (NLP) system extracts coded concepts from patient records in VUMC’s electronic health record (EHR) repository using UMLS terminology. Through this process, the Word Cloud represents the most prominent concepts found in the clinical documentation of a specific patient or population. This viewpoint describes a use case for how the VUMC Center for Knowledge Management leverages the condition-disease associations in the Word Cloud to aid in knowledge generation to inform interpretation of phenome-wide association studies. A pilot concept extraction example is also presented to demonstrate the potential for LLMs to facilitate extracting coded concepts from clinical texts at scale, potentially enabling replication of the Word Cloud’s features at other institutions.