Programmatic Theme: Clinical Research Informatics

Abstract: To extract information from unstructured text, a dictionary lookup task, which links text to standard terminologies, is typically required. In the context of a larger project to automatically extract relevant information from the EHR with high accuracy to then improve the completeness, correctness, and timeliness of electronic lists of medical problems and allergies, we evaluated several dictionary lookup tools (Apache Lucene, Apache UIMA ConceptMapper, and cTAKES fast lookup) and compared their speed and accuracy.

Learning Objective: Contrast accuracy and speed characteristics of a selection of dictionary lookup tools applied to clinical text notes.


Stephane Meystre (Presenter)
Medical University of South Carolina

Andrew Trice, Clinacuity, Inc.
Youngjun Kim, Medical University of South Carolina
Paul Heider, Medical University of South Carolina

Keywords, Themes & Types