Description
Medication information from electronic health records (EHRs) provides invaluable research data. We present medExtractR, a natural language processing system that extracts such information from clinical notes with accuracy suitable for statistical modeling using medication data. We found that medExtractR performed better than three existing natural language processing systems: MedEx, MedXN, and CLAMP. Our results suggest a better alternative for extracting medication information from EHRs. Improved entity-level extraction ultimately leads to higher quality EHR-based research datasets.
Learning Objective: After participating in this session, the learner should be able to 1) identify key challenges in extracting medication information from free-text clinical notes and 2) discuss the impact of using this data from electronic health records in statistical analyses.
Authors:
Hannah Weeks (Presenter)
Vanderbilt University
Cole Beck, Vanderbilt University
Elizabeth McNeer, Vanderbilt University
Cosmin Bejan, Vanderbilt University
Joshua Denny, Vanderbilt University
Leena Choi, Vanderbilt University
Presentation Materials: