Materials associated with the Analytic Solutions for Real-Time Biosurveillance: Negation Processing in Free Text Emergency Department Data for Public Health Surveillance consultancy held January 19-20, 2017 at the University of Utah, Salt Lake City.
Problem Summary
False positive syndrome hits are created when a syndromic classification process cannot properly identify negated terms. For example, a visit is classified into a fever syndrome when the chief complaint or triage note says “denies fever.”
NC DETECT’s current approach is to use a combination of Emergency Medical Text Processor (EMT-P) (available at https://www.ibridgenetwork.org/#!/profiles/6065458510418/innovations/33/ ) and NegEx (available at http://toolfinder.chpc.utah.edu/content/contextnegex ). This approach works for the majority of negation in triage notes.
Challenges appear, however, when negation is complex and/or doesn’t follow set rules. In addition, some EHRs are generating triage note text through the use of templates. Templates can speed up data entry for clinicians but can result in text that is difficult to process for negation using current NLP approaches.
Attachments
- Use case summary
- Consultancy agenda
- List of attendees