Whole record surveillance is superior to chief complaint surveillance for predicting influenza

Matched records of positive and negative influenza cases were parsed with a Natural Language Processor, the Multi-threaded Clinical Vocabulary Server (MCVS). Output was coded into SNOMED-CT reference terminology and compared to the SNOMED case definition of influenza. Odds ratios for each element of...

Full description

Saved in:
Bibliographic Details
Published in:AMIA ... Annual Symposium proceedings 2008-11, p.1173-1173
Main Authors: Welsh, Gail, Wahner-Roedler, Dietlind, Froehling, David Arthur, Trusko, Brett, Elkin, Peter
Format: Article
Language:English
Subjects:
Artificial Intelligence
Decision Support Systems, Clinical
Diagnosis, Computer-Assisted - methods
Disease Notification
Humans
Influenza, Human - diagnosis
Medical Records Systems, Computerized
Natural Language Processing
Population Surveillance - methods
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Matched records of positive and negative influenza cases were parsed with a Natural Language Processor, the Multi-threaded Clinical Vocabulary Server (MCVS). Output was coded into SNOMED-CT reference terminology and compared to the SNOMED case definition of influenza. Odds ratios for each element of the influenza case definition by each section of the record were used to generate ROC curves. C-statistics showed that whole record surveillance was superior to chief complaint surveillance for predicting influenza.
ISSN:1559-4076