Displaying results 1 - 1 of 1
-
Identifying Contextual Features to Improve the Performance of an Influenza-Like Illness Text Classifier
Content Type: Abstract
To understand the types of false positive cases identified by an Influenza-like illness (ILI) text classifier by measuring the prevalence of ILI-related concepts that are negated, hypothetical, include explicit mention of temporality, experienced by… read more… on symptoms, problems, or findings from electronic note documents. False positive extractions may be due to concepts … of these strings. Two reviewers annotated the same document set with a third reviewer completing a blinded … version of the text classifier applied to surveillance document sources was 75% and 27% with 569(4%) false positive …