作者: Shaun J. Grannis , J. Marc Overhage , Jeff Friedlin
DOI:
关键词:
摘要: We examined whether using a natural language processing (NLP) system results in improved accuracy and completeness of automated electronic laboratory reporting (ELR) notifiable conditions. used data from community-wide health information exchange that has ELR functionality. focused on methicillin-resistant Staphylococcus Aureus (MRSA), reportable infection found unstructured, free-text culture result reports. the Regenstrief EXtraction tool (REX) for this work. REX processed 64,554 reports mentioned MRSA we compared its output to gold standard (human review). correctly identified 39,491(99.96%) 39,508 positive MRSA, committed only 74 false errors. It achieved high sensitivity, specificity, predicted value F-measure. over two times as many without NLP. Using NLP can improve ELR.