Linked to from EHR Usability, Workflow & NLP Presentations at AMIA 2012 on the EHR Workflow Management Systems blog.

AMIA 2012: Identifying Catheter-Associated Urinary Tract Infections with NLP and Machine-Learning Techniques While Embedded in Live Clinical Operations

ABSTRACT: Urinary tract infections from urinary catheters(CAUTIs) account for 40% of hospital-acquired infections(HAIs)1, which carry substantial morbidity and cost2. CAUTI identification3 requires time-intensive medical-record reviews. We are building an automated, real-time, web-based platform utilizing NLP and machine-learning to identify presence of urinary catheters and CAUTIs, and will continually retrain its algorithms against CAUTI documented from ongoing Infection Control surveillance. Preliminary NLP performance for presence of urinary catheters: 57 inpatients, 8 hospital-days, precision=1, recall=0.875, 54/58 catheter-days identified.

AMIA 2012 Itinerary Planner

Related:

Clinical NLP at 2012 NAACL Human Language Technology Conference