Abstract: Current procedures for evaluating trauma severity with risk scores are performed retrospectively by certified coders for quality reporting and prognostication. Manual annotation at point-of-care is not feasible and automated approaches can improve trauma care throughput and triaging. We demonstrate that natural language processing with machine learning from routinely collected radiology reports can provide excellent discrimination for prediction of severe chest injury. Our approach is a viable solution to augment clinical decision support in trauma care.

Describe the new knowledge and additional skills the participant will gain after attending your presentation.: The attendee will be able to learn the value in radiology reports collected during trauma care for analytics in clinical decision support.

Authors:

Sujay Kulshrestha (Presenter)
Loyola University Chicago

Brihat Sharma, Rush University Medical Center
Richard Gonzalez, Loyola University Chicago
Cara Joyce, Loyola University Chicago
Dmitriy Dligach, Loyola University Chicago
Matthew Churpek, University of Wisconsin
Majid Afshar, Loyola University Chicago