Radiology exams can reveal findings that require the immediate attention of the referring clinician. Such urgent results can include pneumothorax, a severely misplaced line, or any condition that could immediately endanger a patient's well-being, according to presenter Scott Werwath of the University of California, San Francisco (UCSF). It's the radiologist's responsibility to ensure that these urgent results are communicated to the referring physician.
"Given the importance of such communication, a machine-learning model which could automatically detect such findings from the text of a radiology report could be valuable in clinical settings as a quality-assurance tool," Werwath said.
The researchers used machine-learning and natural language processing techniques to build a model that could predict the likelihood of a radiology report including findings that required urgent communication.
"The model we developed performed well and could be integrated into clinical workflow to flag findings that require communication and ensure that such communication takes place," Werwath told AuntMinnie.com. "The model could also be used for an auditing of past reports to ensure that radiologists are following communication guidelines."
Get more details by sitting in on this Tuesday talk.
