Lung nodules are among the most common incidental imaging findings, and evidence-based management of these cases begins with the radiology report, according to senior author Dr. Stella Kang of NYU Langone Medical Center.
"We wondered how frequently incidental lung nodules were being reported with recommendations that followed the Fleischner Society guidelines," she told AuntMinnie.com. "Natural language processing provided a means of rapidly identifying these incidental lung nodules in our electronic health record, as we assess the outcomes and how to improve care."
The researchers found that their NLP tool could be trained to detect most incidental lung nodules on radiology reports with high sensitivity and moderate specificity.
"We then looked at longitudinal trends in reporting of these nodules and recommendations and found that the rate of providing recommendations was steady even after we implemented a department-wide macro with the Fleischner Society guidelines, and that usage of the guideline macro had room for improvement," Kang said. "The findings will help us provide continual feedback for process improvement and assess the guideline concordance of the recommendations we provide."
What else did they find? Sit in on presenter Dr. Ryan Chung's talk to learn more.
