Interactive Natural Language Processing for Identifying Incidental Findings

We present an interactive tool to help clinicians find relevant information in patient notes. Our tool builds Natural Language Processing models to highlight parts of a document.

It allows users to make any necessary corrections, thus forming a feedback loop to improve the NLP models.

In this demo, we demonstrate our tool for identifying incidental findings in a trauma service. Radiologic imaging often uncovers findings that are unrelated to the trauma. These are called “incidental" findings and identifying them is necessary for proper follow-up.

Gaurav Trivedi, Charmgil Hong, Esmaeel R. Dadashzadeh, Robert M. Handzel, Wendy Chapman, Harry Hochheiser, Shyam Visweswaran. Interactive NLP in Clinical Care: Identifying Incidental Findings in Radiology Reports. Applied Clinical Informatics, 10(04), Pages 655-669, 2019. DOI: 10.1055/s-0039-1695791

Narrated by: Bryan Andrews. Revision 25.02.19-34b3456.