Identifying Patients With Pulmonary Nodules From CT Radiology Reports Using Natural Language Processing (NLP)
Published in ATS Abstracts, 2023
Recommended citation: Dotson T.L., Gasimova A., Watkins J., Chometon Q., Bellinger C.R., (2023). "Identifying Patients With Pulmonary Nodules From CT Radiology Reports Using Natural Language Processing (NLP)." ATS.
Up to 31% of chest CT examinations report an incidental lung nodule (ILN). Previous studies estimate that around 60% of patients with ILNs can be lost to follow-up. One proposed reason is the lack of personnel to manage longitudinal surveillance, leading to either no follow-up or follow-up that does not adhere to guidelines, especially for patients coming from the emergency department. There is an unmet need for health care systems to automatically and accurately identify ILNs for appropriate follow-up. To help address this problem, we evaluated an automated method of identifying nodule patients from free-text CT radiology reports that can be tracked by a dedicated nodule navigator.