Investigating the feasibility of using AI to detect unreported chronic disease findings on chest X-rays in a retrospective aged patient dataset
Author
Luchs, Jonathan Stephen | Premier Radiology Services, US
Scientific poster presentation (W5A-SPCH-3) at RSNA 2023, 26 – 30. November 2023 in Chicago, US
Purpose
Can comprehensive AI decision support improve the detection of chronic diseases on chest X-rays in aged patients?
Method
Retrospective analysis of 1,223 CXR cases of patients ³65 years from US outpatient centers. Comparison of Annalise Enterprise CXR result and clinical report on 15 chronic findings, ground-truth review in case of disagreement.
Results
Agreement between AI model and report in 71.2% (Range 43.1 – 97.7%), with certain chronic diseases underreported in clinical routine.
Key Takeaway
AI may improve patient outcomes with its high sensitivity, flagging findings omitted by radiologists’ reports for further evaluation.
