Ophthalmology
Machine learning outperformed human graders on eye images
For left eyes, a score measuring how well estimates distinguished glaucoma status was 88% for machine learning versus 79% for human graders.For left eyes, a score measuring how well estimates distinguished glaucoma status was 88% for machine learning versus 79% for human graders.
Cross-sectional · 6,304 peopleBiomarkers
Who was studied6,304 EPIC-Norfolk Eye Study participants, against trained human graders
What changedAUROC for glaucoma status in left eyes was higher with machine learning: 88% against 79% for human graders
The catchOnly participants with images from both eyes gradable by both methods were included


















