Optometrists outperform AI in diagnosing and managing age-related macular degeneration

Licensed optometrists demonstrated significantly higher accuracy than current artificial intelligence (AI) models in diagnosing and managing age-related macular degeneration (ARMD), according to a study presented at ARVO 2026 Annual Meeting.

In this study, optometrists reviewed clinical vignettes representing varying stages of ARMD. Participants provided diagnostic and treatment decisions, which were compared with responses generated by multiple AI models. Retinal specialists scored all responses against predetermined answer keys.

Optometrists achieved a mean diagnostic accuracy of 94.4% and treatment accuracy of 91.7%, while AI models demonstrated 46.7% accuracy in both categories. The differences were statistically significant (P < 0.05), with human clinicians showing a stronger ability to interpret complex findings and develop individualized management plans.

The authors emphasize the importance of clinician oversight and note that further refinement of AI with larger datasets and ophthalmic-specific training may improve its utility in the future.

Reference

Graf J, et al. AI vs Optometrists in the Diagnosis and Management of Age-Related Macular Degeneration. Poster presented at: ARVO 2026 Annual Meeting; May 3-7, 2026; Denver, CO.