Commentary - Synergistic AI-resident approach achieves superior diagnostic accuracy in tertiary ophthalmic care for glaucoma and retinal disease

Sep 28, 2026

Uploaded by Logic

A commentary analyzes a study demonstrating that a synergistic collaboration between artificial intelligence systems and resident physicians improves diagnostic accuracy in tertiary ophthalmic care for glaucoma and retinal disease. The discussion contrasts combined performance with standalone AI and early-year residents, highlighting gains in glaucoma-suspect classification accuracy and retinal sensitivity, as well as tracking of CDR estimates. It also addresses educational implications, particularly how reduced cognitive load and altered disease exposure may affect trainees’ clinical judgment.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages2

Reading Time5 min

AccessFree to download

Key results and comparison

AI alone vs residents vs synergistic AI-resident partnership

Educational implications for trainees

Cognitive effort and altered exposure to disease spectrum

Opportunities for deliberate system design

Teaching modes and explainable AI

TYPE General Commentary PUBLISHED 19 December 2025 DOI 10.3389/fopht.2025.1749411 OPEN ACCESS EDITED BY Michael Wormstone, The University of Nottingham Ningbo (China), China REVIEWED BY Alessio Martucci, University of Rome Tor Vergata, Italy *CORRESPONDENCE Rosa Corro [rosaacorro@gmail.com](rosaacorro@gmail.com) RECEIVED 19 November 2025 REVISED 02 December 2025 ACCEPTED 08 December 2025 PUBLISHED 19 December 2025 CITATION Williams-Gaona K and Corro R (2025) Commentary: Synergistic AI-resident approach achieves superior diagnostic accuracy in tertiary ophthalmic care for glaucoma and retinal disease. Front. Ophthalmol. 5:1749411 . doi: 10.3389/fopht.2025.1749411 COPYRIGHT © 2025 Williams-Gaona and Corro. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Commentary: Synergistic AI-resident approach achieves superior diagnostic accuracy in tertiary ophthalmic care for glaucoma and retinal disease Kevin Williams-Gaona 1 and Rosa Corro 2* 1Caja de Seguro Social (CSS) Dr. Hugo Spadafora, Colón, Panama, 2 Department of Research, Mayo Clinic Florida, Jacksonville, FL, United States KEYWORDS artiﬁcial intelligence, clinical training, diagnostic workﬂow, medical education, ophthalmology A Commentary on: Synergistic AI-resident approach achieves superior diagnostic accuracy in tertiary ophthalmic care for glaucoma and retinal disease By Camacho-Garca-Forment D, Bayln-Vzquez G, Arriozola-Rodrguez K, Avalos-Ramirez E, Hartleben-Matkin C, Valdez-Flores H, Hodelin-Fuentes D and Noriega A (2025) Front. Ophthalmol. 5:1581212 . doi: 10 .3389/fopht.2025.1581212 Camacho-Garćıa-Forment́ı et al. present an impressive demonstration of how a synergistic collaboration between artiﬁcial intelligence systems and resident physicians can improve diagnostic accuracy in glaucoma and retinal disease. Not only do they show that an AI-resident partnership can outperform either one alone, they also demonstrate something that many groups struggle to capture: how these systems behave in a real tertiary-care environment. Anyone who has worked in a busy ophthalmology service knows that elegant results on paper are one thing, and making them coexist with high patient load, mixed pathology, and time pressure is another. Their team managed both. Even though the combined approach was the top performer, the study also shows that AI on its own outperformed ﬁrst-year residents across several key measures, including higher accuracy in glaucoma suspect classiﬁcation (88.6% vs. 82.9%) and much higher sensitivity for retinal disease (76% vs. 52%, and 100% for high-risk ﬁndings) . AI’s CDR estimates also tracked more closely with expert measurements than those of residents (r = 0.728 vs. 0.538), although the system could only evaluate CDR in 61.6% of patients because of image-quality issues. That detail matters: even when an algorithm performs well, realworld imaging is variable, and someone still needs to handle the cases it cannot process. As we read through their results, a central question emerged. We talk a lot about“human in the loop” AI, and it’s usually framed as a reassuring idea. The algorithm assists, but the clinician remains in control. That logic works today, when clinicians have years of pattern recognition behind them. But what happens when the human entering the loop has had fewer chances to build the very skills the loop depends on? Frontiers in Ophthalmology 01 [frontiersin.org](frontiersin.org) Ophthalmology is built on repetition. Residents grow by seeing normal variants, borderline OCTs, unusu

FAQs

What is the main finding discussed in the commentary?

The commentary emphasizes that combining AI systems with resident physicians can achieve superior diagnostic accuracy for glaucoma and retinal disease in a real tertiary-care setting.

How does the commentary compare AI performance to first-year residents?

It notes that AI alone outperformed first-year residents on measures including glaucoma-suspect classification accuracy and retinal disease sensitivity, while also explaining limitations related to image quality for CDR evaluation.

What concerns does the commentary raise about using AI in training settings?

It argues that when trainees rely on AI outputs, educational risk may shift toward reduced cognitive effort and narrower disease exposure, potentially changing how clinical judgment develops.

Sep 28, 2026

Uploaded by Logic