Artificial intelligence (AI) in diabetic retinopathy screening: evaluating the impact of IDx-DR and EyeArt in global eye care

Background: Diabetic retinopathy is a preventable cause of vision loss, but screening coverage is limited when retinal imaging, expert interpretation, referral completion, and access to treatment are not uniformly available. Objective: The purpose of this review is to assess the clinical evidence, regulatory status, and health-system applicability of three FDA-cleared autonomous diabetic retinopathy screening platforms: LumineticsCore (formerly IDx-DR), EyeArt, and AEYE-DS. Scope: The review will address diagnostic performance, imageability, camera compatibility, acquisition workflow, evidence provenance, geographic validation, referral burden, and deployment context. CE-marked and international non-FDA systems are examined as comparators to provide context for global innovation. Methods: A narrative synthesis of evidence from January 2010 to April 2026 was performed using peer-reviewed studies, regulatory submissions, and real-world reports. Endpoints, reference standards, camera protocols, study populations, and ungradable-image handling differed across sources, precluding quantitative pooling. Key Findings: Autonomous DR screening has progressed from proving technical feasibility to demonstrating pathway fit. LumineticsCore aligns with controlled-workflow autonomy, EyeArt with program-scale triage, and AEYE-DS with portable access expansion. CE-marked and international systems broaden the field through handheld, smartphone-based, offline, and locally adapted models. Clinical value is shaped not only by sensitivity and specificity, but also by imageability, true screening denominator, referral burden, evidence maturity, patient follow-up, and local care capacity. Conclusions: Autonomous screening is most likely to reduce preventable vision loss when it operates as a care pathway linking image capture, referral completion, and treatment initiation. Future evidence should evaluate systems under fair local conditions, measure completed screening and treatment outcomes, and align each platform with the clinical capacity of the setting in which it is deployed