25 Aug 2025 • A large real-world study of 54,353 fundus exam records showed that an AI-assisted screening system achieved high diagnostic accuracy for diabetic retinopathy (DR) and other retinal abnormalities.

Key Findings:

• About 25% of individuals screened had fundus lesions, with the AI reaching ≥96% negative predictive value and ≥90% positive predictive value.
• Beyond DR, it reliably detected retinal vascular sclerosis, drusen, maculopathy, optic cup enlargement, and hemorrhage.
• Detection rates were higher in those with diabetes, hypertension, high myopia, and longer disease duration.

Findings highlight AI’s potential for scalable, efficient early detection and broader utility in systemic disease monitoring, particularly in underserved populations.

Source: Dovepress | Read Full Story