Diagnostic accuracy and clinical performance of deep learning models for grading diabetic retinopathy: a systematic review and meta-analysis.
Deep learning detects no diabetic retinopathy with very high sensitivity on fundus photos
Diagnostic accuracy and clinical performance of deep learning models for grading diabetic retinopathy: a systematic review and meta-analysis.
Diabetic retinopathy (diabetic retinopathy) is a leading cause of preventable visual impairment worldwide, and its precise severity grading is critical for optimizing clinical management.
This systematic review and meta-analysis aimed to comprehensively assess the diagnostic accuracy of fundus image-based deep learning models in the grading of diabetic retinopathy.
PubMed, Embase, Web of Science, and the Cochrane Library were systematically searched for relevant studies published up to October 28, 2025.
correctly flags nearly all eyes that truly have no diabetic retinopathy
In the simplified four-class classification task, sensitivities markedly improved across all grades: 96.85% (95% CI: 90.18%-99.93%) for stage 0, 92.94% (95% CI: 79.50%-99.72%) for stage 1, 92.75% (95% CI: 79.31%-99.61%) for stage 2, and 88.19% (95% CI: 68.99%-98.93%) for stage 3.
Deep learning exhibits high sensitivity and substantial potential for diabetic retinopathy grading, particularly in screening for no diabetic retinopathy and vision-threatening diabetic retinopathy.
Nevertheless, precisely differentiating between adjacent non-proliferative stages remains a clinical challenge.
The observed heterogeneity underscores the imperative for methodological standardization, rigorous external validation, and multimodal data integration.