Inteligência artificial no rastreamento da retinopatia diabética

Diabetic retinopathy can lead to vision loss even when early retinal changes cause no symptoms. Artificial intelligence has therefore been investigated as a means of expanding retinal image screening. This article reviews eight scientific publications on diagnostic accuracy, implementation in healthcare services, and practical limitations. The findings indicate good performance for certain algorithms and higher screening completion when results are available during the visit. However, performance differs across systems, and detecting abnormalities does not in itself ensure access to specialized care. Artificial intelligence may strengthen screening when image quality is assessed and referral pathways are clearly established.