Artificial intelligence and screening for visual impairment related to diabetic retinopathy and macular edema

We read with interest the article by Pérez et al. 1 titled "Visual impairment associated with diabetic retinopathy and macular edema: a hospital-based study," which aimed to estimate the prevalence of visual disability and associated ocular diseases in patients with type 2 diabetes; it emphasizes the need to optimize resources to promote screening for diabetic retinopathy.We would like to contribute the following:Artificial intelligence (AI) algorithms play a promising role in the areas of diagnosis, prognosis, and support in clinical decision-making.It has been demonstrated that AI algorithms can achieve high levels of diagnostic performance for the detection of diabetic retinopathy, and leveraging these technologies could help improve the quality of health services in low-and middle-income countries. 2 Wong et al. 3 analyzed to 352 patients classified based on their degree of macular edema determined through photographs obtained by optical coherence tomography.They found that their AI had a sensitivity rate of 81.5% and a specificity rate of 88.8% for interpreting images; however, images of diabetic macular edema of lesser severity with preserved architecture and minimal retinal thickening could be interpreted as normal.Peris et al. 4 conducted 2680 AI-based examinations, which provided an adequate diagnosis in 86.80% of patients and a result of insufficient quality in 13.20%.They point out that compared to manual classification by ophthalmologists, AI is faster and more accurate, has a sensitivity rate of 100% and a specificity rate of