Automated Machine Learning for Diabetic Retinopathy Progression

Using deep learning models to detect and grade eye diseases from color fundus photography, including diabetic retinopathy (DR), has proven feasible, and some constructs have US Food and Drug Administration approval. Potentially expanding on this concept, Silva et al demonstrate an automated machine learning (ML) model from ultra-widefield (UWF) retinal images for assessing the risk of progression for mild or moderate nonproliferative DR (NPDR). Using approaches such as this could be important as the number of persons with DR increases dramatically in the coming decades.
Source: JAMA Ophthalmology - Category: Opthalmology Source Type: research