Automated machine-learning models are revolutionizing the prediction of diabetic retinopathy progression, a groundbreaking study published in JAMA Ophthalmology reveals. Led by Dr. Paolo S. Silva and his team at the Beetham Eye Institute, the research showcases the accuracy of these models in identifying the risk of diabetic retinopathy progression through the analysis of ultra-widefield retinal images.
With a comprehensive dataset of 1,179 deidentified retinal images, captured using the cutting-edge California retinal imager (Optos), the study focused on individuals with mild nonproliferative DR (NPDR) and moderate NPDR. The findings highlighted the significant role of automated machine-learning models in estimating the risk of DR progression, a critical aspect of diabetic eye disease management.
The study reported impressive results, with the model accurately identifying 77.5% of eyes with mild NPDR and 85.4% of eyes with moderate NPDR that progressed at least two steps. Notably, the model successfully detected all eyes with mild NPDR and 85% of eyes with moderate NPDR that progressed within a year, reflecting its potential to transform clinical decision-making and improve patient outcomes.
Dr. Silva emphasized the importance of prospective validation and regulatory approval before implementing these AI models in clinical practice. Still, the study’s findings underscore the accessibility of machine-learning applications in addressing clinical needs and enhancing screening processes for individuals with diabetes. As technology continues to advance, these AI models offer a promising solution to enhance the early detection and management of diabetic retinopathy, ultimately safeguarding the vision and well-being of patients worldwide.
