The AI technology uses a Deep Learning System (DLS), which thinks and makes decisions like human intelligence, to automatically assess retinal photographs of patients with and without diabetic eye conditions. The architecture of the DLS consists of a convolutional neural network to implicitly recognize characteristics of referable diabetic retinopathy, possible glaucoma suspect, and age-related macular degeneration from the appearance in retinal images. Training of the DLS entailed exposure of multiple examples of retinal images (with and without each of the 3 conditions) to the neural networks, allowing the networks to gradually adapt their weight parameters to model and differentiate between conditions.
Once the training was completed, it was tested and validated to detect the said eye conditions using separate datasets, including 10 external datasets from multi-ethnic populations across different countries. The 30 co-investigators used the world’s first and largest dataset consisting of nearly 500,000 retinal images to evaluate the use of the DLS.