Diabetic retinopathy (DR) is an important cause of vision loss in working-age adults, affecting around one in three of the 463 million people with diabetes worldwide. Professor of Retinal Research and Consultant Vitreoretinal Surgeon Timothy Jackson and his team recently completed a study in which they developed deep learning (DL) models, a type of artificial intelligence, to predict when DR would reach a sight-threatening stage up to three years in the future.
Using retinal images and participant characteristics data from the southeast London DR screening service, the DL models were able to accurately predict when retinal images, which were not part of the model training dataset, would progress to sight-threatening disease over one, two and three years.
Our grant will now enable Professor Jackson to test his findings against a different geographic group, distinct from the southeast London population. Once evaluated against a broader population base, these results have the potential to reduce the risk of sight loss for the four million people who attend DR screening across the UK every year.
Not only that, fully developed DL models could enable accurate, individualised risk predictions for people with diabetes who regularly attend DR screening services. Studies have shown that individualised follow-up, which can be enabled by the developed DL models, could reduce appointment needs by 43% and save 20% in screening costs. This would result in fewer unnecessary visits for people who are at low risk of DR progression, and free up resources for high-risk individuals, such as closer monitoring and earlier treatment.
