Researchers at Kings College London have built an AI model that can accurately predict who is at a high risk of developing sight-threatening diabetic retinopathy (DR) up to three years in advance.

Diabetes patients aged 12 and older in the UK are asked to attend an annual eye check, which tests for DR. Called the NHS Diabetic Eye Screening Program (DESP), approximately 3.2 million people are screened each year, and it costs the health service £85m per year in England, according to the announcement.

Image data from the UK’s National Health Service (NHS) was used to develop the tool, which can predict if someone is at low or high risk of developing sight-threatening DR in one year, two years or three years using images from the back of the eye.

To develop the new AI tool, researchers utilised over a million retinal images from diabetes patients in the Southeast London DESP. The model’s accuracy was validated using a dataset of 70,000 images from the INSIGHT health data research hub.

INSIGHT is an NHS-led initiative housing retinal images linked to clinical data. The images are routinely collected from patients in Moorfields Eye Hospital and University Hospitals Birmingham NHS Foundation Trust. The dataset includes contributions from over 200,000 patients enrolled in the Birmingham, Solihull and Black Country DESP, the largest urban screening programme in Europe.

The NHS said that if implemented, this AI-based technique could reduce the screening burden for people at low risk of vision loss, while ensuring individuals at a high risk of vision loss are seen urgently, saving the NHS millions of pounds, and thousands of appointments every year.

The research team now plans to conduct a prospective clinical trial to assess if the AI model is safe, efficacious, and cost-effective for use within DESP. The trial – which aims to include data from over 50,000 diabetes patients – could provide evidence to support the implementation of predictive AI in the NHS DESP.

In the announcement accompanying the study, Pearse Keane, director of INSIGHT and professor of artificial medical intelligence at UCL’s Institute of Ophthalmology, said: “This is an exciting use case for the value of curated NHS eye data in research for the benefit of patients and the wider healthcare system. Although there are AI models that can detect the presence of diabetic eye disease with the accuracy of a retinal specialist, this new model breaks ground in predicting the risk of developing sight-threatening diabetic eye disease up to three years in the future.”