A single retinal scan could replace years of genetic testing for people with inherited sight loss. Eye2Gene is an artificial intelligence algorithm that predicts which of over 300 disease-causing genes is responsible for a patient’s retinal degeneration, using only an image of the back of the eye. The problem is stark. Inherited retinal diseases are a leading cause of vision loss in working-age adults, but getting a genetic diagnosis currently takes more than five years and costs £10,000 per patient. More than 40 per cent of patients—over 10,000 people in the UK—remain undiagnosed. Without knowing the specific gene, clinicians cannot offer personalised treatment plans, predict disease progression, or advise on family planning. This grant funds the next phase: turning Eye2Gene from a research tool into a certified medical device. The team will validate the algorithm on independent datasets from three external sites—Oxford, Liverpool, and Tokyo—and add explainability by identifying specific abnormalities in retinal scans. A health economic evaluation will assess cost savings for the NHS. If successful, Eye2Gene could cut diagnosis time from years to minutes, reduce costs dramatically, and give equitable access to genetic testing across the country. For patients, that means faster entry into clinical trials and emerging treatments. For the NHS, it means fewer undiagnosed patients and reduced economic burden.
View original technical description
Background Inherited retinal disease (IRDs) are a leading cause of visual impairment in the working age population. Mutations in over 300 genes are associated with IRDs and a genetic diagnosis, which involves identifying the affected gene in a patient, is the first step towards care and management of the patient. However, genetics diagnosis is currently slow, expensive and not widely accessible: Average time to diagnosis is over 5 years and costs £10,000 per patient. More than 40% of patients are undiagnosed (>10,000 individuals). Aims_and_Objectives We aim to address the evidence gap in IRD diagnosis with an AI algorithm, Eye2Gene, to accelerate and democratise the IRD diagnosis service. Currently Eye2Gene predicts the gene causing the disease from a retinal scan, with a top-5 accuracy of 88% on our internal dataset. As part of this Phase 2 grant we plan to develop Eye2Gene into a medical device by: Validating Eye2Gene on independent datasets from three external sites (Oxford, Liverpool and Tokyo) Providing explainability by identifying specific abnormalities (IRD-specific features) in retinal scans. Work_plan_and_t imelines_for_delivery Year 1 External site onboarding Extension of a deep-learning segmentation algorithm to segment IRD-specific features Datasets preparation for benchmarking against "best-practice" Year 2 Validation in external sites Health economic evaluation Year 3 Application to variant proritisation Regulatory approvals Commercial exploitation Anticipated_Impact_and_Dissemination By increasing the diagnostic rate for IRDs at a decreased cost, and by offering equitable access to a genetic diagnosis, the anticipated impact for patients are: Improved health outcomes Personalised treatment plans (emerging treatments or clinical trials) Better understanding of the condition and its heritability for family planning For the NHS: Improved operational efficiency Increased genetic diagnostic rate at eye hospital. Reduced economic burden Dissemination channels for healthcare professionals will include: High-impact publications Conferences Research networks Professional bodies For patients and public: Patient day Charities Lay summaries Media
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know