By G. Mudalige, Jadetimes Staff
G. Mudalige is a Jadetimes news reporter covering Technology & Innovation
Artificial intelligence (AI) is emerging as a promising tool in the fight to prevent diabetes-related vision loss, particularly through early detection of diabetic retinopathy. Diabetic retinopathy is a leading cause of blindness, caused by damage to the retina's blood vessels due to prolonged high blood sugar levels. Vision loss can be devastating, as experienced by Terry Quinn from West Yorkshire, who lost his sight despite undergoing laser treatments and injections to control the condition. Like many individuals living with diabetes, he was unaware of the severity of the risk to his vision until it was too late. His story highlights the critical importance of early detection and intervention for diabetic retinopathy, a problem that AI-based tools aim to address.
Diabetic eye screening is recommended regularly for individuals with diabetes to catch complications early. In the UK, patients are invited for screening every one to two years, while in the US, annual screenings are advised for adults with type 2 diabetes. Despite these recommendations, many people fail to get screened due to barriers such as cost, inconvenience, and lack of access to healthcare facilities. Roomasa Channa, a retina specialist from the University of Wisconsin-Madison, stresses that timely screenings can prevent vision loss. However, manually interpreting the eye images used for screenings, known as fundus images, is time-consuming and repetitive for medical professionals. AI technology has the potential to simplify this process, reduce costs, and make screenings more widely accessible.
AI algorithms can analyze fundus images to detect the stages of diabetic retinopathy and determine whether a patient requires further examination by a specialist. For instance, Retmarker, a Portugal-based health technology company, has developed an AI-powered system to identify concerning images and refer them to human experts for verification. This hybrid approach combines the speed of AI with the precision of human judgment, reducing the burden on healthcare professionals while maintaining accuracy. Independent studies indicate that tools like Retmarker Screening and Eyenuk's EyeArt achieve acceptable rates of sensitivity and specificity. Sensitivity measures a test's ability to detect disease, while specificity measures its ability to confirm its absence.
However, challenges remain. AI systems often rely on pristine images for accurate analysis, which may not reflect real-world conditions like poor lighting or dirty lenses. Google Health, which developed an AI model for diabetic retinopathy detection, found its performance varied significantly during trials in Thailand due to such inconsistencies. To address these issues, Google has been working to refine its model and is now partnering with Thailand and India to assess its cost-effectiveness and integrate it into local healthcare systems.
Cost is another significant factor. AI tools can offer affordable screenings, with Retmarker estimating costs of around €5 per scan, though prices vary based on location and volume. Singapore, for example, has successfully implemented a cost-effective hybrid screening model, where AI performs initial analysis and human specialists validate results. This model will be rolled out in the Singapore Health Service in 2025, showcasing how robust healthcare infrastructure can make AI a viable solution for diabetic retinopathy.
Despite its promise, AI's benefits are not yet universally accessible. Bilal Mateen, Chief AI Officer at health NGO PATH, emphasizes the need for equitable implementation, ensuring AI tools are not limited to high-income countries. Middle- and low-income regions often lack the infrastructure and resources to adopt AI technology, widening the global health equity gap. Dr Channa echoes this concern, stressing the need for expanded access to eye care in underserved areas and encouraging ongoing attention to other eye conditions like glaucoma and myopia, which remain harder for AI to detect.
For individuals like Terry Quinn, who experienced the life-altering consequences of diabetic retinopathy, AI offers hope for earlier intervention and prevention. “If AI had existed for earlier detection, I’d have grabbed it with both hands,” Quinn says. As diabetes rates rise worldwide, AI-driven eye screening systems could transform care by making timely and affordable screening a reality for millions, ultimately saving vision and improving lives.
Comentarios