AI Could Help Prevent Diabetes-Related Sight Loss by Revolutionizing Eye Screening
Diabetes is known to cause a range of complications, with diabetic retinopathy—damage to the blood vessels in the retina—being one of the most serious, potentially leading to vision loss. Terry Quinn, a diabetes patient from West Yorkshire, experienced this firsthand, struggling with vision deterioration that eventually led to the loss of his driving privileges and difficulty recognizing faces.
While eye screening for diabetic retinopathy is crucial, many patients don’t receive it regularly, especially in the US where barriers such as cost and accessibility prevent timely tests. “There’s clear evidence that screening prevents vision loss,” says Roomasa Channa, a retina specialist at the University of Wisconsin-Madison.
Currently, screening involves taking images of the eye’s fundus, but the process of interpreting these images manually is time-consuming and repetitive. Experts believe artificial intelligence (AI) could make these screenings faster and more affordable, especially since diabetic retinopathy progresses in distinct stages that AI can learn to detect.
Retmarker, a health technology company based in Portugal, has developed an AI-powered system that identifies potential issues in fundus images and sends them to human specialists for further review. AI tools like Retmarker and Eyenuk’s EyeArt have shown promise, with acceptable sensitivity and specificity rates for detecting diabetic retinopathy.
However, challenges remain, particularly concerning the quality of the images fed into AI systems. Google Health researchers found that AI models performed inconsistently when images were less than ideal, such as when camera lenses were dirty or lighting was poor.
Despite these challenges, AI’s cost-effectiveness is a key consideration, especially in low-resource areas. In Singapore, a hybrid model using AI for initial screening has proven to be more affordable than fully manual or fully automated approaches. The hybrid model is expected to go live in Singapore’s health service platform by 2025.
While AI may offer significant benefits, there are concerns about health equity, particularly in low-income countries where access to eye care is limited. Bilal Mateen, chief AI officer at PATH, emphasizes that the future of AI in healthcare should prioritize global accessibility, not just effectiveness in wealthy countries.
Experts like Dr. Channa remain hopeful, acknowledging that AI has the potential to revolutionize diabetes-related eye care. “Given the burden of diabetes, this is a really potentially great solution,” she says. However, the need for comprehensive eye care and screening for other conditions, such as glaucoma and myopia, should not be overlooked as AI continues to evolve.
