CHENNAI: A simple photograph of the retina, analysed using AI, may help detect diabetes and even early-stage pre-diabetes without the need for blood tests, according to a new study conducted among Indians and published in an international peer-reviewed journal.
Researchers from Yenepoya (Deemed to be) University, Madras Diabetes Research Foundation (MDRF) and Emory University have demonstrated that subtle changes in retinal blood vessels, particularly veins, can reliably distinguish people with diabetes from those with normal blood sugar levels.
The findings, published in Diabetes Technology & Therapeutics recently, point to a non-invasive, quick screening tool that could aid early diagnosis in large populations. The retina is the only part of the body where blood vessels can be viewed directly without surgery.
By applying AI-based image analysis to routine retinal photographs, the researchers identified microvascular patterns that are invisible to the human eye but strongly associated with diabetes. “Think of the eye as a window to the body,” said Dr Soujanya Kaup, lead author of the study and associate professor of Ophthalmology at Yenepoya University. “Using AI to read these tiny vascular clues, we can identify diabetes from a simple retinal image, without fasting or blood draws.”
The study analysed 273 high-quality retinal images from 139 participants, including individuals with normal blood sugar, pre-diabetes and diabetes. The AI model achieved a sensitivity of 95% in detecting diabetes, with particularly strong performance when analysing retinal veins rather than arteries.
Dr Sudeshna Sil Kar of Emory University, co-lead author, said: “The system is trained to detect specific geometric patterns in retinal vessels. It functions like a digital investigator, identifying disease-linked changes from a single image with high accuracy.”
Importantly, the analysis also detected early vascular alterations in people with pre-diabetes, a stage where lifestyle changes can delay or prevent progression to diabetes. “These subtle vessel changes appear even before full-blown diabetes sets in,” said Dr R Rajalakshmi, senior author, and Head of Medical Retina and Ocular Research at MDRF.
The researchers emphasised the public health relevance of the findings, particularly for India, which has over 100 million people living with diabetes, many undiagnosed. “A low-cost, non-invasive screening method using existing eye exams could be a game-changer,” said Dr V Mohan, chairman, MDRF. “But, bigger and multi-centre studies are needed before clinical rollout.”
Dr KM Venkat Narayan, Emory Global Diabetes Research Center, opined that the study strengthened the case for using the eye as an early indicator of systemic disease. The authors described the findings as proof of concept and called for broader validation, but said the approach holds promise for community-based screening, especially in resource-limited settings.