AI Eye Scan Could Transform Diabetes Screening in India, Study Shows

Is

there any other non-invasive way to detect diabetes early, other than pricking your finger for your blood? The answer may soon be yes. In a breakthrough for early diabetes detection, researchers have developed an AI-powered retinal screening tool that may be able to identify the metabolic disorder using a single high-resolution photograph of your eye. The study, conducted by researchers from Yenepoya (Deemed to be) University, Madras Diabetes Research Foundation, and Emory University, has been published in the journal Diabetes Technology and Therapeutics. The retina is the only place in the body where doctors can directly view live blood vessels without surgery or tools.

How is the detection through the retina done?

According to the researchers, for analysing retinal photographs, Artificial Intelligence was used to study subtle features of the vessels. It was found that veins show measurable changes in those who have diabetes, a chronic condition marked by persistently high blood sugar levels.

Why does the retina hold the key to diabetes detection?

The retina is located at the back of the eye and is the only place in the body where blood vessels can be observed directly and non-invasively. And so, analysing these vessels, doctors can gain insights into systemic diseases like diabetes. “Think of the eye as a window into the rest of the body,” said lead author Dr Soujanya Kaup, Associate Professor of Ophthalmology at Yenepoya University and Adjunct Associate Professor of Global Health at Emory University. “With AI reading tiny clues in this window, diabetes can be detected through a quick retinal photo,” he added.

How does AI spot diabetes from a single eye image?

The research team trained an AI model to analyse specific vein patterns, shapes, and structural changes using retinal photographs from people with and without diabetes. According to co-lead author Dr. Sudeshna Sil Kar from Emory University, the AI acts like a “digital detective,” achieving 95 per cent accuracy in identifying diabetes from just one retinal image. Importantly, the system uses routine retinal images already captured during standard eye exams, making it highly practical for real-world screening. One of the most significant findings is that the AI tool could detect early vascular changes linked to prediabetes, a stage where lifestyle changes can still prevent or delay full-blown diabetes. “We observed subtle alterations in eye blood vessels that begin even before diabetes develops,” said senior author Dr. R. Rajalakshmi, Head of Medical Retina and Ocular Research at Dr. Mohan’s Diabetes Specialities Centre.

Can lessen India’s diabetes burden

India has more than 100 million people living with diabetes - many of whom remain undiagnosed. According to Dr V. Mohan, Chairman of Dr. Mohan’s Diabetes Specialities Centre, AI-based retinal screening could transform population-level diabetes detection. “If validated in larger populations, this technology could enable real-time, mass screening using simple eye photographs,” he said. The tool also requires no expensive lab infrastructure, making it especially useful in low-resource and rural settings.Also read: There’s Now A Type 5 Diabetes And Doctors Say It’s Been Ignored For Decades, Has Unusual Symptoms

What are the key findings of the study?

• 95 per cent accuracy in detecting diabetes using retinal images
• Early detection of prediabetes
• Non-invasive, needle-free screening
• Uses existing eye exam images; no extra procedures needed

Experts say this research strengthens the role of AI-driven eye screening as an early warning system for chronic diseases, potentially reshaping how the disease is diagnosed across the world.