Smart Scan: AI Identifies Keratoconus Before Vision Fades

In a world where artificial intelligence (AI) is reshaping nearly every industry, from finance to agriculture, healthcare remains one of its most promising frontiers.

Now, researchers in Europe have successfully used AI to predict which young patients with a progressive eye disease—keratoconus—are at risk of vision loss and need urgent treatment.

The study, presented at the 43rd Congress of the European Society of Cataract and Refractive Surgeons (ESCRS), is being hailed as a game-changer in the field of ophthalmology.

Led by Ophthalmologist Dr. Shafi Balal and his team from Moorfields Eye Hospital NHS Foundation Trust and University College London (UCL), the study has the potential to transform not only eye care practices in the West but also in countries like India, where the burden of undiagnosed keratoconus is significantly high.

Keratoconus is a progressive thinning and bulging of the cornea, the transparent front layer of the eye, which distorts vision and can lead to blindness if untreated. Though manageable in early stages, diagnosis is often delayed—leading to severe complications, particularly in young, working-age populations.

India, with over 2% of its population estimated to be affected by the condition, faces a dual challenge: lack of awareness and limited access to specialised diagnostics in many parts of the country. Often, the disease is diagnosed only after the damage has become irreversible.

That’s where AI steps in.

The AI model developed by Dr. Balal’s team analysed over 36,000 optical coherence tomography (OCT) scans from nearly 6,700 patients. Using this data, it could accurately identify which patients would see their condition deteriorate—and who would remain stable—from their first clinical visit itself.

“Two-thirds of patients were classified as low-risk, needing only monitoring, while the remaining third were identified for immediate treatment,” Dr. Balal said.

This new research reaffirms that AI is no longer confined to labs and prototypes but is now entering the consultation rooms—and even the retina—of modern medicine.

“In keratoconus, the cornea—the transparent front part of the eye—bulges outward, leading to visual impairment,” explained Dr. Balal. “It remains the leading cause of corneal transplants in the Western world, particularly in young, working-age individuals.”

Until now, ophthalmologists were unable to predict which patients would need intervention, relying instead on years of repeated monitoring. This delay often meant that treatment—a one-time, non-invasive procedure called corneal cross-linking—came after irreversible damage.

Using AI trained on 36,673 optical coherence tomography (OCT) images from 6,684 patients, the researchers developed a model that could, with impressive accuracy, predict disease progression from a single clinic visit.

“With AI, we could categorise nearly two-thirds of patients as low-risk, sparing them from unnecessary hospital visits, and the remaining third as high-risk who require prompt treatment,” Dr. Balal noted.

When data from a second visit was added, the accuracy of predictions rose to 90%.

The cross-linking procedure, which uses ultraviolet light and vitamin B2 (riboflavin) to strengthen the cornea, is successful in over 95% of cases when performed early. By deploying AI to predict who needs the procedure early, the researchers believe many patients can avoid corneal transplants and long-term vision loss.

Dr. Balal emphasized that the algorithm could streamline healthcare delivery, allowing overstretched eye care professionals to focus resources where most needed. “This is the first study of its kind to reach this level of predictive accuracy for keratoconus, and it uses one of the largest patient datasets monitored over two years,” he said.

The AI model is currently limited to data from a single imaging device, but researchers plan to expand its application across platforms after further safety testing.

The research team is now working on a more advanced AI system, trained on millions of eye scans, which can extend its predictive abilities beyond keratoconus—to eye infections, inherited eye diseases, and more.

As AI cements its role in diagnostics and treatment across sectors, this breakthrough is a reminder of how human lives—and vision—can be preserved through digital precision.

“For a teenager facing vision loss, this technology could mean the difference between a life of limitations and one of possibility,” said Dr. Balal.