Researchers at Moorfields Eye Hospital and University College London have created an artificial intelligence system capable of diagnosing keratoconus, a condition that causes thinning and deformation of the cornea. Computers can now track the progression and development of the disease.
The authors of the method analyzed 36,673 optical coherence tomography (OCT) scans from 6,684 patients. Using only data from the patients' first visit, the AI accurately determined which patients' condition would worsen and which would remain stable.
The patients were then divided into two groups. The first group included two-thirds of low-risk patients, for whom observation was sufficient. The second group included one-third of high-risk patients, who required urgent preventive treatment—cross-linking. This is the only effective way to halt the progression of keratoconus. This method involves using ultraviolet light and vitamin B2 (riboflavin) drops to strengthen the cornea. This treatment has a high success rate in over 95% of cases. Early intervention can prevent irreversible corneal damage and the need for corneal transplants.
In the future, this AI could save the sight of thousands of people and reduce the burden on healthcare systems.
