Artificial intelligence can diagnose infectious keratitis as well as ophthalmologists, according to a new study.
- Infectious keratitis is an inflammation of the cornea that can lead to blindness if not managed properly.
- A study shows that AI is capable of diagnosing this eye disease effectively and safely.
- This technology could improve disease management, particularly where access to specialized eye care is limited.
Infectious keratitis is a serious infection of the cornea, the outer membrane covering the eye. The most severe cases can lead to blindness. Prompt treatment reduces the risk of serious complications. Artificial intelligence (AI) could be a solution to improve the diagnosis of this eye disease, according to a new study from the University of Birmingham published in the November 2024 issue of the journal eClinicalMedicine.
Diagnosis of keratitis: AI as effective as a doctor
To assess the value and effectiveness of artificial intelligence in the diagnosis of infectious keratitis, researchers reviewed 35 studies that used deep learning models to identify patients with the eye infection. . In total, more than 136,000 corneal images were screened by the AIs tested.
Analyzes showed that the technology had similar diagnostic accuracy to ophthalmologists. Indeed, it recorded sensitivity (capacity to detect a maximum number of patients while avoiding false negatives, Editor’s note) of 89.2% and specificity (ability to detect only sick people while avoiding false positives, Editor’s note) by 93.2%. With doctors, these rates were 82.2% and 89.6%, respectively.
AI models have also proven effective in differentiating between healthy eyes, infected corneas, and various underlying causes of disease such as bacterial or fungal origin.
Infectious keratitis: AI to improve access to care?
For Dr Darren Ting, lead author of the study, the various data compiled show that AI could help improve the management of infectious keratitis.
“Our study shows that AI has the potential to provide rapid and reliable diagnostics, which could revolutionize the way we manage corneal infections globally. This is particularly promising for regions where the Access to specialist eye care is limited and can help reduce the burden of avoidable blindness worldwide.”explains the scientist from the British university in a press release.
However, he recognizes that additional research and testing is needed to further increase the reliability of algorithms before their use in the clinical field.