AI can find patients who need treatment to prevent vision loss in young adults

Using artificial intelligence (AI), researchers predict which patients will need treatment to stabilize the cornea and maintain their vision, published in a study published in the European Association of Refractive Surgeons (ESCR).

This study focuses on the people of Keratconus. Keratconus is a visual impairment that commonly occurs in teens and young people and tends to worsen during adulthood. It affects up to 1 in 350 people. In some cases, the condition can be managed with contact lenses, but other conditions may deteriorate quickly and if not treated, the patient may need a corneal transplant. Now, the only way to tell who needs treatment is to monitor patients over time.

Using AI, the researchers evaluated patient eye images in combination with other data, successfully predicting which patients would need rapid treatment and which would continue to monitor.

This study was by Dr. Shafibalal and colleagues of Moore Fields Eye Hospital NHS Foundation Trust in London and University College London (UCL) in the UK. He said, “The cornea, the window in front of the eyes, bulges outward in people with the cornea.

“A single treatment called “crosslinking” can halt disease progression. When done before permanent scarring occurs, crosslinking prevents the need for corneal implantation. However, doctors are currently unable to predict which patients will progress and need treatment, and which patients are stable with just surveillance.

“This means that patients need to be monitored frequently over the years, and usually crosslinked after progression has occurred.”

This study involved a group of patients referred to the Moorfields Eye Hospital NHS Foundation Trust for evaluation and monitoring of Keratoconus. Using AI, the researchers studied 36,673 OCT images of 6,684 patients along with other patient data.

AI algorithms can accurately predict whether a patient's condition will worsen or remain stable, using images and data from the first visit only.

Using AI, researchers were able to classify two-thirds of patients into a low-risk group. The low-risk group did not require treatment. If information from a second hospital visit is included, the algorithm can successfully classify up to 90% of patients.

In crosslinking treatments, ultraviolet rays and vitamin B2 (riboflavin) are used to stiffen the cornea, and have been successful in over 95% of cases.

“Our research shows that using AI can predict which patients will need treatment and which patients will continue monitoring. This is the first study to obtain this level of accuracy to predict the risk of keratconos progression from a combination of scans and patient data, using a large cohort of patients monitored for more than two years.”

“This study is limited to the use of one specific OCT device, but the research methods used and AI algorithms can be applied to other devices. The algorithms can undergo further safety testing before they are deployed in a clinical environment.

“Our results mean that high-risk keratitis patients can receive preventative treatment before their condition progresses, preventing vision loss and avoiding the need for corneal transplant surgery with associated complications and recovery burdens.

“Low-risk patients avoid unnecessary frequent monitoring and free up medical resources. Effective patient sorting with algorithms allows specialists to be redirected to areas where they are most needy.”

Researchers are currently developing more powerful AI algorithms trained with millions of popsicles, such as predicting the progression of keratconos, trained for a specific task, including other tasks such as eye infections and hereditary eye diseases.

“The keratconos is in a manageable state, but when will patients hire patients, they need invasive implants or implants,” said Dr. José Luis Güell, the ESCRS trustee and head of the cornea, and the Refractive Surgery Department of Microcaryugia Ophthalmology, Barcelona, ​​Spain, in the Department of Cataract and Refractive Surgery.

“This study suggests that AI can be used to predict who will progress even from the initial routine consultation, meaning that patients can be treated early before progression or secondary changes.

“It can similarly reduce unnecessary surveillance of patients with stable conditions. If it consistently demonstrates its effectiveness, this technique ultimately prevents vision loss and more difficult management strategies in younger working-age patients.”

Provided by the European Cataract Association and Refractive Surgeons