AI-Driven Eye Exams May Increase Screening Rates Among Diabetic Youth

A research team from Johns Hopkins Children’s Center has found that an autonomous artificial intelligence (AI) tool to screen youth populations for diabetic eye disease (DED) can also increase the likelihood of screening completion, according to a recent study published in Nature Communications.

Specifically, the researchers hypothesized that the AI would increase screening and follow-up rates for diabetic retinopathy in a racially and ethnically diverse cohort of children and adolescents. The condition can occur when blood sugar levels are not well controlled, leading to damaged tissues and blood vessels in the retina. In turn, diabetic retinopathy can lead to permanent vision loss and blindness.

The research team noted that DED prevalence is typically lower in patients under the age of 21, but diabetic retinopathy impacts between four and nine percent of youth with type 1 diabetes and an estimated four to 15 percent of youth with type 2 diabetes.

Regular eye exams to screen for DED can help identify the presence of disease and facilitate treatment before the condition can progress, but barriers to screening exist.

The researchers indicated that experts recommend annual eye exams for diabetic youth. However, these exams often require a separate visit to an eye care provider and the use of eye drops to dilate the pupils prior to screening.

These factors may contribute to screening gaps, they posited, underscoring that anywhere from 35 to 72 percent of children and adolescents with diabetes receive these recommended screenings. The screening gap is larger among poor and minoritized youth.

Barriers to DED screening include lack of time, transportation, and access to specialists, alongside confusion and a feeling of inconvenience around the need for screening.

The research team’s previous work in the area of DED screening revealed that autonomous AI could successfully be used to diagnose diabetic eye conditions. The tool leverages a camera to take four pictures of the backs of the eyes without the need for dilation. These images are then fed to the AI, which analyzes them to determine whether or not diabetic retinopathy is present. If the disease is present, the patient can then be referred to an optometrist or ophthalmologist for follow-up.

To assess the AI’s impact on screening rates, the researchers gathered data from patients aged 8 to 21 years who received care at the Johns Hopkins Pediatric Diabetes Center between November 24, 2021, and June 6, 2022.

The cohort included 164 patients, 41 percent of whom were from minority groups, while 47 percent of the group had Medicaid insurance. Each participant was randomly assigned to one of two groups: the first, which contained 83 patients, received standard screening instructions and care, after which they were referred to either an optometrist or ophthalmologist for an eye exam; in the second group of 81 patients, each underwent a five-to-10-minute diabetic eye exam with the AI tool during a standard visit to their endocrinologists, during which they also received the results of the exam.

The research team found that all of the patients in the AI screening group completed their exams, while only 22 percent of those from the standard screening group followed up within six months to complete an eye exam with an eye care provider.

Upon further investigation, the researchers noted that there were no statistical differences in socioeconomic status, gender, or race in terms of whether participants in the second group scheduled a separate screening with an optometrist or ophthalmologist.

Further, approximately 31 percent of the AI group were identified as having DED.

These findings suggest that AI-driven DED exams could help close screening gaps.

“With AI technology, more people can get screened, which could then help identify more people who need follow-up evaluation,” said Risa Wolf, MD, a pediatric endocrinologist at Johns Hopkins Children’s Center who led the research, in a news release. “If we can offer this more conveniently at the point of care with their diabetes doctor, then we can also potentially improve health equity, and prevent the progression of diabetic eye disease.”