AI-driven screening could improve health equity for youth with diabetes
AI technology is taking the lead in transforming diabetes care, particularly in the realm of eye examinations, according to an intriguing study from Johns Hopkins Children’s Center (MD, USA). The research highlights the efficacy of autonomous AI-driven cameras in enhancing screening rates for potentially blinding diabetes eye diseases (DED) in children and youth with diabetes.
DED refers to a group of eye problems, but the most common is diabetic retinopathy: a complication of diabetes causing damage to blood vessels and nerve tissues in the retina. Diabetic retinopathy affects a substantial percentage of youth with diabetes (4–9% with type 1 diabetes and 4–15% with type 2 diabetes). Early detection of DED through frequent screenings is crucial in preventing its progression.
Traditionally, diabetic eye examinations require dilation for a detailed view of the retina, and a separate visit to an eye care professional, making these screenings inconvenient for numerous individuals. Although diabetes specialists recommend yearly screenings, research indicates that only 35%–72% of youth with diabetes actually undergo the recommended screenings, and care gap rates are notably higher among racial and ethnic minority youth.
The study focused on individuals under 21 with type 1 and type 2 diabetes and discovered that 100% of patients undergoing AI exams completed the eye assessment. Through simplifying the imaging process without the need for dilation, this AI-driven approach may address care disparities among racial and ethnic minority youth with diabetes, who have historically faced elevated rates of DED and limited access to routine eye damage screenings.
164 participants aged 8 to 21 years were enrolled, with 41% belonging to minority groups. They were randomized into two groups: one (83 participants) receiving traditional screening instructions and care, and the other (81 participants) undergoing the autonomous AI system diabetic eye exam during their endocrinologist visit. The latter group received their results in the same visit.
The AI system diabetic eye exam consisted of capturing four pictures of the eye without dilation. These images were then processed through an algorithm to ascertain the presence or absence of diabetic retinopathy. If diabetic retinopathy was detected, the participant was referred to an eye doctor.
Remarkably, 100% of patients in the AI group completed their eye exam the same day, compared to only 22% in the traditional screening group who followed up with an optometrist or ophthalmologist within 6 months. Additionally, there were no statistically significant variations based on gender, race or socioeconomic status for participants in the second group regarding their decision to arrange an additional screening appointment with an eye doctor.
“With AI technology, more people can get screened, which could then help identify more people who need follow-up evaluation,” explained Risa Wolf, a pediatric endocrinologist at Johns Hopkins Children’s Center. “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.”
While the AI used in the study is not FDA-approved for individuals under 21, the researchers emphasize the promising potential of integrating AI technology into diabetes care, addressing health equity concerns and preventing the progression of DED, as well as improving disease management.
