The promise of artificial intelligence (AI) and the role it can play in helping doctors to detect and diagnose diabetic retinopathy—a common complication of diabetes that can lead to blindness—has the potential to become a real game changer. A recent study by eyecare nonprofit Orbis International has found that AI can accurately detect diabetic retinopathy in children and young adults, an important breakthrough that can mean the difference between healthy sight and irreversible vision loss for youngsters with diabetes.

The peer-reviewed study, published in Clinical Medicine Insights: Endocrinology and Diabetes, shows that Cybersight AI, a component of Orbis’s telemedicine and e-learning platform, can be an effective tool to support medical staff, who are often overburdened with patient caseloads, to care for children with diabetes, especially in low-resource settings with limited numbers of health care professionals.

“To date, AI has been studied to detect diabetic retinopathy in adults,” said Nicolas Jaccard, principal architect, telehealth and program technology for Orbis International. “These studies have shown that AI is highly effective and accurate, but almost none have been tested on children. Adults and children with diabetes both require regular eye screenings to detect diabetic retinopathy and keep the condition from progressing.

“However, trained ECPs cannot meet the growing demand for diabetic retinopathy screening as the prevalence of diabetes continues to rise. A tool like Cybersight AI can help to meet this burden, especially for children and young adults who have been shown are less likely to seek out routine eye screenings,” Jaccard said.

The study in Bangladesh screened more than 1,300 children and young adults between the ages of 3 and 26 diagnosed with diabetes. Each patient had images of their retinas taken on a fundus camera, which were then evaluated by Cybersight AI and by a fully qualified optometrist certified to grade for diabetic retinopathy. Results showed that Cybersight AI accurately detected any signs of diabetic retinopathy among children and young adults, despite the algorithms having been trained on adults.