14-OR: Autonomous Artificial Intelligence Diabetic Eye Exams Mitigates Disparities in Screening Completion for Youth

Diabetic retinal disease (DRD) is a complication of diabetes mellitus that can lead to vision loss; early detection through screening and treatment can prevent this. Few individuals with diabetes meet recommended DRD screening guidelines, and racial/ethnic minority youth are even less likely to undergo recommended screening. We sought to determine if implementing point of care (POC) autonomous artificial intelligence (AI) screening could mitigate disparities in diabetic eye exam completion. In ACCESS2, a preregistered prospective pre-post study design, youth with type 1(T1D) and type 2 diabetes (T2D) meeting American Diabetes Association criteria for needing a diabetic eye exam underwent POC autonomous AI diabetic eye exams at routine diabetes clinic visits. Completion rates of diabetic eye exams were compared prior to and after implementation of autonomous AI. A total of 380 youth with T1D (71.1%) and T2D (28.9%) were enrolled, mean age 15.2y, 46.8% Non-Hispanic (NH) White, with mean duration of diabetes of 6.0y, and median hemoglobin A1c of 8.1%. A greater percentage of NH White participants reported any prior diabetic eye exam (81.6 v 62.0%, p<0.001) compared to non-white and Hispanic participants. Multivariable analysis demonstrated that the strongest predictor for prior eye exam was diabetes duration. After undergoing POC autonomous AI diabetic eye exams, completion rates were 98% among non-white and Hispanic participants, and 99% among white participants. Introducing autonomous AI at the point of care enhances the accessibility and completion rates of diabetic screening eye exams, and promotes health equity for minority youth with diabetes.