Diabetic retinopathy (DR) and diabetic macular edema (DME) are debilitating complications of diabetes; however, patients in the most vulnerable populations often face barriers in accessing recommended ocular exams, leading to major health inequity. These barriers are especially true for FQHCs, as they serve the most under resourced and underserved populations. More recently, rigorously validated, FDA cleared, autonomous AI for the detection of DR and DME has been considered as an alternative to deployed remote reading networks. We examined the impact of autonomous AI compared to remote reading on patient access and clinical workflow efficiency, measured as number of steps by the operator. During 90 days in mid-2021, at two FQHC clinics in the southeastern US, autonomous AI was provided in place of a remote reading network. All steps were completed at the point-of-care: qualified patients with diabetes were identified in the electronic health record (EHR) , diagnosed by the autonomous AI output, provided results at time of the exam, and referred to specialists where appropriate. After 90 days, 1patients with diabetes underwent autonomous AI diagnostic testing, and 30/1patients with diabetes (28.6%) were identified by the AI as having signs of vision threatening DR or DME. Testing was accomplished with fewer steps as compared to exams completed with a remote reading network at the same clinics. We conclude that implementing autonomous AI point-of-care testing for the identification of DR and DME in an FQHC setting improved workflow efficiency and patient access compared to remote reading, demonstrating a potential to address health inequities in underserved and vulnerable populations.
J.Goldstein: Employee; Digital Diagnostics. D.Weitzman: Employee; Digital Diagnostics. M.D.Abràmoff: Advisory Panel; NovaGo Therapeutics AG, Board Member; Digital Diagnostics, Consultant; Digital Diagnostics, Speaker's Bureau; AbbVie Inc., Stock/Shareholder; Digital Diagnostics.