Michael S. Ip
KOLOA, Hawaii — Artificial intelligence can play a role in meeting the significant need for diabetic retinopathy screening in the US, where more than 30 million adults have the disease and only half of them obtain yearly eye screenings, Michael S. Ip, MD, said in a presentation here at Retina 2020.
The EyeArt AI screening and referral workflow system (Eyenuk), which is limited to investigational use in the US, has undergone “rigorous clinical validation” in retrospective studies, according to Ip. Of more than 100,000 consecutive patient visits, the software sensitivity for the identification of referable diabetic retinopathy was 91.3%, specificity was 91.1%, and sensitivity for detecting treatable DR was 98.5%, Ip said.
In the most recent study, a prospective pivotal clinical trial, the objective was to evaluate the EyeArt system on 2-field non-mydriatic fundus photographs to detect referable DR and vision-threatening DR, using the clinical reference standard of standardized, adjudicated ETDRS grading of 4-widefield stereoscopic mydriatic fundus photographs (equivalent to 7-field ETDRS photographs).
“The EyeArt AI screening system for DR in this prospective clinical trial had high sensitivity and specificity compared with standardized grading at the University of Wisconsin fundus photograph reading center where they evaluated 4-widefield stereo images and met its predetermined sensitivity and specificity endpoints (P < .0001),” Ip said.
Furthermore, the system can enable point-of-care DR screening making it more accessible and obviating the need for trained personnel to evaluate fundus images to save time and financial resources, he said. – by Patricia Nale, ELS
Reference: Ip MS. Artificial intelligence screening for diabetic retinopathy: Results from the EyeArt pivotal multicenter, prospective, clinical trial. Presented at: Retina 2020; Jan. 19-24, 2020; Koloa, Hawaii.
Disclosure: Ip reports he is a consultant for Alimera, Allergan, Amgen, Astellas, Boehringer Ingelheim, Genentech, Novartis, Omeros, Quark and Thrombogenics, and receives research support from Biogen, Clearside, Genentech and Novartis.