Researchers from the University of Tennessee Health Sciences Center in Memphis have trained an artificial intelligence (AI)-enabled robot to successfully perform glaucoma assessments and interact with ophthalmologists and patients, according to data presented at ARVO 2023 in New Orleans. Siamak Yousefi, PhD, and colleagues are training Alborz, a humanoid robot, to assist ophthalmologists in screening and diagnosing various ocular disorders based on fundus photographs, ocular coherence tomography images, and visual fields, he said in an interview with PracticeUpdate.

The team programmed Alborz to identify glaucoma from retinal fundus photographs and visual fields. They developed a probabilistic deep convolutional neural network model based on 1,851 fundus/visual field pairs and validated the model using an independent dataset of 196 fundus/visual field pairs. The area under the curve of Alborz to detect glaucoma from fundus photos and/or visual field data was >0.97 (95% CI, 0.93-0.99). Dr. Yousefi explained that Alborz may reduce burden on ophthalmologists and could be used in rural and underserved areas.