News

United States Department of Veterans Affairs Publishes Research with Head-to-Head AI Performance Comparison Results for Diabetic Retinopathy Detection

January 6, 2021 – The Diabetes Care Journal just published a landmark study with patient data from the United States Department of Veterans Affairs (VA).  In this multicenter, noninterventional study on more than 300,000 retinal images from over 20,000 veterans at the Puget Sound and Atlanta VA Healthcare Systems, Aaron Y. Lee MD MSCI and his VA colleagues performed an independent, head-to-head validation to compare automated artificial intelligence (AI) diabetic retinopathy (DR) screening systems.  These data were from a real-world teleretinal service for diabetic patients undergoing routine diabetic retinopathy screening across 15 years.  Among the 7 AI algorithms included in this study, only one algorithm performed similarly to VA human graders by sensitivity and specificity without a single case of severe non-proliferative or proliferative diabetic retinopathy being missed in the arbitrated set.

Eyenuk is proud that its FDA-cleared EyeArt System was part of this VA study.  This is the second large-scale, independent, published head-to-head AI performance comparison study in which Eyenuk participated.  In 2016, the Health Technology Assessment (HTA) from the Moorfields Eye Hospital (London, UK) involving more than 20,000 consecutive patients was conducted to determine the screening performance and cost-effectiveness of multiple automated retinal image analysis systems. This study demonstrated that the EyeArt AI System delivered much higher sensitivity (indicative of patient safety) for DR screening than other automated DR screening technologies investigated and that its use is cost-effective alternative to the current, purely manual grading approach.

The study conclusion states, “The DR screening algorithms showed significant performance differences. These results argue for rigorous testing of all such algorithms on real-world data before clinical implementation.” With two independently published head-to-head AI performance comparison studies, strong results from the prospective multi-center pivotal clinical trial which led to FDA clearance, and multiple peer-reviewed publications with superior validation results, EyeArt continues to be the world’s most validated AI system for diabetic retinopathy detection.

#EyeArt  #Eyenuk  #diabetescare  #diabeticretinopathy  #artificialintelligence  #veteransaffairs

 

Share This Post