The U.S. Food and Drug Administration (FDA) has provided the clearance of a Cloud-based, Artificial Intelligence system that is capable of diagnosing diabetic retinopathy utilizing retinal images highlights the potential for deep learning and algorithmic analysis to help and replace diagnosticians in medical tests.
The FDA’s clearance of the Coralville, Iowa-based IDx Technologies’ IDx-DR AI diagnostic system, showed that how image-based AI systems might assist clinicians, anatomic pathologists, and other care providers diagnose disease and lead therapy decisions. The device, already in use at the University of Iowa Hospitals and Clinics (UIHC), utilizes Cloud Computing and algorithms to autonomously evaluate images of the retina for marks of diabetic retinopathy. It enables IDx-DR to offer a screening decision about 20 seconds after image capture. The IDx-DR system delivers a twofold outcome. The system, when signs of diabetic retinopathy are present, proposes a follow-up with an ophthalmologist. The system suggests a follow-up screening in one year if it detects no signs of the condition.
According to a clinical study including 900 participants released in Nature, showed that an analogous AI system gained 87.2 percent sensitivity and 90.7 percent specificity in the identification of diabetic retinopathy, more than pre-specified primary endpoint goals. The ability to arrive at a diagnosis, not including a clinician already holds the potential to radically impact the workflows and services of medical laboratories and other diagnosticians. While describing IDx Technologies integrated IDx-DR with the EHR (Electronic Health Record) systems of UIHC, they also illustrated that how AI might power data interfacing.