"The first and only autonomous AI system authorized by the FDA"

On April 11, 2018, the U.S. Food and Drug Administration (FDA) permitted marketing of a medical device that utilizes artificial intelligence to diagnose eye disease in diabetic adults.  The device, called IDx-DR from the company IDx, screens patients for diabetic retinopathy using deep learning algorithms.  The screening involves standard retinal imaging, takes less than a minute, and can be performed without a clinician's interpretation of the images or results.  Accordingly, IDx-DR may be used in primary care offices for early detection of retinopathy, which can lead to vision impairment and blindness.

IDx founder Michael Abramoff, a retinal specialist at the University of Iowa, told Science News that the medical device is unique because it operates autonomously and there is no specialist "looking over the shoulder of the algorithm."  "It makes the clinical decision on its own."

The FDA based its decision to permit marketing of IDx-DR, in part, on a clinical trial of the system involving the screening recommendations and corresponding retinal images from over 800 diabetic patients at ten different primary care sites.  The study indicated that IDx-DR correctly identified the presence of more than mild diabetic retinopathy 87.4% of the time and correctly identified patients with less than mild diabetic retinopathy 89.5% of the time.  The former group was referred to a specialist and the latter group was scheduled for a checkup in 12 months.

Tweets from FDA Commissioner Indicate AI Push

On the same day as the IDx-DR announcement, FDA Commissioner Scott Gottlieb M.D. unleashed a string of tweets that strongly endorsed AI-based health care solutions.  Based on Dr. Gottlieb's comments, it appears that the FDA may be currently accelerating review of AI-based systems.  Looking ahead, Gottlieb also signaled that AI-specific fast track regulatory pathways are under development.

Despite such relaxed AI regulations at the FDA, however, Dr. Gottlieb tweeted that the agency's approach would maintain consumer safety by establishing "appropriate guardrails."  Based on Dr. Gottlieb's comments, it will be interesting to track the FDA approval rate and timeline of AI-based healthcare solutions.  Additionally, in the case of new AI-specific regulatory pathways, time will tell what "guardrails" the FDA will put in place to maintain adequate safety as future AI innovations are made available to patients and providers.  Data privacy, algorithmic transparency, and "unbiased" learning datasets are just a few broad areas in which the FDA might regulate artificial intelligence as applied to health care.

For additional information regarding this topic, please see:

• In the future, an AI may diagnose eye problems (Science News, March 4, 2018)
• IDx-DR (IDx LLC Website)
• FDA permits marketing of artificial intelligence-based device to detect certain diabetes-related eye problems (FDA News Release, April 11, 2018)
• Scott Gottlieb, M.D. – Commissioner, U.S. Food and Drug Administration (Twitter: @SGottliebFDA)