Where is the FDA on AI?

In the 7th episode in our AI Conversation series, we tackle the BIG FDA question…the FDA’s stance on AI today. Join TeraRecon President and CEO, Jeff Sorenson, as he explains how the FDA has continued to show their dedication to fostering the development of AI in healthcare.

How does the FDA Classify Artificial Intelligence?

The FDA is really evolving rapidly in terms of artificial intelligence in healthcare. There have been a lot of landmark clearances, as of late, which are true AI based systems. The FDA has taken many actions to foster the development of artificial intelligence to really embrace the technology. The challenge has been the industry and the medical community itself.

The medical community is really looking for something different, when there is nothing different. What I mean by that is, they are getting together and looking for whole new ways of approving medical devices and whole new societies on how these things are approved. When, in fact, the FDA has been clear, and thankfully so, that AI engines are class-II medical devices, just like TeraRecon’s image processing solutions are class-II medical devices. That does mean it’s not class-I, so you can’t just be a company, register on a website, and then go to market. There are rules and processes for this that are very well understood. Companies know exactly how to do them. So, we know how to put AI responsibly into the radiology workflow, through careful version control, traceability, and all other things that medical device companies have to do.

We at TeraRecon bet very strongly on that trend. It is why we built the Northstar AI Explorer. Northstar is all about taking the results of AI and allowing the physician to be control, even to adjust those results, to understand them, to believe them, and to be in control of what’s persisted into the patient’s medical record. When we started this project in 2013, the FDA classification wasn’t known and that was a big risk for our company. We really believed that the FDA would insist on the physician standing in between computers and a diagnosis. That remains true today and we absolutely believe that is the right approach.