Is AI Regulated as a Medical Device?
What the Next FDA Commissioner Needs to Do To Support Innovation
By: Sherri Douville, CEO & Board Member at Medigram
Is software and in particular, AI classified as a medical device under FDA? Well, it’s confusing. It’s fair to say that AI “advises” a healthcare professional rather than “drives” for example, a medical device.
According to a March 8, 2019 article by Bradley Merrill Thompson in HiMSS mobihealthnews, two main events have occurred.
- In December 2017, FDA published a draft guidance building on 21st Century Cures legislation, providing the agency’s perspective on the clinical decision support software that FDA should regulate and what it should not. That guidance has profound implications for artificial intelligence used to guide healthcare decision-making.
- Throughout 2018, FDA announced a year-long process of completely revamping the agency’s approach to regulating software. FDA calls its new program the “Pre-certification Program.”
From an entrepreneur’s view, the proposed pre certification program unfairly advantages legacy companies. Not just in a way to their financial advantage, in a way I believe can be harmful to patients. For example, “pre-certified” companies could use their pre-certified status as a company to leverage past performance for a past class of product into a new, irrelevant class for which they have little known capability (desktop to mobile or consumer to medical etc.). The 21st Century Cures Act Sec. 3060 (a congressional statute) on Software moved some things pertaining to regulating software out of FDA and much of that has not been addressed in the current FDA website. Perhaps, HHS will take this opportunity to succeed the highly respected and effective outgoing commissioner, Dr. Scott Gottlieb with an excellent administrator with a different focus (a job that I could never do myself).
Perhaps the successor to the respected commissioner could harmonize what’s current to the law with what’s posted on the FDA website now in order to provide clarity to the market and unlock innovation. FDA could make it a real priority to update regulations and guidance for the half dozen or so outdated guidance documents that are currently on the website. In all fairness, FDA had published a new guidance document that describes the range of problems with the existing documents. To also be fair, some say that modifying regulations and guidance documents requires going through an extensive process that requires a cascade of approvals both within and outside the agency. That’s a really hard job that we can only hope someone special will be willing to do. Entrepreneurs and investors need to be able to plan and budget for our regulatory roadmaps. We shouldn’t defer to pre-certifying a handful of legacy companies and expect them to produce all the innovation we need. Our healthcare system and in particular, physicians need solutions to make their jobs easier and help relieve some of the enormous burdens contributing to their burnout. Let’s foster this vision by creating a highly competent, competitive, and entrepreneurial ecosystem.
FDA Update: April 2, 2019 courtesy of Bradley Merrill Thompson
“On Tuesday (April 2, 2019), FDA published its discussion paper on a “Proposed Regulatory Framework for Modifications of Artificial Intelligence/Machine Learning AI/ML based Software as a Medical Device.” To announce the proposal, Commissioner Gottlieb published a statement explaining that the purpose of the new framework is to allow FDA to approve medical software that includes adaptive machine learning, rather than insisting that sponsors lock their algorithms so that they don’t continually adapt or learn as the algorithm is used. FDA’s basic idea is that, when seeking approval, sponsors would come to the agency with a plan for monitoring and controlling the performance of their AI software over time, to make sure that it doesn’t degrade. FDA would clear or approve the device then based on the commitment by the sponsor to follow that plan.”
Statement from FDA Commissioner Scott Gottlieb, M.D. on steps toward a new, tailored review…
Artificial intelligence and machine learning have the potential to fundamentally transform the delivery of health care…
By Sherri Douville, CEO at Medigram, the Mobile Medicine company. Recognized in 8 categories of top CEOs by Board Room Media (Across SMS, mHealth, iOS, IT, Database, Big Data, Android, Healthcare). Top ranked medical market executive worldwide and #1 ranked in mobile technologycategories (mhealth, iOS, Android), #1–2 (on any given day) for the cybersecurity market in the U.S. on Crunchbase. Best selling editor/author, Mobile Medicine: Overcoming People, Culture, and Governance & Advanced Health Technology: Managing Risk While Tackling Barriers to Rapid Acceleration