FDA released a discussion paper and request for feedback that explores a tailored approach to reviewing technologies that use advanced artificial intelligence algorithms.
FDA Commissioner Scott Gottlieb is making the most of his final week at the agency.
In the month that has passed since Gottlieb rattled the medical device industry with news of his impending resignation, the commissioner has issued 18 public statements pertaining to nearly all corners of the agency's realm, from food, tobacco, and cosmetics to drugs and devices. Friday is Gottlieb's last day on the job.
On Tuesday, Gottlieb said the agency will consider a new regulatory framework for reviewing medical devices that use advanced artificial intelligence algorithms. AI has been making headlines in medtech for a while now, and this is certainly not the first time FDA has turned its attention to how AI-based medical devices should be regulated. Currently, most of these technologies fall under the clinical decision support software classification with FDA, but there is a significant amount of gray area where AI-based medical devices are concerned, as one regulatory affairs expert pointed out in an exclusive MD+DI article last year.
"The ability of artificial intelligence and machine learning software to learn from real-world feedback and improve its performance is spurring innovation and leading to the development of novel medical devices," Gottlieb said.
AI algorithms are already being used to aid in screening for diseases and to provide treatment recommendations, he pointed out. Last year FDA authorized an AI-based device developed by Coralville IA-based IDx-DR for detecting retinopathy, an eye disease that can cause vision loss. The agency also authorized San Francisco-based Viz.ai’s Contact application, a type of clinical decision support software designed to analyze CT results that could notify providers of a potential stroke in their patients.
Gottlieb said the authorization of these technologies was a "harbinger of progress" that FDA expects to see as more medical devices incorporate advanced AI algorithms to improve their performance and safety.
"Artificial intelligence has helped transform industries like finance and manufacturing, and I’m confident that these technologies will have a profound and positive impact on healthcare," Gottlieb said. "I can envision a world where, one day, artificial intelligence can help detect and treat challenging health problems, for example by recognizing the signs of disease well in advance of what we can do today. These tools can provide more time for intervention, identifying effective therapies, and ultimately saving lives."
The first step toward developing a tailored approach to reviewing these technologies comes in the form of a 20-page discussion paper. Down the road, the industry can expect to see draft guidance based on input the agency receives on the paper.
"We plan to apply our current authorities in new ways to keep up with the rapid pace of innovation and ensure the safety of these devices," Gottlieb said.
He noted that the AI-based technologies that have already been authorized and cleared for marketing are generally called locked algorithms that don't continually adapt or learn every time the algorithm is used. These locked algorithms are modified by the manufacturer at intervals, which includes training of the algorithm using new data, followed by manual verification and validation of the updated algorithm, Gottlieb explained.
"But there’s a great deal of promise beyond locked algorithms that’s ripe for application in the healthcare space, and which requires careful oversight to ensure the benefits of these advanced technologies outweigh the risks to patients," he said in the statement. "These machine learning algorithms that continually evolve often called 'adaptive' or 'continuously learning' algorithms, don’t need manual modification to incorporate learning or updates. Adaptive algorithms can learn from new user data presented to the algorithm through real-world use. For example, an algorithm that detects breast cancer lesions on mammograms could learn to improve the confidence with which it identifies lesions as cancerous or may learn to identify specific subtypes of breast cancer by continually learning from real-world use and feedback."
Gottlieb said the agency is considering how an approach that enables the evaluation and monitoring of a software product from its premarket development to post-market performance could provide reasonable assurance of safety and effectiveness and allow FDA’s regulatory oversight to embrace the iterative nature of these artificial intelligence products while ensuring the agency's standards for safety and effectiveness are maintained.
"This first step in developing our approach outlines information specific to devices that include artificial intelligence algorithms that make real-world modifications that the agency might require for premarket review," he said. "They include the algorithm’s performance, the manufacturer’s plan for modifications and the ability of the manufacturer to manage and control risks of the modifications."
In recent weeks, Gottlieb also has reached out to the public regarding shutdowns of facilities that use ethylene oxide to sterilize medical devices, post-market compliance of breast implants, medical device materials. On Tuesday he also spoke out about the need for more competition in the insulin market to lower prices and expand access.
After Gottlieb walks out the door Friday, Norman Sharpless will become the acting commissioner of FDA. Sharpless is currently director of the National Cancer Institute.