Latest News

FDA Proposes Rules for Devices Using Artificial Intelligence

The Food and Drug Administration has unveiled a proposed regulatory framework for medical devices that use artificial intelligence algorithms.

artificial intelligence

Source: Thinkstock

By Fred Donovan

- The Food and Drug Administration (FDA) has released a proposed regulatory framework for medical devices that use artificial intelligence (AI) algorithms.

The agency is seeking industry feedback on the proposal, with the goal of issuing a draft guidance based on the feedback.

“Our approach will focus on the continually-evolving nature of these promising technologies. 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,” explained FDA Commissioner Scott Gottlieb, MD.

The FDA is considering an approach that would enable the evaluation and monitoring of an AI-based device from its premarket development to post-market performance in order to provide reasonable assurance of safety and effectiveness.

The approach would allow the FDA’s regulatory oversight to embrace the iterative nature of these artificial intelligence products while ensuring that its standards for safety and effectiveness are maintained.

READ MORE: FDA Grants Breakthrough Device Status to Healthcare AI Software

Last year, the FDA approved the first AI-based medical devices, one for detecting retinopathy and the other for alerting providers of a potential stroke.

The device for detecting retinopathy is an autonomous diagnostic AI system that can be used at the point of care and no human reviewer or oversight is necessary. This shifts specialty diagnostics from the academic setting to the primary care setting, increasing the number of patients who can be tested and reducing testing costs, said Michael Abramoff, founder and CEO of IDx Technologies, which developed the AI-based device.

Abramoff said that he had to raise $22 million to develop his system and get FDA approval, which he just received this year, almost two decades after he first came up with the idea.

The device for alerting providers of a potential stroke was developed by Viz.AI. The device is designed to analyze computer tomography (CT) images of the brain and send a text notification to a neurovascular specialist about a suspected large vessel blockage.

The algorithm automatically notifies the specialist at the same time the provider is conducting a standard review of the images, thereby involving the specialist sooner than the usual standard of care in which patients wait for a radiologist to review CT images and notify a neurovascular specialist. The notification can be sent to a mobile device, such as a smart phone or tablet, but the specialist still needs to review the images on a clinical workstation.

AI Will Have a Profound Impact on Healthcare, Says Gottlieb

In announcing the proposed regulatory framework, Gottlieb said that AI “has helped transform industries like finance and manufacturing, and I’m confident that these technologies will have a profound and positive impact on healthcare.”

“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,” he said.

Gottlieb explained that the AI technologies that his agency has cleared use “locked” algorithms that do not adopt or learn each time the algorithm is used. The locked algorithms are modified by manufacturers on a periodic basis, which involves training the algorithm using new data and manual verification and validation of the updated algorithm.

But “adaptive” algorithms do not need manual modification to learn and adapt. These algorithms learn from new user data through real-word use. Gottlieb cited the example of an algorithm used for detecting breast cancer lesions on mammograms, which could learn to identify specific subtypes of breach cancer by learning from use and feedback.

“We are exploring a framework that would allow for modifications to algorithms to be made from real-world learning and adaptation, while still ensuring safety and effectiveness of the software as a medical device is maintained,” said Gottlieb.

“A new approach to these technologies would address the need for the algorithms to learn and adapt when used in the real world. It would be a more tailored fit than our existing regulatory paradigm for software as a medical device,” he added.

Gottlieb said that the FDA will take a total product lifecycle approach for regulating devices that use adative algorithms.

“We’re 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 the FDA’s regulatory oversight to embrace the iterative nature of these artificial intelligence products while ensuring that our standards for safety and effectiveness are maintained,” he concluded.