Traditional medical devices have a largely episodic regulatory process, with a major FDA approval acting as the gatekeeper into the market.
But how does this paradigm need to change in a healthcare ecosystem with the within category of software as a medical device (SaMD) that is obstensibly continually iterating and improving?
With the Rise of AI, What IP Disputes in Healthcare Are Likely to Emerge?
Munck Wilson Mandala Partner Greg Howison shared his perspective on some of the legal ramifications around AI, IP, connected devices and the data they generate, in response to emailed questions.
That transition was the topic of a panel at the Digital Health & Medtech Showcase in San Francisco featuring a host of companies hoping to make their way in a changing regulatory environment.
The discussion held particular relevance after FDA commissioner Scott Gottlieb recently unveiled Version 1.0 of the agency’s working model for its Pre-Cert program, intended to streamline the regulatory process for digital health companies with software-based tools.
The working model relies on an initial judgement of the developers of the program through a so-called Excellence Appraisal which would put companies on a pathway to more personalized regulatory review and approval.
Two of the speakers on the panel were representatives of companies initially selected for the pre-cert program: Pear Therapeutics and Roche.
A Personalized Approach to Medication Nonadherence
At the Abarca Forward conference earlier this year, George Van Antwerp, managing director at Deloitte, discussed how social determinants of health and a personalized member experience can improve medication adherence and health outcomes.
“The pre-cert pilot is a very interesting experiment for FDA because in general the agency has done a pilot they have the idea of what the pilot is going to be and they release that and they test that,” said Lesley Maloney, the head of US regualtory policy for Roche Diagnostics.
“This is the reverse. FDA is saying ‘we don’t have all the answers the public should help us figure out what the pilot should actually be.'”
Dave Amor, the vice president of quality and regulatory affairs at Pear Therapeutics said the agency’s pre-cert pilot approach has been striking because of the collaborative nature of the program, which actually required a bit of mindset change from traditional regulatory and quality management executives.
“It took some us regulatory and quality nerds some time to remove ourselves from this very entrenched statutory framework of what we were used to doing and really understand the FDA was really trying to learn from us,” Amor said.
Still, as most of the panel applauded the overall aims of the FDA’s vision when it comes to SaMD regulation, there were also plenty in the way of challenges which could result in a rocky transition period as details get hammered out.
One potential stumbling block is the disconnect created between the larger strategic intentions of FDA leaders and the agency’s employees doing the on-the-ground work of reviewing product submissions.
Sam Surette, the regulatory affairs and quality assurance manager at AI-based cardiovascular imaging company Bay Labs and a former FDA reviewer, said the product review process still largely operates bottom up with a single reviewer acting as the company’s main point of contact.
“It can create some tension as things are changing because eventually as pre-cert rolls out of the working model into normal practice, the rubber is going to have to hit the road at some point and the review staff is going to have to implement these policies,” Surette said.
Among the potential solutions suggested by the panel to ease this implementation would be increased education and training for review staff as well as more proactive participation by reviewers in the initial Excellence Appraisal process.
These challenges may especially be compounded with the increasing use of deep learning and advanced AI analytics tools in digital health. These technologies rely on training data which may have questionable quality or provenance. As the adage goes, garbage in means garbage out.
“Hardware tends not to adapt, it is what it is. Software adapts you add the Wild West of AI and it could be adapting everyday,” said WellDoc Chief Strategy Officer Anand Iyer. “How do I actually ensure that the training data is accurate such that it is actually going to deliver against the intentions of my product?”
Pear Therapeutic’s Amor said that in his opinion, the pre-cert program provides a good framework for the review of AI-based products because of the nature of trusting the company to pursue best practices and bake in quality assurances and security at the development level.
“If you’re establishing a complex neural network you are going to change based on the data sets that you have and that’s only going to increase as the company progresses in iterating on that model,” Amor said. “In that system you have to have a trust in the organization’s process.”
Picture: Getty Images, z_wei
