AdvaMed’s Zach Rothstein sat down with Bakul Patel, Director of the Digital Health Center of Excellence at the US Food and Drug Administration, for updates on the Center, its AI/ML Action Plan, Pre-Cert and the concept of “care at the person”.
ZR: Bakul, thanks so much for taking time to join us today to talk about what’s going on at FDA and its Digital Health Center of Excellence. Can you give us an update on how the Digital Health Center of Excellence is going, where it is organizationally and what its expectations are over the next 6-12 months?
BP: Thank you, Zach, for having me. As you can imagine, since the launch of the Center of Excellence late last year, we’ve been working hard to try to establish the center’s operations. First, we’re trying to get coordinated within CDRH, and then set up what that coordination should look like across FDA’s centers. This isn’t to say that we would start directing other centers, but rather helping them more quickly get to the same place, because of the Center of Excellence. I think we would provide them with the expertise that we’ve built up over time at CDRH. So that’s the goal.
One of the things we did at the beginning of the year is the AI/ML Action Plan, so down the road you’ll start seeing us tackle what the common areas of interest that require regulatory science research or scientific work in that area will look like. We’re looking into what the most common interests for all centers that will help digital health rise to the next level will be, and which opportunities we can explore. Another thing that we want to start thinking about is where do stakeholders feel that things need to be done pre-market. So those are the main things we’re working on, while we also work on organizational items and building up our team.
ZR: That sounds great. So as you mature the center, and it stands itself up a bit more, is that something industry will be able to interact with and engage with during the lifecycle of a medical device?
BP: Yes, that’s exactly what we’re hoping to do. Right now we provide a very limited approach toward engaging with the device center – we have a digital health inbox. We can expand that to a help desk on digital health policy and eventually expand that to other centers and say, as you start seeing some of these aspects, if you have questions, we’ll answer them, and of course we’ll connect you with the right experts and get you answers as fast as possible. It’s less about solving for other centers’ regulatory challenges, and more about getting them the tools and expertise to solve the problems they may have, because they’re all different. So that’s the next step in terms of what we’re thinking about.
ZR: You mentioned earlier the AI Action Plan, can you talk a bit about where that stands within the agency and what you see as the most high-level, important items that need to be worked on in this space?
BP: Yeah, I can tell you obviously FDA has to play a role. We’ll come out with guidance on the SPS/ACP part, that’s something we’re actively working on. The other two areas where I feel more conversation needs to happen are around transparency and GMLP. If you talk about good machine learning practices, what does that look like? It seems like plenty of people are working on that, but plenty of people also aren’t working on that, so it’s really one of those things that we feel from an FDA perspective that we can provide, from a regulatory approach, some principles that would be beneficial for us to think about for good machine learning practices. In my mind, I think about it as a manifesto – what should that manifesto look like for good machine learning practices? The second part is about transparency, which is probably the most interesting and unsolved part because there’s transparency at different levels. One is, what does a user know, what doesn’t a user know, or need to know at the time of using the AI/ML products. And then, what does a reviewer, who is reviewing independently from the maker of the product, need to know so they can review the product easily and get that confidence of safety and effectiveness? So those are the different aspects of what we call transparency. We are looking at announcing a public workshop this year, we’re hoping to do it before or around the fall timeframe. So as we start moving in that space, we will probably start looking at what we should be doing, and bringing the community together to resolve some of these problems.
ZR: Really interesting. There’s another area – speaking of joint participation – and that’s the Pre-Cert pilot program, and that had been front page news for a number of years. I know that last year was a learning year and a building year, based on the pilot. Can you give an update on where things stand with the pilot, and what we can expect down the road?
BP: Yes, absolutely. The Pre-Cert pilot is still very much one of the fundamental things we have to work on in addition to all the stuff we talked about already. We do need to support the centers, to make sure we’re coordinated and consistent, and we still need to make progress on AI/ML, but to connect the dots, the AI/ML discussion paper we put out in 2018 is really a use case of the Pre-Cert pilot program. It basically brings together the entire lifecycle. So last year, we worked on the building, and trying to take and ingest the content we’ve collected so far and make sense of it. One of the things that became apparent during the pandemic was that we really can’t go out to organizations and start doing interviews and start building a program on a product-by-product basis. So we’ve been trying to design a way that we can take a chapter out of industry’s book, where there’s lots of modeling and simulation that happens before a product goes to market. So we’re working on a research collaboration with MITRE to start thinking about how we can de-risk the program’s issues or topics or areas that we could do in silico, or we could start modeling and create a platform for simulating some of the scenarios that we would anticipate if the program was live. We’re trying to convert the program into more of a digital domain so we can test the program faster. I feel like that approach will allow us to better learn about the program, it will probably help us get more confidence in some of the concepts we’ve laid out in the working model, and help us identify what else to try when something doesn’t work. So that’s the plan for pre-cert going forward. We’re just embarking on that journey, literally speaking, a few days back.
ZR: That’s exciting. Bakul, are there any other topics that you’d like to talk about while we have time?
BP: The biggest thing to underscore is that the pandemic has taught us how telecare is here to stay. And as you start thinking about the concept of remote, in the care setting, digital health is right smack in the middle of this. We at FDA anticipated a while back that this is where the world was heading, so the pandemic and public health emergency completely changed the direction of that, and accelerated that. And as a result, some of the concepts we laid out in pre-cert are becoming real, and becoming much more urgent for us to establish something like this. Understanding how we can better execute our processes for ensuring safety and effectiveness is becoming much more real. I also feel like as a result of this remoteness, we will probably see an increase in demand for de-centralized clinical trials or getting remote participation in clinical studies. I feel like all that “care at the person” is going to change how we do business, how people in our country will be cared for by our health care systems. We are on the verge of a big shift, and probably a better shift, in how health care will be offered.
ZR: That’s a great reminder of how much the COVID pandemic has really impacted and altered the way the world views health care delivery, and I think you’re right, Bakul – our members would agree the importance of digital health is now front and center. Thanks, Bakul, we really appreciate you spending a little time previewing some of these topics with us. For those of you reading the transcript here, we hope you’ll join us for The Digital MedTech Conference on May 6th. Thanks so much.