AI Is a Tool, Not a Replacement for MedTech Expertise with Mike Stuedemann, Co-Founder and Principal Consultant at agilityIRL

[00:00:03] Hi everyone, welcome back to Outcomes Rocket. We are live here in Boston at Device Talks and I'm fortunate to be here with Mike Stuedemann. Thank you so much for being with us on today's episode. Well thanks for having me Alexis. Thrilled to be joining you and talking with the Outcomes Rocket team here at Device Talks Boston. Amazing. Speaking of the events and being here in Boston, how are you finding things?

[00:00:26] Yeah, it's great. We're connecting with a lot of medical device companies. AgilityIRL, we try to help device companies bring product to market faster in an FDA compliant way using agile techniques. And so we're seeing a lot of interest in space. A lot of interest on a topic that I'm sure is near and dear to your heart is on Outcomes. The focus on patients, not to get too philosophical, the word focus in general.

[00:00:49] So many technologies, robotics, AI, focus is one of my favorite F-words. My other one is feedback, but some of your listeners might have thought a different F-word, but it's not that type of podcast. So yeah, what about you? What are you folks, Anne? Yeah, I think for me, one of the things that's been on my mind recently is like 10 years ago, I heard a lot about how healthcare wasn't very tech forward, wasn't very fast-paced moving.

[00:01:16] I heard a lot of people saying, why can't healthcare be like Amazon? Why can't it be like this? But actually, I would argue that post-COVID, but especially now in the last couple of years, things have really flipped. Like just walking around here on the floors and seeing some of the robotics and seeing some of the devices that are possible. And I want to say that healthcare is maybe like much more cutting edge now than what I would have seen. Absolutely. And one of the things that fascinates me about the space is how broad it is. I mean, everything from an aura ring. Yeah, I'm wearing one right now.

[00:01:46] You're all right. Everything from an aura ring, subtle plug there, aura, if you're looking for a marketing deal. All the way to devices that require follow-in cardiac surgery using a robot to perform. And that's part of the fun of the space. It's also though, I think, to your point, personalized healthcare is the future. I agree. It's not a therapy for a middle-aged man with this condition. It's a therapy for Mike and his specific genetic profile in this condition. That's true.

[00:02:13] And so in terms of the submissions and the process and that FDA experience, how are you seeing that play out in what's coming onto your desk and your team's desks? The FDA is still risk-averse. And that's not a political comment. It's just the way it is as regulators. I think the FDA is struggling right now with the personalization of medicine, particularly with they've given some guidance on AI, but it's not really actionable. Yeah, okay.

[00:02:37] When I think the big question that we all have to ask in regards to AI and FDA regulation is, how do you regulate something that's constantly changing on a minute-by-minute basis? Yeah. So if you have an AI-enabled device and it's learning about you based upon the number of steps that you take a day and what you're eating and your particular therapy prescribed, how do you prove to the regulators that you didn't break something? So true. We're seeing a lot of interest around that.

[00:03:05] We're seeing a lot of focus on how do we speed up product development. You can't afford to have a product take multiple years because the market's changing so quick. You're basically eroding your own market by waiting. Something's already extinct by the time it's happening. Absolutely. Yeah. I was just talking to someone yesterday about how she needed to pull out her laptop from university for something. And just even holding her laptop then, remembering, thinking, oh, I'm going to have a laptop that's portable.

[00:03:34] Until she took it out of her bag, it felt like a 10-pound bag. Right, right, right, right, right. So it's like, wow, I guess we're not even realizing how. We know that tech is changing so quickly. We know that. But holding a piece of tech that's not even from that many years ago and just really seeing and feeling the differences. It's a really fun time to be in the space because of that accelerated innovation. A lot of people are like, yeah, but it's also changing. Yeah, but the change is also exciting. For sure. The change is also exciting.

[00:03:59] Are you seeing a lot of people out of curiosity who are trying to use AI to help with their submissions and things like that?

[00:04:34] We are. Just, hey, here are three requirements that you need to maybe go back and revisit. Okay. Which that used to be a manual process. It was also error prone. It was dependent upon you as the quality engineer going, oh, yeah, I need to go check this and then kind of pull through the document stack all the changes. Well, now you have an AI agent that's a little bit more intelligent, can go out and say, yeah, hey, you should look at these based upon the risk. Love that. But I want to be clear. I don't think AI is going to replace anyone.

[00:05:03] I think it's a tool. Yeah, I agree. I think I had heard a lot of people in the beginning waves of AI. And I say beginning. That's like two years ago. Not even the beginning of AI, but the beginning of like my use. But I used to hear some fear. I also agree with you now, having been someone who we're all interacting with AI on a daily basis, I feel less afraid that it's going to take my role. But I do think that if I don't use it to increase my efficiency and kind of some of the things you're pointing out, then that will become the piece that becomes a non-negotiable.

[00:05:33] Yeah, I actually had an opportunity here a couple weeks ago at a health information management symposium to hear the chief AI officer of a local health care system in Minnesota where I based. And she was asked the question, of course, after a presentation, is this going to cost people jobs? And she laughed and she goes, maybe in my grandchildren's generation. Yeah. But she goes, not certainly in my working career, not in my children's working career, but maybe that.

[00:06:03] Wow. She goes, the technology is good, but the stakes are too high in medical and med device. So right. And she's seeing a lot of use in like ambient listening. So you're my doctor. We're having a conversation. The notes. The notes. But also agents suggesting, hey, Dr. Alexis, have you thought about this? Maybe he's describing these symptoms. How about that? Okay. So it's almost like opening up other possibilities as opposed to just relying on that. Right. And clinicians experience. Exactly. Exactly. Yeah.

[00:06:31] And she also talked a lot about boring AI, which I thought was funny. And she told the story of heart transplant doctors. And this organization had built a tool to recommend who's eligible for a heart transplant, which is, I've never done that, but I imagine it's fairly complicated. Well, the doctors weren't using the tool. Oh. And so they went to these physicians and they said, look, this is so complicated. We can't rely on a tool.

[00:06:57] But what you could do for us is have AI take all these documents that we're looking at and synthesize them and put it up on a board while we're having our discussion. Yeah. Okay. Yeah. And that's, Alexis, that's never something that's going to be on the front page of Info Week. It's nothing that you and I are going to be like, oh, that's really cool. Yeah. But it's helping them make a better decision as human beings. Yeah. You know what's funny to me is when we talk about like medical product design, there's

[00:07:25] that whole wave of patient centricity and innovating based on what the genuine needs are of a population. Sure. But it's funny to me because that had been in the healthcare rhetoric for years. And then when it comes to AI, everyone's like, that's out the window. I just want to do something that sounds sexy and cool. Yeah. And that's another theme that I've picked up on here through our conversations. And I'm curious if you've heard this is let's not forget about patient centricity. Let's not forget about at the end of the day, we have to make this accessible. For pain points. Exactly. Yeah. Exactly.

[00:07:55] Great. I love that. Anything else you're seeing just in terms of like yourself or your team, or it could be this event, but any bets you're willing to make on where the future is going to go or hurdles that we have to overcome as an industry? Well, there's always the regulatory hurdle. And I want to be clear from my FDA friends who are listening. Yeah. You have a, I'm not saying let's get rid of regulation, but we have to get over that hurdle. I think we need to figure out a regulatory framework that works for AI. We talked a little bit about that.

[00:08:22] And I think continuing to focus on the patient and then the ever present issue of cost in the U.S. To go back to outcome rocket, we're not getting the outcomes as a society that we want from our healthcare and the cost is going up. So we have to find a way to address that. If I had to bet, I think we're going to see AI enabled devices in some form. I think there's going to be some limit on the learning that can occur. But I think we'll see that. I think we're going to see more individualized medicine.

[00:08:51] I think we'll see therapies that are going to be developed based by some of the bigger players based upon AI analysis of data. I mean, the big players are sitting on parabytes, gigabytes of data. Yeah. That has been something that has been on my mind for years on like, what a waste that we can't leverage this data effectively to either identify opportunities or just going to have learnings. Like you hear back that healthcare data sets aren't representative and are inclusive.

[00:09:21] And I'm like, but there's all this data. Everyone's collecting my data all the time. And you raise a legitimate point that we do have to look at is the data biased in some way, gender wise, ethnicity wise. But there's to use a really old phrase, there's golden then their hills in terms of new therapies, more targeted therapies. And we finally have the computing power to be able to go in and dig that out.

[00:09:47] So that would be the other bet is I think we'll see AI enabled devices in some form. I don't say I don't see true learning devices like real time, but maybe some learning with some constraints or some learning with a little bit of a leash on it. Yeah, no, I love that. My personal hope is that we're going to see AI become like personal assistants. Oh, absolutely. Every time we needed to make an appointment, it would just already look at my calendar, already know the appointment I need to make, like preventative checkups, things like that. Just make it happen.

[00:10:16] I can just show up and just tell me how to get there and give me the directions as well. So I don't have to look that up. But who knows? We're a few years out from that. Oh, it's interesting. One gentleman that we talked to here yesterday, a little bit different application, but he is the drone operator for a local health care system in New York. Oh, wow. And they're using drones, autonomous drones, to fly themselves to deliver medications between their outpatient pharmacy and the hospital. Wow.

[00:10:43] And between the hospital and one of the sister hospitals, that's a 45 minute flight away. Wow. And I said, so what's the business case? And he goes, we've all been in the hospital. We know that the longest period of time is from when the doctor says you can leave to when you can actually leave. And he goes, part of that delay is because the outpatient pharmacy has to fill all your going home with drugs. Yeah. He goes, it used to take us two hours to send the order over to outpatient pharmacy, have the pharmacist fill it, send it back. He goes, with the drone, it takes about 10 minutes.

[00:11:13] Wow. And most of that time is the pharmacist divvying up the drugs. Yeah, yeah. The flight. Not even the flight. We say these things are a long ways away. Yeah, maybe not. That story kind of makes me feel like maybe we're not that far away. Yeah, yeah. The future is pretty much here. Yeah. I love that. Yeah. Well, I always like to ask folks on these episodes, going into the second half of the year, look ahead for next year. Is there anything that you want listeners at home to kind of stay tuned for, keep an eye on you and your team for, or anything you want to tease?

[00:11:41] Yeah, I think continue to become AI native. Love that. Go beyond the chat GPTs, go beyond the clods. Whether you think that the development of AI is a good thing or a bad thing for society, it's going to be here. So learn how to build a basic agent. You don't have to be an engineer or a technologist to have to do that. Yeah. I think from a med device perspective, recognize that pressure to deal with change and iterate

[00:12:10] quickly while keeping the patient in mind is going to become, is already ever present. The pressure is only going to ramp up. I love that. And is there anywhere that you want to let the audience know in terms of to stay in touch with you? Oh, absolutely. To have your team, your company? Yeah. Check us out at agilityirl.com. We also have what's called the Take Five podcast. Oh, I love that. We'd love to have you on. We call it Take Five because it's just a quick five-minute conversation about a topic. Love that. So check out Take Five IRL as well. Love that. That's great. Well, thank you all so much. And thank you, Mike, for being here with us. Thanks for having me.

[00:12:40] And yeah, stay tuned to click our show notes to learn more. We'll link all the different pages and sites that Mike talked about. And yeah, maybe we'll see you on Take Five. Sounds great. Thanks, Alexis.