Utilizing AI-Enabled Remote Wearables for Lifesaving Care

Chris Darland, President and CEO, Peerbridge Health 

In this interview with a guest expert, learn about the potential for successes and challenges in implementing AI-enabled remote wearables in health care settings. 

Read the full transcript:

Chris Darland: My name is Chris Darland. I'm the CEO of Peerbridge Health. The first 15 years or so of my career was with GE in businesses all over the world and joined Peerbridge Health about 2 years ago, following a little bit of a passion on heart disease and something that's been close to my family for a long time. I went down the entrepreneurial path and have been CEO here since the end of 2022.

What specific advancements in artificial intelligence (AI) technology have enabled remote wearables to provide lifesaving care? 

Darland: I think there is 2 things. One is a proliferation of new sensors, which can give you more and more useful information that potentially in the past only came from hospital settings. And the way to think through that is the sensors maybe aren't anything novel but the fidelity you can get from these sensors is better. I kind of live in the ECG world and being able to get higher fidelity signals to get more of a full view of the heart are advances that didn't necessarily even exist 4 or 5 years ago. Having better inputs is a big part of it. 

The other big part of it which has really come into its own in the past couple of years is the processing of the signal. You've got this better signal and the reality is there's been lots of fantastic research going back 50 or 60 years on what an ECG signal can tell you, or what blood oxygen saturation can tell you, or what accelerometers in your chest can tell you, but the real world is always way too messy to do anything with it outside of what you might see in a very sterile academic setting. 

Being able to take improved signal, combine that with signal processing tools that AI does a fantastic job of helping to develop. Now you can make some of that academic lab work into reality, which is fantastic. Now you're really pulling hospital-grade work to the home and you're enabling a virtual care appointment. I'm living in the cardiac world, things you'd have to go into a big city for or find a professional cardiologist for, you can do over telehealth if you have the tools that allow the cardiologists to do their job, and that's what's lacking. And it feels like we're at the point where now you can really enable virtual care to the extent like you can with a sinus infection, where you can get a ZPAC prescribed over the phone that I think we've proven can work. Cardiology and some of the more specialty care should be next with the tools we have going out. 

It's kind of like slow and then all at once. And I think a lot of it, and this came up in another conversation I was having, is there's a lot of really, really smart people working on these problems now. And I think, one, health care has been quicker to adopt AI than I would have guessed if you asked me 3 years ago, because it's not a notoriously fast-moving industry. And because there's adoption, I think a lot of people who have built careers in AI, although popular the last 2 years, it's really been around for 30 years, now see it as an opportunity to make an impact and really help people on top of just building cool products or working on whatever that the best app was 3 years ago. You've got this huge influx of talent, of course, and when talent comes in, then you innovate much faster and then it all builds on itself in this huge ecosystem. 

Can you provide examples of successful cases where remote wearables have facilitated timely intervention and improved patient outcomes? 

Darland: We specifically are working on ECG and trying to identify heart failure. And just like I mentioned, we're trying to get more and more high-fidelity diagnostics to a cardiologist so the cardiologist can actually go act faster with a patient. We actually rolling out pilots for this right now. We had our patch on a patient just 10 days ago, the first patient actually for our entire pilot. He wore it for 3 days. He sent it back to us. We were able to know, or at least detect, heart failure with preserved ejection fractions, like diastolic failure in the patient. And he's going in for a pacemaker on Tuesday. 

This is all care that the cardiologist would have found it at some point, I'm sure. The tools were there, and the symptoms had started to show up, kind of lightheadedness and a little bit of dizziness. But the goal is to shortcut that by months, potentially years, which is fantastic. And can you do it without having to have this patient, for example, travel to see a cardiologist? Like, let's get all the work done first. If it was nothing, you're not wasting an appointment for a very expensive cardiology visit, but if it's something, let's fix it now. And the beauty of cardiology, and I'm sure it's the same in oncology and all kinds of different areas, if you catch it early, we know exactly how to treat it. And you can avoid what can be a super stressful caregiver process, not to mention for the patient and potential pain you're going through. But we know how to solve it if we can catch it, it's just a matter of finding it early enough. 

How can health care providers integrate AI-enabled wearables into their practice to better monitor and support their patients remotely? 

Darland: I think it'll happen in a couple steps. There is established reimbursement for a lot of tools today that I think will only get better with AI. You don't necessarily need to reinvent new workflows or integrate consumer wearables into some kind of diagnostic potential or monitoring potential because I think there's a different standard when you're coming up with a ring or a watch to monitor health than what you're actually going to go treat a patient with. I think you leverage patches and wearables that are already pretty common in this space today, but you can kind of 10x the utility with AI tools. And then what I think you'll see next then is you kind of enter into a little bit of the gray area or the unknown. If you can start to monitor, in the case of cardiac hemodynamic blood flow items continuously to avoid ER visits or to predict when a potential ER visit is coming, some of that is still is still uncharted territory with no real reimbursement. Because without having done it, it's hard to establish reimbursement and everybody is very cautious to do anything that doesn't have reimbursement rightfully so especially in rural clinics because money is always very tight. 

I think you have to work within the system today so use existing reimbursable items that are in workflow and kind of elevate them while we continue to kind of test out what the newest tools may be. And the newest tools are going to be this, and really what enables the hospital at home and this whole next generation of not needing the acute facility as much. But that that'll take more time. It will come for sure. And the health systems need it because there's not new hospitals coming up in rural cases. There's 20 closing a year because it's hard to sustain. We need to find a way to get care home, but have to do it in a safe and very thoughtful way so that we're not just shifting risk to the patients. To be thoughtful about it will take a few years, but I am convinced it's inevitable. 

What are the potential challenges or limitations in implementing AI-enabled remote wearables in health care settings, and how can these be addressed? 

Darland: The biggest concern I hear is “I don't want the machine doing my job for me or telling me what the patient has or does not have”. And I think what's going to be really important in this next phase of AI is what I kind of call explainable AI. You have a machine that's coming to a conclusion, but the machine needs to say, “Here's how I came to that conclusion. Here's what I saw in ECG, or here's what I saw on the MRI, or here's what I saw on the CT that drove that conclusion”. And then it's more like a superpower or super tool for the cardiologist to ultimately make the decision. I think a lot of the work using kind of these deep learning models or you're putting lots of data into a model and you're coming out with a correlation, but you don't really know how the machine came to the conclusion. I think there'll be a really hard time driving adoption because then you end up double-checking everything. And if you're double-checking everything, you're still using all the tools that you traditionally use and if you save time, probably not on a diagnostic level. I think explainability is going to be huge in the near term to really get adoption for any kind of remote diagnostics. 

My focus is all rural. My family, my grandfather was a 20-year heart failure patient, which he would argue was like a miracle of modern medicine. I would argue we actually didn't fix the problem, we just dragged him along a 20-year heart failure journey. But maybe from a different generation if we would have caught it earlier, we could have done it. But living in a rural area, being proactive is very tough. Like I said, hospitals aren't growing. They're going away. I think the health, general health, which you probably saw in Kentucky in some of the rural areas isn't as good. You're 6 times as likely to be obese, there's a 20% higher chance of heart failure. 

I think a good example, and these are all what leads to heart disease and these chronic diseases, right? There's a good example of Walmart, who was in perfect position to help early detection in a matter of 2 months, pulled out of primary care, and then added Burger King discounts to their membership. The cards are dramatically stacked against rural communities, for sure. And we're not helping thematically as a society. We're not helping those communities at all. I think we're actually making it worse. Which means that any tools to help have to be exceptionally inexpensive, because nobody is going to go invest thousands of dollars. We know how to find heart failure, but you're not going to go take ultrasounds around in Mt. Washington, Kentucky to try to save a bunch of money and find heart failure earlier. It's got to be really inexpensive. 

What I'm pushing quite hard, and I think it's kind of a challenge for a lot of the startups, is not to go maximize and spend a bunch of money lobbying for the highest reimbursement possible. But how do you invest in yourselves, ourselves, to get the cost down as much as possible? Because I think that's the only way you get reach. And business-wise, if you get reach, you'll make money. This isn't a charity, it'll work, but a cost focus for accessibility versus feature or upselling focus, I think is the only way we really reach communities that are lost today. But I'm optimistic. I mean, part of the beauty of AI is you've got all this even back office functions, invoicing, and how we track trials. And all these things we can do infinitely cheaper than we could even 3 years ago. The challenges for us is like, all right, now that we've saved all this money, can we put it towards expanding access or do we just put it towards making the bank account a little bit bigger? And I think there is a way to do both. And not just us, lots of really fantastic startups at the same time I think are pursuing the same thing. It's a very exciting time. I think the opportunity, finally the opportunity is there to make an impact. There's no shortage of hurdles in front of us, Walmart is it is a decent example. But somebody has got to solve it and it feels like we have a shot.