Peter Arduini - AI's Impact on Healthcare Transformation

AI and data are no longer just tools. There should be your sovereign assets in this new world. For many though, AI still feels more storytelling than a strategy. However, the winners here aren't louder, they're just more deliberate. They're reshaping industries, governments and society. AI and data horizons is your fast track guide to that deliberate future, presented by EDB, the leading sovereign AI and data company. Each episode brings an expert practitioner navigating and shaping this new world. I'm your host, Michael Gail, Wall Street Journal, best selling author of the Digital Helix, and the host and creator of your award-winning podcast from Forbes, Futures and Fingers. This is AI and data horizons from EDB. In this podcast, we talk with experts about their insights and ideas about the sovereign futures of the AI and data worlds they live and work with. This is a very personal podcast for me because you may not know who Peter Arduine is, but he's present and CEO of G Healthcare, one of the largest, maybe more, most pivotal healthcare technology providers on the planet. So I sat down with Peter a while ago and really at a pivotal moment and the company had just been spun out as an independent leader medical technology from GE. This is a really critical moment for how they attacked the future world of five, ten, fifteen years ahead of them. So listing back today is remarkable how relevant and prescient his vision remains. And again, hearing this again, it's striking how much this vision has actually already taken shape in the world around us. He framed the mission around one simple but bold purpose. Quote, "We put a lot of energy around this focus of creating a world where healthcare has no limits." I think we think a lot about that when we think about being around AI and data platforms, the sense of almost infinite possibility. He spoke candidly about AI and how AI and data were ready reshaping care delivery from reviving 10-year-old MRI machines with state-of-the-art imaging quality to dramatically reduce and scan times. Quote, "With the right AI algorithm you can actually eliminate noise, increase the signal and even cut through put time by 50%. You also push just to imagine what happens when advanced tools like handheld, ultrasound and AI powered workflows become as common as a stethoscope, which is a great metaphor. Quote, "If every primary healthcare physician had an ultrasound, how much could you find earlier? How much could you bend the curve of better outcomes and lower costs?" Really remarkable statement. He underscored that the real power of AI isn't just in diagnosis, it's redesigning how care is delivered. As I think we sit here in our industries and realise this immense potential for sovereignty over AI and data, the sense of unbound possibility should really be driving us all. He made also a great quote, "People think AI means a system telling you what disease you have, but it's equally about smarter devices, more precise therapeutics and even helping hospitals run like air traffic control. Two years on, many of these ideas have moved from vision to reality, earlier screening, multi-mode, modal diagnostic and AI-driven workflows in medical and healthcare system, and now driving the very transformation he described. So, let's roll the tape. We're lucky today to have a guest who I think is leading a revolution or the next revolution in how healthcare, intelligent systems healthcare using AI, using data, can radically shift the way we as individuals on the planet, receive not just preventative or diagnostic support, but actually during a whole life cycle, can use data, modeling and systems to fundamentally change how we manage our lifestyle. Look, COVID had certainly changed the way we maybe interact with the medical industry with remote monitoring. We've all got some form of digital watch, probably on our partners in life that track heartbeat. There are monitors that contract your possibility of a heart event occurring. We are literally surrounded by data in our normal living lives inside and outside the medical environment. My guest today is a CEO of G Healthcare, which actually serves about one in seven, one in eight people on the planet every single year, and their understanding of how all these systems work together, they have actually their own Edison AI platform is something we're going to talk about today. So Peter, I've set you up, and that's maybe too much to digest, but thank you for joining me today. Thank you, Michael, for having me on. I very much appreciate it. Look for the discussion. Good. Well, I don't know where to start. I think in some ways I'd like to talk about AI. It's the hottest subject, probably outside Taylor Swift and Travis Keltchee, but fundamentally the promise and the possibilities are very tangible. So can you talk to us given the infrastructure, how the technologies, the patents, the investments you make, how you're thinking about AI in the context of GE Healthcare? Yeah, no, look, it's a great question, Michael. I would start with this broader point about as a new company coming out of GE. We put a lot of energy around what our purpose is. This focus of creating a world where healthcare has no limits. It was bold statement, but when you think about our system, there's quite a few limits. How do you choose some of those to go after? Either it be access of patients or just the quality of the broader delivery of care, or the other big one, which is obviously the cost of care, which is now above 18% in the United States. And so when I think about AI and data, it actually has the potential to positively affect all three of those, better care because of earlier care, better input and capabilities because of the quant capabilities and then finding and addressing things sooner by definition reduces the total cost. Yeah, and it healthcare is, I think the ultimate tip of the ISB for where AI will make a difference because as you said, it basically embraces all stages of that healthcare paradigm. Where are you seeing its most immediate net returns? Obviously, that breaking confidentiality for you as a company, the institutions you work with and as patients. Yeah, no, I would say right now what it is helping the most and actual products that are out there. And we have a good amount of devices that have AI integrated in them. It's fundamentally changing some of the physics or capabilities that we thought were kind of the way things work. And just to give you an example, how an MRI processes images, your body is full of water and that positive charge of those hydrogen molecules aligns to the magnet. And much of that signal is just noise. It's not used to make an image, but with the right AI algorithm, you can actually eliminate some of that noise that would never be used to make the image increase the signal noise, which means a better image. And then also because you have less data crunch, make it faster. So in that case, there's an example where maybe a 10 year old system, you could make it have state of the art image quality today, which gives better cost and capabilities. And then secondly, be able to have 50% improved throughput. So that's a simple example of how you make something that is very tangible to date better because of the way you can manage algorithms. I would say the parts that we're starting to see evolving or clearly in how diagnosis are made and capabilities that can help enhance the outcome from a standpoint of advising standards, advising to look at different areas. And again, this is just the tip of the iceberg as you referenced. So Mark will be quick. It's the ability to up modernized hardware through software is I think clearly not just environmentally, but economically where we're going. Do you see that construct of the algorithmic update that almost over the air type process occurring across a range of equipment? We're doing particularly with MRI. So naturally suited to this is tough to replicate that sort of success. No, no, we think it's going to happen across more broadly, even much more broadly than the equipment that we touch. So if you actually think about something like devices and deploying products for structured heart to different imaging modalities to IV, individual diagnostics, everyone at some level is experimenting in the area and finding positive ways to drive change. One that I'll give you an example which I think is going to be a really important evolving modality for cost and access and quality is ultrasound. It doesn't use radiation. it can be used in many, many different situations. And the combination of microization of the technology, cloud computing to enable REACH and AI, these three together, really has the opportunity to transform care delivery. So as an example, with handheld ultrasound, what we have as well as others has some of the power of what the best systems did just eight years ago. And then when you couple that with AI, in our case, we have a capability that can actually help a novice user be good at doing a basic screening. And so in that case, having a philbot in this group, someone who's going to someone's house to the clueless blood potentially trained up to be able to do an ultrasound opens up a whole host of access capabilities. In the future, your primary care physician having a handheld ultrasound, if you think about that, if every primary care physician had an ultrasound that they used versus a stethoscope, how much more you could find earlier, how much you could actually kind of bend the curve of better outcomes, but I would submit to you as also better cost. So there's a lot of interesting ways that we think this is going to head. - When you speak to medical institutions and the medical system, because this is such an obvious, 30 word description, everybody was going to be jumping at you for this, because this combination of cost, you know, really capacity and prevention, it is a sort of, gaudy not of the medical industry. Have you seen institutions and it's sort of insurance bodies jumping at this, or is it still embryonic in it sort of from lab to basic test environment? - Yeah, I think it's a great question, Michael, and the answer is it obviously varies quite a bit around the world, but I would say in the United States, it's still early innings from a standpoint of larger institutions and hospital systems, trying to figure out how to pick their bets. Do you want to partner with many different companies? Do you want to try to kind of solve specific use cases? Are you trying to do a broader reach and also realistically in a scenario where, you know, many health systems have come out of some challenging environments plus COVID and labor costs. And so, you know, our view is, and this is, you know, this idea of helping change insight out, as opposed to outside in, and what I mean by that, as you create smarter, you know, devices and drugs, and you integrate them into the already pathways, whether it be neuroscience, or oncology, or whatnot, and help enhance it. It's a way that you can obviously drive better outcomes around a specific use case, then trying to actually create, you know, an infrastructure that says, it solves for this problem. And so, we're quite focused on that. I mean, the ultrasound, just to come back to that, it's a great example of specific scenarios where, you know, you might be trying to reach a population that's under diagnosed heart valve challenges, and this is a way of actually doing that with a technology to help reach that. Another scenario might be, you're trying to manage what your total costs are, and you're trying to take a look at how you can actually better leverage your labor, and having a command center type product that helps you manage where your employees are, and the patient flow real time can help do that. And so, we see it more around, I have a problem, and I'm trying to solve it. How can you help me or others help us solve this problem? And with it comes AI workflow, and in many cases cloud computing. - What sort of skills are gonna be needed in this new world to handle that technology? You talk about flabobindus being able to go to a house and do a hand-based ultrasound, but how much are we gonna have to re-skill certain levels of the workforce to be able to handle this? - Yeah, look, I think there are we clearly areas, you know, throughout healthcare, where being obviously comfortable with digital dashboards and analytics is just gonna be increased. I think ultimately for clinicians, it's gonna give more time to do what they're most valuable about, which is being face-to-face with that patient, being able to actually help drive the diagnosis. But clearly the idea of actually interacting with devices of all types and being able to use data or statistics to make more effective outcome decisions is just gonna be core. I would argue that statistics relative to everyone and actually how do you think about the best outcomes, the best scenarios is going to be important here. I mean, the machine learning models by definition are trying to find the best scenarios. But on our side of things, we're trying to make it as seamless as possible. And, you know, I would give you an example that would say in the future, you know, the next 10 plus years, the idea of having multi-modal data on one pane of glass, right? A clinician can look at one screen and see everything about, if it's you Michael, what's going on with your history where you are today? And the understand, did we do everything that the standard say you should do today and actually the system highlight, did you or did you not? But even in a valve phase, be able to actually have the algorithm say, what we saw on this CT scan in the past or this ultrasound or this cardiac scan or this pathology, be able to, you know, assimilate data from multi-modal areas and not necessarily give you or give the clinician the answer, but say, here's what maybe the next one or two steps should be. And if you think about that with a quality algorithm, whether you're at one of the best institutions in the United States or you happen to be in, you know, Africa as an example, the ability for actually to have the next course of action be high quality can really change how we think about the delivery care. - And, to a completely agree. So, Q from you've taken one of the most famous healthcare units in the world out separately since January. When you look at the philosophy behind why this was done, what was the basic tent for freeing up this asset for growth and how has that transformation sort of started and evolved? And I know nine months is not a long time to talk about that journey, but it'd be interesting to know because this isn't a startup in itself, but the energy seems very startup-like for you. - Yeah, well look, I think our chairman had the fortitude and the courage really to say, what is GE can be three very special focus company, you know, a great aviation company, a great energy company, a great healthcare company, and we're the first out. And I think the point here is is where the future is going and care is very much around precision and personalization is definitely one that is actually evolving from us just being a company that took images primarily to a company that's heavily involved in the diagnosis process. And I think as a separate company, you know, the ability to have a dedicated board of deep clinical folks, the ability for us to actually put our balance sheet, all of our focus around this and bring investors in that are solely focused on healthcare. I think it helps with thinking about time horizon investments, it helps thinking about how we would think about acquisitions and partnerships, but in general focus matters. And in particularly with this window of time where there's gonna be a lot of rapid change, having a very deep organizations, you know, just focus solely on healthcare helps you be able to move quicker and faster. And, you know, look, it's early as you mentioned, but what we've been really trying to do is say, we've been given a great gift, the assets of a hundred year old company, but a culture that we're trying to drive is very much an entrepreneurial culture that is people inspired to make change. We talk about this internally that, you know, the next 10 years we may see some of the most profound changes in healthcare more so than the whole last 100 put together. And a lot of that is because of what's happening again with microization and robotics, what's happening with cloud computing. And in particular, the evolution of AI and it's effect on all of this. - I really put, what's the one thing that surprised you when you have these conversations in a positive way with clinicians, institutions that they go, oh, I haven't thought about it this way. Is there a unique clarity both in the version and the capability you guys have that seem to hit home? Where is that sort of unique? - Yeah, I think, Michael, the interesting thing is, when people think about, and I'll just use artificial intelligence, they're tensed that everyone goes to, oh, this is gonna say, you have, you know, this disease or you have, this element, and it's going to give me the answer. And as we spend time on it, when they think about actually how it could change the delivery model, we there makes a device smarter or more effective, or actually allows more precision delivery of a therapeutic, or it changes the workflow, right, where it's kind of an air traffic control for the institution and allows you to make real-time decisions on better deployment of people. I think that's when folks eyes open up a lot broader that yes, there are components that could enhance diagnosis, but there's just so many things that come together in this very complicated system called delivery of care that can be simplified and enhanced. And as we all saw coming out of COVID, it was a very challenging time for clinicians. The cognitive burden on taking care of patients, some laying all this information, I think many of the more enlightened organizations when they think about it say, wow, I could create a really great environment for my clinicians, take a lot of this work off that they don't love to do anyhow, actually make it more effective, and use my most important assets, my people and my clinicians to actually bring better care. And so there's still a journey on that path for sure, but I think that's when the light comes on. - Yeah, it is. I think if you look at what Hampton's practitioner is over the last seven years, they've been elevated in their range of diagnostic and prescriptive recommendation, as that sort of bridge between a nurse and a doctor, you can see a much more agile understanding of resource allocation because of that enormous cognitive and cost pressure that every health system underwent over the proceeding five years. At some point, do you envisage that 15, 18% of GDP declining on healthcare? What do you think is gonna go up but the value creation is greater? - Yeah, I think it's a really interesting question. And I'm not sure I necessarily have the answer. I think the age old question is, for your care, for your life, are you willing to pay 17%, 18, 20, 23, 25? It is a really interesting question that can be had. I do think that there's clearly limits to it and the more that we can create, more capacity and capabilities, that's ultimately gonna bring more care to more folks. But having a percent escalation in a given year over year, the reality would say it's not a sustainable model. And I think the interesting part again with the different kind of tools in the toolbox we talked about, things can be changed for the better. At the same time, we're right now living in an interesting world where there's more pharmaceutical therapeutics coming out that really have a chance to kind of change the game, whether it be in oncology, what immun oncology has done in the space, the new cell therapies that are actually coming out and then the Alzheimer's drugs. Some of the other interesting ones too are these ideas of actually having radioactive, small isotopes taken to a cancer and unloading the payload and killing the cancer. These are all really happening in the last five to seven years and I think that's going to drive better outcomes. But when it comes to this whole question about total cost, I think our view is is that we just have to do our very best job to be able to find ways to reduce that overall cost. The question is as other innovation comes in, how will that be layered on? And I do think we're in this window of time where we're going to continue to see some really interesting innovations on therapies continue to arise. Yes, I mean, the whole ladder of what we used to be considered exotic ideas has increasingly becoming sort of practical realities. And it's fascinating when we look through this. But if you step forward, let's say to 2033, what do you think the industry is going to be most proud about in terms of the changes that it's helped orchestrate? Yeah, I hope as an industry that we'll be able to look at and say access to all was increased significantly. And that's more than an insurance point. It's a point that says, how do you get to an elderly person that can't come in because we can come to you? I think this idea of everything like I mentioned in ultrasound, if you play that for in wearable monitors, you opened up Michael talking about the Fitbits of the World in Apple Watch is this idea of a longitudinal data that's collected on you wherever you are. And if you're homebound, actually can provide a really positive experience of taking care of you in the house. I think we're going to continue to see more of these alternate sites, but it's going to be enabled because we can collect the data, use artificial intelligence to actually help proactively manage that population. And then we're going to see this rise of earlier capabilities. I think of, again, the idea of screening and fine disease much earlier. If you can, it obviously has a great impact on a better outcome for the patient, but in nine times out of 10, it's less expensive for the health care system. And whether it be lung cancer or cardiovascular or valvular issues, whatever it might be, I think that's a really important part of this model. And this is where reach and changing of technologies can have a pretty profound effect. I'll just leave you one on this on the monitoring front, which we're very excited about. Most patient monitoring that you see in a hospital and such, all of that data really isn't archived. I mean, once it's on the monitor, it gets the right side of screen. It's gone forever. In the future platforms, that will be actually archived. And whether you're in your home in a acute care or a ward, be able to apply an algorithm against that to say, if your oxygen goes to this level, or your CO2 goes here, your respiratory changes, be able to predict when you might have an issue early enough in advance that it's way before the issue happens so that you can deploy a nurse or someone to intervene. Very, very much is what the future of care can look like. And that'll be brought to you by again, remote sensors, wireless capabilities, cloud computing, and artificial intelligent algorithms that can really help again drive better positive experience and also have better cost scenario in that case. But we think there's a lot of interesting things here on the horizon. And by 2033, I would also say this idea of multi-modal diagnosis, being able to assimilate these different components of data and actually have a single pane of glass that can actually help direct a clinician where they should go next. That's very much the path that we see is where the future's headed. If you were on the Nobel laureate selection committee, what one idea right now in your opinion in this domain has the most singular capacity for changing healthcare management in the world? - I would have to say how the foundational models are created, the ponds of the data that the algorithms focus on so that they are optimized in tune to be able to really leverage all the great things that we have talked about. Something in the AI space around that probably will play a larger role in that. And it goes both sides of being able to actually have a very clear view of all of the related ethics issues, the quality of the models to make sure that the output is checked and verified on the right types of standards and how that evolves actually into being kind of a standard, if you will, to drive confidence. Because I do think one of the things in the future that is today is how we increase people's confidence with the right infrastructure, the right support that the AI outcomes of these algorithms, not only are they reliable, but they're obviously consistent and that we have ways to check and verify that people can understand and see. - As you go through this, where do you think maybe the top one or two barriers is gonna be to get this future accelerated as much as possible? What's the thing, the friction that's gonna get in the way that we can sort of try and solve for? - Well, I think there's some of the things that I just mentioned, which is around security and privacy and being able to get whether it is industry enabled whether it becomes from different government policies, but actually having some ground rules that people feel comfortable with and confident it is probably a better term. than comfortable to be able to move quickly. I think that's one aspect. And from our standpoint, via our industry trade associations, working with FDA, in this case, to be able to drive some of those is how we think about it. I think the other aspect is that this point about our health care delivery, whether it be government-based systems around the world or private system here in the US, aren't going to have the money to just say, I want to build out a new system. It has to be done from the inside out as I referenced. And so the more that we can actually integrate, our devices with other products-- and again, what I mean by that is less about us acquiring and more about thinking about the ecosystem, how things operate in that care pathway. That's how we're ultimately going to get there. If an example of saying prostate cancer from early screening through diagnosis, through personalized delivery, well, what's the right medicine, what's the right follow-up structure, and that demonstrates a better outcome and ultimately a better cost from the system, how we duplicate and replicate those in other disease state areas, we think is ultimately the way to go about it, barring some larger transformation questions. But I think that's ultimately is how we think about it, which is why we get back to precision care saying, smart integrated devices we've talked about, this idea of thinking about certain disease pathways, and how do you solve for those? And then this integration of data, and particularly AI, to fundamentally help change the curve of a faster, more effective outcome, and a better cost associated with it. So I'm not going to ask you the dangerous question. You can definitely say, am I unable to answer it? So we've looked at 2033, we've looked about a disease pathways and diagnosis. We've looked at the fact that outside the government and military, healthcare is the largest industrial sector in the US, and probably in the Western world. When you look at this application of AI, you look at the pathway diagnostic, how much do you think we could compress and shift the cost structures in the industry, assuming that the ceiling is always going to be 15 or 18%. How much could you shift the net performance of the industry with these ideas? - The reality of it is, you end up typically escalating somewhat as you're getting the capabilities in place. But I think over the longer term, the ability to flatten them out. Again, in a case where you're not, you know, just take long cancer screening. We have reimbursement capabilities today in this country, for CT scan, for Lotus, lung scan. Only about 10% of lung cancer cases are found via screening. If that was 70%, they would be at such earlier stages the cost to treat them, and then not only that, the outcome results would be significant. But people don't come in early. And honestly, the longitudinal data and the focus of our system to actually drive people there, today isn't there. And I'm an optimist by nature and would say, with the changes that we're talking about here in driving, you are going to have primary care physicians, as well as the system be more incentivized to have earlier interactions, which also may be going to drive better health outcomes, and they're ultimately going to drive lower costs. But I do think there is, you know, more setup and structures that will need to take place. And with that, you know, we at GE Healthcare want to play a significant role, and now I have to kind of optimize your health system to take costs out, but help you think about how you implement the technologies to best get the best results for your patients, whether it be a new Alzheimer's drug or an oncology, they're an Aztec's drug, or you know, Tavar. And how do you find the right patients for an ARTC valve replacement? All of those need imaging, planning, and interventional and follow up. And, you know, the more we can make that more precise to rule out those that don't need it, but rule in those that need it sooner, it's just better care and better costs in the long run. - Yeah, one of the interesting things that's underlie this conversation, particularly around imaging, is there's always been this preset that AI is about data. Well, actually, imaging is data, visuals are an enormous portion of the world that we live around. And it is interesting that you've taken that start point because you can accumulate enormous volume of image, but if you can't have thoughtful, an almost instant diagnosis of them or at least prescriptive analytics, it's sort of wasted. And so I think this is a great example of how AI can get used beyond just the pure mass of a spreadsheet, in really practical ways with people's bodies. We all have partners that go through scans, you know, for whatever reason, and you're absolutely right. Early, easier diagnosis with more simplistic choice for next step is really a major step forward in healthcare where we spend a lot of our money at trying to solve a problem when we probably spend even more time on money on trying to prevent it or find it. - Yeah, and Michael, to your point, you know, I think you probably know this, maybe some of your listers down is, you know, a given hospital, one hospital in the United States has about 50 petabytes of data a year that they collect of some sort. And you know, that's, I think the Library of Congress would have the equivalent of about half that, just to put it in perspective, 'cause this is 4K, 8K data, over half of that by far is imaging data because of the size of that. And the fact is less than 25% of that really gets used at any level, less than that. That would be at a more premier institution. And so the data is there. The key is is how do you take that unstructured data and turn it into insights? And again, you mentioned that, as Sen, you mentioned some of the things that we're focused on, others are as well, but this is the journey. The data is there. How do you turn that into insights and then how do you have a more longitudinal view which ultimately helps drive better care? What a wonderful conversation. Peter, thank you for today. I will be referencing this in the new book in Visage about the world 10 years from now. But this was an inspiring, practical, transformative and extremely helpful. So thank you for joining us today. My pleasure, thank you, Michael. This is AI Data Riders from EDB. (upbeat music) [MUSIC]