In this special episode of HealthTech Remedy, recorded live from HLTH 2025, Dr. Paul Gerrard sits down with Roland Dias, a leader at GE Healthcare, to discuss the profound transformation happening at the intersection of medical devices, data, and artificial intelligence.
Roland shares his unique career journey—from Medtronic to Microsoft and now GE Healthcare—and explains how his background in both devices and cloud AI is shaping their new patient-centric strategy.
Tune in to explore:
The "Three-Legged Stool": Roland's framework for driving real-world AI adoption by balancing clinical benefits, IT simplicity, and a strong ROI.
AI Beyond Diagnostics: How GE is moving AI out of the lab and into therapeutic devices to enable adaptive therapies and better decision-making.
Patient-Centric Care Pathways: Why GE is shifting its focus from selling individual machines to providing end-to-end solutions for oncology, cardiology, and neurology.
A 10-Year Vision: Roland's exciting look at the future of healthcare, including AI-driven early detection in your local pharmacy, true personalized medicine, and the rise of the connected "home care" ecosystem.
This conversation is essential listening for anyone interested in how one of healthcare's largest incumbents is pivoting to become a digital-first company and what it means for the future of patient care.
Introduction
Dr. Paul Gerrard: I want to welcome Roland Dias from GE Healthcare. Roland, welcome to HealthTech Remedy. Thanks for joining us.
Roland Dias: Thank you for having me. I appreciate that.
Dr. Paul Gerrard: Let me start off with a question about how you got to where you are. You've had an impressive career at the intersection of engineering, healthcare, and leadership. You had roles at Microsoft and Medtronic before coming to GE Healthcare. Can you share with us a little bit of that history and the path that brought you to GE Healthcare and what drew you to your current position?
Roland Dias’s Journey: From Medical Devices to Cloud AI
Roland Dias: Oh, good question. I started back in Medtronic many years ago. And what drove me to Medtronic was this opportunity to really help customers, physicians, and patients. That was the desire to see the vision and mission that was at Medtronic and how we can develop customers.
And manufacture and then commercialize devices that can actually help patients, whether it's a pacemaker, a defibrillator, a neurostimulator, or we're building an at-home hemodialysis machine. That was my driver to go there. And it allowed me to get really close with our customers. The intimacy was beyond, beyond. Typically in companies, they talk about during surveys and market research, and it's all good to get good data, but unless you get into their environment and truly understand what's going on and understanding their challenges by observing, that's when you really begin to appreciate what their challenges are. And then when you come back with the solution, now you're making the connection.
That was my way in. I'm an engineering by training, so I love devices. And it allowed me to leverage my engineering piece, but also to leverage my desire to help and to be in the healthcare space. So it was a nice match made for me. And from there, I moved on to Microsoft because there was an area that I felt I needed to know about, and that was more about cloud and LLM and AI. And although Medtronic has done a pretty good job of incorporating AI and algorithms into their devices, they're not the Microsoft of the world. And so I was called by Microsoft to actually come on board to start up a business where they had been working on it for some time, but couldn't really get it going.
The idea was, how do we take clinical AI, integrate that into workflows within the radiology channels, and then also downstream to other ologies? So whether you're a cardiologist or a neurologist or a pulmonologist, how do we take that good data and then make sure it seamlessly integrates into those workflows so they can then use that to make a decision? And for the three and a half years that I was there, that's what I was doing, was building that business. And so I was working with providers, but also working for payers and working with medtech and pharmaceutical companies to say, how do we help you now that we have access to that data?
How can we help you, for example, on the payer side for prior authorizations? What can we do there to streamline that? That's a huge problem within the hospital system. And it takes a lot of time and energy to do that. But how do we streamline that as well? So we were leveraging our technology to do that. But then after about three and a half years, I got this urge to go closer to the physician and the customers, like I had with Medtronic. And GE had called me, GE Healthcare had called and said, hey, we'd love to have someone who has both the device and the tech side experience. Because as we look forward to care pathways, we want you to look at solving challenges, both thinking about it from a device perspective, but also from a tech perspective.
And so my first meeting back at GE was with a group of interventional cardiologists. And you can imagine sitting in an interim advisory board and talking about structural heart and having them talk about their challenges today. And then for me to look at it and say, I've got an idea from an IT perspective, but then I also have an idea in terms of a hardware perspective, how we can help you to overcome your challenges. So for me, it was coming back home. And that's been the beauty of GE for me, because now I had the best of both worlds.
Dr. Paul Gerrard: You have all these prior experiences. Can you talk about how that shaped your approach when you joined GE and how that perspective helps you lead the Care Pathways agenda today?
Defining GE’s Patient-Centric Care Pathways
Roland Dias: Yeah, that's a good question. And so from a Care Pathway perspective, and just to let you know, we're still evolving Care Pathways. But we wanted to focus on three main areas from a care pathway perspective. One is in the area of oncology, then cardiology and neurology. And within oncology, we're not looking at every single patient pathway, but we're looking at breast, lung, and prostate. When we go into cardiology, we're looking at CAD, coronary artery disease, EP, electrophysiology, and then structural heart. And then within neurology, we're looking at Alzheimer's. There's a lot of different neurodegenerative diseases. We can't hit them all at once. We want to start with Alzheimer's first. So those are the three areas that we're looking at, and then within them, those are the patient pathways.
I think the way that you can think about care pathways within GE is we start with the physician and the patient first. So I don't go to a physician and say, hey, we've got this wonderful MR that does all these things, or this wonderful CT that does all these things. You start with, let's talk about the patient pathway for breast or for lung. From the very beginning, from the point of diagnosing to treatment to monitoring, but then even before diagnosing, what happens to that patient before? And then as monitoring, what happens when that patient goes home? So from a care pathway, we're truly looking at the end-to-end journey. It's not episodic treatment. It's longitudinal. How do we better care for our patients along that journey? And what I like to do is to say, tell me what the journey looks like today for you, customer.
Tell me what grade looks like for you, customer, based on what's best for the patient, what's best for you as a provider. And then show me where those barriers to grade are, and let's talk about those. Because some of those barriers, we may have in our roadmap already. You just don't know about them yet. And then there may be some that are not on our roadmap and we could start developing them. Now we could do it on our own or this could be a co-development opportunity for us. And so we look at those and then what we do is we come back and we put it all together and say, if we're able to bridge these gaps or these barriers to grade, this is what the end-to-end solution looks like for this patient pathway for Alzheimer's or for prostate or for breast cancer. And this is what allows GE then to differentiate.
From the competition from that perspective, because we're not looking at just selling an MR or just a CT. We're looking at end-to-end, how do we help to take better care of that patient as they move through that journey?
Dr. Paul Gerrard: Yeah. So just to reflect on that, when I go into the hospital, I can see lots of machines and devices with GE on it. But what you're really talking about is not at all device or machine-centric. It's really patient-centric, healthcare provider-centric. How does that end-to-end process look? So, somebody's thinking about GE as a machine and device company. Do they really have the wrong rubric in their mind?
Roland Dias: Well, they have that in their mind because that's who we were in the past. But as we go forward, GE has to evolve. And that's the digital transformation that we're trying to move towards. Now, again, it's not easy because we've been the GE that you just mentioned. We've been a hardware company forever. And that's where people think about us. They think about digital and AI and large language models. They think about the hyperscalers like Microsoft, AWS, Google, when you start talking about that. But it would be great, and it will be great, when they start thinking about LLM and AI, they think about GE as well. Because we need to move beyond just a device.
And today, an MR rep talking with a physician is only thinking about that patient who's going to have an MR. It's faster, better, higher quality, blah, blah, blah. But they never ask what happens and what are your challenges before the patient gets to the MR? And then what happens when the patient comes through that MR? What happens to that patient? Those questions are not being asked, but now they are through care pathways.
Dr. Paul Gerrard: Let me pivot a little bit here and mention that one of the early promises of AI in healthcare has really been focused on diagnostics. My full-time job is at a diagnostics company and same thing with my co-hosts. But we've seen some announcements from GE about starting to integrate AI into some therapeutic devices. Can you tell us a little bit about that? Because that seems to be something new where we haven't really, at least we in the healthcare community, haven't really been thinking about an application of AI.
Beyond Diagnostics: Integrating AI into Therapeutic Devices
Roland Dias: That's a very good question. And I think AI for diagnostics is not going away. That will always be there. And we'll talk about the adoption of AI in that world because there's some challenges there. But AI beyond diagnostics, when it comes to helping to make a better decision for the care of that patient, it's certainly going to evolve, for sure. And so you think about whether it's in the world of oncology, when the patient has been diagnosed with breast cancer, lung cancer, prostate cancer, what's the best treatment plan for that patient? And if we can create LLMs that have been trained and have access to multimodal data, genomics, pathology, imaging, labs, et cetera, all the ones that are out there, and then can take all that information historically and then present to the physician some options.
Given what this patient has, this would be the best path for this patient and you would get the best outcome. So that decision-making becomes a lot clearer and there's a higher probability for success. Or if you are in the RAD-ONC space and you're doing beam therapy and you've done all the planning and maybe you didn't leverage LLMs to give you that first plan and you start out. And as you're going through, if you see that your first plan is not exactly working to where it should be working, working at different metrics like PSA, for example, on the prostate side, then you can adapt your therapy along the way. So adaptive therapy is going to be another area where you can leverage AI to say, here's what we've decided. This is the dose. However, it's not working. Let's adjust accordingly and then look at it again and then adjust accordingly and look at it again. That's going to be another area where we can actually help as well.
So I think you're going to see a lot more of that coming in. You'll also see a lot more in upstream, for example, with patients that have dementia and Alzheimer's. So if way upstream, we can create tools where not only does it go through the, let's make sure this patient's, the reason why they have cognitive decline is because it's leading to dementia or Alzheimer's. But can we also assess the risk of that patient for ARIA, for example? And if we can do that, then we can start risk stratifying these patients. But leveraging the tool to risk stratify because not everyone needs the same amount of resources. So we can start leveraging those tools to start risk stratifying.
Then you put the patient on the right pathway, which also helps the hospital because everyone is overwhelmed and very busy and the burnout is high, but we can help to risk stratify so that every patient gets the same amount of resources because we can help risk stratify using these tools to do that.
Dr. Paul Gerrard: You just mentioned at least three different diseases and maybe more. I think I lost count there. It sounds like you're trying to boil the ocean here. Obviously no one company can do that. So how do you think about among these various diseases that are all very important? How do you prioritize?
Roland Dias: Yeah, that's a good question. And I think when we look at the three areas that I talked about, and again, if you think about cancer, how many different cancers are there? The liver, the pancreas, they're all different diseases. They're all different diseases. And so for us, it's where we think, based on evidence and based on what we know, communicating with customers, what are the ones we should really focus on that if we were to solve for them, have the greatest impact, not just impact for GE, but impact for the customer and that patient. And that's what guides us. And that's why I only said there's only three in oncology, there's three in cardiology. It could be a lot more. And then one in neurology. So we're starting with those because during our research, we understand in having those conversations that those are the areas where if we could solve for these problems within those pathways, we're going to make a big impact along that. So I think it's been a great journey. And again, to your point, we don't want to spread ourselves too thin and not be effective. And so we really do want to focus on a few things.
And there's tons of pathways out there, as you know. And everyone talks about what about that pathway? What about this pathway? But we happily say no at this point, because we need to focus on these to really make a big impact. But in the future, once we make this right, then we can start to explore and other things. But I'll tell you that, and I know you know this because you've been in it long enough, but within my role within Microsoft, I had a lot of great experience with AI companies, typically startup companies. And you probably know a lot of them that are out in the marketplace. And they come and go. And these are clinical AI tools because Microsoft didn't build those. And so the idea was to put them in a marketplace and then seamlessly deploy them. And the idea was that from a hospital perspective, you think about a hospital, they have a lot of AI companies banging on their doors, the little mama pops, hey, I want to get in there. You can do all these wonderful things.
But the IT folks don't want to take the risk. There's a lot of cybersecurity, et cetera, that goes on. And so they try to minimize that and say, if we could just have one company that we can have access to all these wonderful AI tools for radiologists, oncologists, and other ologies, where we can seamlessly integrate. And then you have a marketplace where we can pick and choose which ones we'd like. That would be hugely helpful from a hospital perspective. That's what we're doing at GE is how do we create that ecosystem so it's one connection into the hospital, into a data fabric layer, if they'd say. So you're pulling in pathology and imaging and labs, and then you're managing that data, and then that data supports what's up above, which is the AI layer and security layer, which then leads to the cloud where all those apps sit.
Creating this tool that we can build for our customers to be able to use different AI algorithms to do things like contouring and segmentation, lesion detection and characterization, which everything is sitting in the cloud that you can access pretty quickly. And it's separated from the machine, but you can use those tools. So say for example, on the oncology side, you want to contour the organs. So you can imagine if it's in the cloud. And you can access that by sending the image, it does it within five to six minutes, pushes it back down again, and now that's ready for you to look at. Rather than you spending hours contouring certain organs. You can imagine how long it takes. But now everything's in a cloud. It's ubiquitous, and you can use it on demand to actually do those things. And it saves you hours of work. So that's where we're going to go, because I think that one connection into the hospital system is going to be huge for us. Because once we have that, everything else is built up and it's easy now to help those hospitals to do what they need to do.
Dr. Paul Gerrard: One of the big challenges that we've seen in healthcare is getting even great innovation that helps patients, getting it adopted, getting it into those clinical workflows. But it sounds like you guys are proactively tackling that as the first step and then plugging innovation in afterwards. Is that a fair assessment?
The Three-Legged Stool: Driving ROI and AI Adoption in Healthcare
Roland Dias: There's a couple of things. One is, one of the challenges that we see is that trusting AI is still not there yet. There's a lot. On the administrative side, it's a little bit easier. But when you get on the clinical side and helping to make decisions, that's not there yet. So we have to do a better job of, one, building models that work. Number two is have a system that monitors the performance of the model over time. So day one, it should work really well. But day 365, it should still work really well. And then beyond that. And honestly, there are some that work pretty well and there's some that don't work so well. And until we get to a point where we can build that trust and it consistently works well, we're going to have this slow roll and adoption. That's number one. Number two is often companies don't put in the time to truly understand what are the challenges and the needs.
And I'll go back to my experiences with Medtronic, but now with GE as well, is that when we're going in to assess needs, we're in there side by side with our customers, truly trying to observe their workflow and understand what actually is going on and seeing it. Because sometimes when you do a survey, you may answer questions, but they may forget a few steps along the way that may cause pain. And you don't realize it until after the fact, when you come out with a solution. It's too late now. Now I'm going to have to revise it. But I think for us to be able to get in and understand that when we're trying to solve problems, certainly it's in the clinical setting. So the RAD-ONC and the MED-ONC and the nurses and the APs that are surrounding that area that are working and caring for that patient. But then there's also other stakeholders in the hospital system. And if you don't think about them, you will certainly will after when you try to launch a product, it doesn't really move the needle. So I call it the three-legged stool. One is it's got to show a benefit from a clinical perspective.
Workflow efficiency, got to be there. Number two is from an IT perspective, it's got to be light and simple to deploy. If it's beyond a month, you're lucky if an IT person will talk to you. And I know you know that. And then number three is you have to have a good ROI story. If it doesn't have reimbursement, not a CPT code one, there's no reimbursement for it's going to be challenging. But if you have something that says based on the workflow efficiency and the clinical benefits, we can turn that into an ROI story that I can either take down your cost by this much, or I can drive incremental revenue by this much by having this tool. Those three things have to be there. If you don't check the box on those three, it's very difficult to drive adoption within a hospital, to even sell it to a hospital. Because as you know, every hospital system is challenged with funding. Everything's already spent. So for them to then buy a digital solution, they have to take from somewhere else. And then it becomes an ROI story. If I take from here to put here, what's the ROI? And many of these companies fail because they don't think about the ROI. They focus so much on the clinical workflow optimization piece, which is important. But if you don't have a good ROI story, it's not going to hit the market. It's going to be very difficult to drive it up. You may get it in one or two.
But from my mind or the way I think about it, scaling is about how fast you scale and how large you scale. So for me, in our first product that we launched, my goal was to be in 1,000 accounts, 2,000 accounts, 3,000 accounts, and not over a 20-year period. I'm talking about the next five to seven years, what do we need to do? But you have to do those three things really well. And so AI algorithms that are taking six months to a year to deploy, very difficult to scale. So when we launched our first iRT product, we were a little bit over 30 days, but the latest readings that we've had have been less than two weeks in deploying this tool. That's where, and we're building an ROI story and it has improved, it has a clinical benefit and workflow efficiency benefit there as well.
Dr. Paul Gerrard: So let me talk about the distant future now, not even five to seven years. Let's look 10 years out. What does success look like and what excites you? And feel free to answer this from the standpoint of GE or even our healthcare system more broadly.
A 10-Year Vision: The Future of Early Detection, Home Care, and Workflow Optimization
Roland Dias: Yeah, there's a lot there. Just coming out of class, there's a lot going through my brain, but there's a couple of things that really get me excited. One is on the early detection screening side. If we can help from a GE perspective to identify before the symptoms. And this gets into preventive care, this is, for me, really, really exciting. So you can imagine having, for example, a portable CT machine. So low-dose, CT machine, no contrast, portable, sitting at CVS, Walgreens, any medic clinic, ubiquitous everywhere, in every country. Where your patient could interact with it, no human in the loop, using agentic AI to put in all the information it sends to your phone all that information plus your PCP you go into this little room looks like a little tube.
And it's kind of like a phone booth, if you will. And only a little more claustrophobic. But now with this, it's all glass, so you can see in and out, which is great. So it takes away that, but you can imagine doing a CT with less than a minute and it's enabled by AI looking at just a chest, for example. And in that you've got AI that looks at heart chambers, looks at the aorta, looks at coronary artery. Looks at emphysema if there is any. So it may have six or seven, eight different things it's looking at and then sending the report to you as a patient via your phone app, but also for you to your PCP, all done in a minute or so. So it's not just the blood pressure cuff you're doing at CVS, now you're doing this. So for me, it gets really exciting because if we can identify early and start treating early, wonderful for the patient, wonderful for the payers and providers. Number one, I think that's going to be huge. That will come in the future for sure. The other one is this personalized care.
It's on its way. The world of and the way of managing patients where one shoe size fits all is not going to work. So we have to get into personalized medicine, whether it's in oncology or neurology, that's going to come. And we're going to leverage LLMs to do that. So you're going to see that continue to go forward. Another area, which I'm going on a limb here, but I think AI autonomous read from mammo, for example, or X-ray, is the future as well. How do we leverage AI to first be the first reader, and then triage only when they're flying something. It's going to be massive. It's not here today, but in the future, that will be there. Because these things that I'm sharing with you, this one in particular, helps to reduce the physician burnout. Because it will take the place of that. Now, radiologists will still see images where they found something, but when they didn't find anything and it's negative, that's not what they're going to see. And so now we start to optimize that. That's going to be big.
I think the workflow optimization is not going to go away. Burnout is here. We've got to figure out how to optimize. Let's take away the administrative tasks so that physicians can focus on what they came to do. And that's care for patients. And so we will continue to do the iRTs as one example of many that will come through. Think about iRT optimizing the workflow within RAD-ONC. And what iRT does is think about a patient who's been diagnosed with cancer. The RAD-ONC has now said beam therapy because two-thirds of patients will go through RAD-ONC and get beam therapy.
So that day the patient has the intake with the RAD-ONC. They have the conversation, they may do the CT sim, and then they say, hey, come back in 30 days, we'll be ready to treat you. Maybe even longer than that, depending on the hospital system. What iRT does is it optimizes the workflow and all the tasks that need to be done. It organizes them and shrinks on that time from the time that I have the first conversation to the time I treat that patient, which is today, weeks down to same day. That's what iRT does. Think about the benefit of the patient who has cancer, wants treatment right away, not 30 days from now, not 60 days from now, that can get treated that same day. That's what iRT really does. And it becomes a hub where all the information that's needed, the planning piece, all of the treatment from the piece from the iRT, plus all the information on that patient, all comes to one location, not 10 different locations, where now it optimizes that process. And everyone can see it and optimize the workflow so that that patient gets treated faster.
And so that's great. So I think the workflow is not going to go away. I think the other piece we'll see is home care. Home care is going to be massive. It's already starting to show that it can benefit. And there's still more work to be done proving that there's a clinical benefit and there's an economic benefit. But I see that going in the future to be massive. And I think GE can play a part there as well. We don't play that now, but I think about miniaturizing and mobilizing our devices. So think about the Vscan Air, which is ultrasound portable, is a perfect example of now you can have the home care team using the scanner. As I mentioned, the portable CT, if it's on a van or a truck, when you roll into the area to care for those patients, you can then do a portable CT, use that portable CT for scanning. So I think home care is going to be massive.
The other piece is this data fragmentation. So if we can leverage our tools that can actually connect the information that we're pulling from the home back to the med-onc who's managing that patient, where you now have information flow, that then helps that physician to make better decisions as to care for that patient. Because right now it's fragmented. If you go to a mini clinic, they have their own information. If you go to the PCP, they have their own information. If you go into ED, and then into the hospital, none of these things are really connected.
Dr. Paul Gerrard: Wait, you can fax notes.
Roland Dias: Yeah, exactly. Can you imagine that? I think this is the only industry where we're still faxing stuff. It's absolutely crazy. So I think those personalized care, early detection screening, AI autonomous, workflow optimization, home care, big things. And AI will always be around for diagnosing and then later on to treatment and therapy and decision support. So I think it's pretty exciting, but it's got to take some time. It's not going to happen overnight.
Dr. Paul Gerrard: All right. Well, I think those are some great positive notes to finish on in a world where we're always talking about healthcare challenges. So let me thank you very much for your time. And that'll be a wrap.
Roland Dias: Thank you, Paul. Appreciate it. Thanks for the opportunity.
