A special ViVE and HIMSS conference sneak peak episode. What can we expect to see from VMware and NVIDIA in collaboration with iCAD and Rhino Health this year? Brad Genereaux, Medical Imaging Alliance Manager for NVIDIA joins Bill to talk about democratizing access to artificial intelligence across healthcare. What are the exciting advances that we can do with AI in the enterprise? How is it helping to drive efficiencies across every single department in the hospital? How do we empower IT departments with the virtualization stack from VMware to really demonstrate what AI enterprise means for hospitals? There will be an increased number of AI models running in hospitals five years from now, what does that look like?
NVIDIA: The way it's meant to be played
VMware - Delivering a Digital Foundation For Businesses
Rhino Health - Healthcare AI with Federated Learning
iCAD - Global medical technology leader providing innovative cancer detection and therapy solutions
Welcome to This Week Health. This is a conference campaign where we chat with our partners about the exciting 📍 initiatives they have going on in healthcare. As you know, we have a couple of great conferences coming up and we want to give you the opportunity to know some of the great solutions that will be at ViVE and HIMSS and how you can find them there. Let's get right to it.
Brad Genereaux Medical Imaging Alliance and Smart Hospital Alliance manager for NVIDIA. Brad, welcome to the show.
Thank you so much, it's such a pleasure to be here. I'm very excited.
I'm looking forward to this conversation. NVIDIA's doing so many really cool things. Such great solution, supporting so many different partners across healthcare. And today we're going to be talking about really democratizing access to artificial intelligence across healthcare. So, I'm really looking forward to that. Talk a little bit about what you're going to be displaying or looking at at the HIMSS conference.
Yeah, no, absolutely. So we're working very closely with our partners at VMware to show off some of the really exciting advances that we can do with AI in the enterprise. You know, AI has really taken off and it's moved much more beyond here's an application. Here's a proof of concept and let's see what we can do with it. And actually bring it into the mainstream and bring it into the enterprise. And take it from running within a single department and really bubbling up into corporate IT that's helping to drive the entire hospital business. What we'll be showing is how we can empower those IT departments with the virtualization stack from VMware to really demonstrate what AI enterprise really means for hospitals.
So AI in the enterprise. A lot of people will hear this and they'll think, oh yeah, really? We're starting to do AI, but I feel like we're at a tipping point. I'm now starting to look at real world solutions that are being implemented in healthcare systems that are using the AI stack. We're literally bringing it into healthcare. Talk about some of the solutions that are starting to implement AI in healthcare.
Yeah, absolutely. It's all across the board. You'll find NVIDIA instruments inside of servers, helping to drive visualization, helping to drive these AI applications to help us clinically and also in terms of doing workflow and departmental analytics. And population health. We'll see it in doing things like detecting cancer.
We'll see it in detecting hospital business and helping to drive efficiencies across every single department in the hospital. Where you'll find it what w what we're seeing with the it departments is as we start to provision these solutions and we'd go out and say, Hey, here's a great breast cancer detection, AI application and we're going to connect it inside of our hospital.
Back two years ago, we would go out and purchase a box, deploy it, set it up and it runs its own little, on its own server. But this is not something that can scale up and it's not something that can scale out. So if we want to add resiliency to those services. What happens if that one box goes down? If we want to add in a lung cancer detector and a pancreas cancer detector and a a liver cancer detector, we can't just keep adding boxes and boxes of boxes. We need to have that virtualization stack. And this is what I really mean when I talk about AI enterprise using VMware and video BGP view to help share that compute power across all the different applications that we're seeing in hospitals today.
You know, I'm one of those people that early on said, eventually we're all gonna have to go to the cloud. Cause you're not gonna be able to build out this AI stack in your own data center. And here we are a couple of years later and I'm eating my words because it sounds like when I talk about democratizing access to AI, that VMware let's us access it with the same resources that we once used to access compute storage and the network. And now we're, we're provisioning this AI stack the same way. And we're accessing that AI stack the same way. It really is kind of exciting and it does open up a lot of opportunities. Talk about some of the partners you're going to have and showcase at the HIMSS booth.
Yeah, absolutely. So, we'll have a number of different demonstrations at the VMware booth. We'll be working with iCAD and there's a session on Tuesday. This was in booth 2121 at the VMware booth. They'll be talking about how their iCAD breast health solutions help work with clinicians in terms of the breast cancer diagnosis. What they'll be showing us how this stack they can deploy now for enterprises run on top of NVIDIA AI enterprise with VMware helping to drive that virtualization layer.
So they'll be talking about what this means for IT departments who are responsible for deploying and supporting these applications driving forward. On Wednesday March 16th, there'll be a session with Rhino Health and the ACR. They'll be demonstrating using federated learning inside of hospitals to help drive the creation of AI across many different institutions. Going back a year, two year, three years ago, when we create AI, we have to pull all of our data together. All of our annotated data where we would do our training of AI models. This is problematic in healthcare because it typically, we want to get data sets from many different modalities, many different instruments, many different instrument vendors.
But these are across many different hospitals and we have to really think very carefully on protecting patient privacy. With federated learning, with what Rhino Health will be demonstrating is how we can actually leave the data at the hospitals, but train the model together. And what we do to drive using NVIDIA flair which is our SDK to help drive this, is move the training from a centralized place and have it run on every individual hospital where the data is. And then we share that AI model back up and create one giant AI model that generalizes for all those different hospitals. But we're still protecting that patient census data at all all the individual locations. So this is what we demonstrating at the booth on March 16.
You know, one of the things I love about these models is they are truly a platform. And when you create platforms like this, the more data that runs through it, and the more times the algorithms run, the smarter it gets, and the more the more value it creates for healthcare. I'm excited about this. I'm really fascinated by the opportunities. As a CIO and I'm looking at this how do I know when I'm just going to use normal compute or when I'm going to use an AI stack?
I think probably most projects start out on normal compute. Right? If I'm saying let's evaluate this proof of concept. Can I detect breast cancer in these mammo images? I could do that one off. One at a time on a single node and, and just do that work. And I could do that up in the cloud. I could do that locally. It probably doesn't matter. As I start to think about scale, right? As I start to think about, this service is going to be depended on by my clinicians. I need to make sure that it's going to remain up. I have, four or five nines of uptime. I had to have multiple boxes where I'm running this. And again, I might run running in my data center.
I might run it on the edge. I might run it up in a hybrid cloud in order for the store. So there's a question of scale and resilience on that front. On just that one AI model. On just that one use case, being able to support what my hospital's going to need. What I look at on the, on the flip side, the number of AI models that I'm going to have to run in my hospital five years from now, what does that look like?
I used a website it's called gamuts. It's put out by RSNA and the ACR and really it lists the number of things we could see in a medical image. And there's about 12,000 things that I could see in medical images, according to that website. And if I need to have infrastructure to run all of those AI models and that's just medical imaging and forget about everything else that's running in the hospital, how do I ensure that my data center is going to be able to support all of those different models that I need to have going forward. Even when I look at AI models that are very much focused on, Hey, I can run just on a CPU. I'd just run it on one box and it's virtualized. That's great. But what happens when I now add three clinics, six clinics, 10 clinics that are growing, that are using those services that are needing you know, a one minute turnaround time. How do I make sure that I'm able to scale up to support the need of those clinics as I could do to build out my hospital enterprise.
Thanks for that explanation. That helped me a lot. As I think about that, we're going to be processing images in ways we never had before. Before we used to just process them, figure out where to store them. We'd put them there, bring them up, let a person look at them and figure it out. But now what we're doing is we're, we're putting these images through and looking for 1200 different items depending on the type of image. But looking for 1200 different items on each one of the images as it goes through. And so now when we talk about all those, the slices of those images and everything else, it just gives me a scale on the amount of compute that we're actually talking about here. I mean, it's, it's a significant amount of compute. And so I imagined because of the platform that VMware springing, I can go from one model where I'm trying it out, I'm doing it on normal compute. I can then make the transition over to a different platform with different technologies that are tapping into those images. So it's it's exciting. So are you going to be at the HIMSS conference?
I will be there. I'm so looking forward to it. It's been a couple of years since I've last been and I'm looking forward to reconnecting with all the great partners, all the great vendors, all the great people that go I'm super excited for that.
So you're going to be at HIMSS booth 2121. I will definitely be stopping by. I want to take a look at the solutions and what you guys are doing. I think it's going to be exciting. That's the VMware booth again. It's at HIMSS. It's 2121. Brad, thanks again for your time. I really appreciate it.
I hope everyone is excited as I am to get together again at the HIMSS conference. Definitely visit us at VMware booth 2121. We're also going to be recording two podcasts with them in their booth, 📍 both on Tuesday and Wednesday with NVIDIA. Thanks for listening. That's all for now. 📍