Search

AI Democratization: CPU-Based AI for Healthcare – Futurum Tech Webcast

AI Democratization: CPU-Based AI for Healthcare - Futurum Tech Webcast

On this episode of the Futurum Tech Webcast, host David Nicholson is joined by Mohan Rokkam, Marketing and Product Manager at Dell Technologies and Steen Graham, Founder at Scalers AI to discuss the democratization of AI in healthcare, focusing on CPU-based solutions.

Their discussion covers:

  • The current state and future of AI in healthcare
  • How Dell Technologies and Scalers AI are working towards democratizing AI
  • Real-world applications and impacts of CPU-based AI in the healthcare sector
  • Challenges and opportunities in the adoption of CPU-based AI technologies
  • Strategies for scaling AI solutions in healthcare environments

Learn more at Dell Technologies and Scalers AI. Download our related report, AI on CPUs – A PoC for Healthcare, here.

Watch the video below, and be sure to subscribe to our YouTube channel, so you never miss an episode.

Or listen to the audio here:

Or grab the audio on your streaming platform of choice here:

Disclaimer: The Futurum Tech Webcast is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded, and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors, and we ask that you do not treat us as such.

Transcript:

Dave Nicholson: Welcome to the Dell Experience Lounge in Round Rock, Texas. I’m Dave Nicholson, Chief Research Officer at The Futurum Group. I am joined by Mohan Rokkam, who’s an engineer in technical marketing at Dell, and Steen Graham, CEO of Scalers AI. We’re here to talk about an interesting use case, a reference implementation that you created. In this case, we’re going into the medical side of things.

Steen Graham: What we built here is a kind of an industry-standard, medical grade human-in-a-loop radiology workflow. And I think what’s really unique about this is what we wanted to show is you don’t always need the highest-end GPUs. In fact, when you look at the EPYC processor line from AMD, we’re able to get a medical imaging stack up and running, compliant with the industry-standard protocols here and looking at all these great radiologists images and hopefully making radiologists and life easier.

Dave Nicholson: What’s connecting these servers together?

Mohan Rokkam: These today are connected using Broadcom 100 gig networking infrastructure. One of the things we have talked about is of course medical images. Right. So your X-rays can be not too big, but they’re a couple of megabytes each. You start going to your MRIs, your CT scans, those are literally hundreds of thousands of images put together. They can easily go into the gigabyte size and to be able to process them quickly and efficiently, having a strong network of… And today’s Ethernet is very efficient, very powerful, very fast. And so with all that, that was the infrastructure we thought was the best to go forward with.

Steen Graham: Just to complement what Mohan said is when you get to 3D medical images, in some cases they could be two terabytes, some of these really complex 3D medical images. And so the other thing about GPUs is you can be memory constrained. Where in a CPU-based architecture, we have a bigger memory footprint as well. And so you run into that problem. And then that’s where I think a modern CPU-based infrastructure really helps out as well.

Dave Nicholson: Often people associate AI with very, very specific tasks that GPUs are tailored to accomplish. Where some of this stuff is actually just fine for a CPU. Isn’t that part of the point?

Steen Graham: Yeah, I think that’s absolutely the case. I mean, what we’re doing here is this is kind of modern deep learning where we’re training a convolutional neural network to detect the medical images, which would be kind of synonymous with modern, modern-based AI techniques. Although today’s AI, everybody’s been inspired by these transformer models and these large language models that inspires everybody. But just like what you see today with the latest multimodal models that are starting to assess images, these modern-based transformer models all have the capability to look at images as well.

Dave Nicholson: So people have choice, Dell offers that choice. We talk about the democratization of AI. Where do you see this going in the future in terms of architecture and infrastructure?

Mohan Rokkam: Dell has always been about open technologies supporting the democratization of IT overall. So I think you may have seen in the news as well, we have a vast portfolio of offerings with a variety of CPUs, GPUs for a variety of use cases. I mean, we have our largest portfolios out there. So I see us giving customers a choice and the capability to solve all of their problems if they come to a one-stop shop that is Dell.

Steen Graham: This is a full on-premise kind of medical grade HIPAA-compliant implementation that you can do. And I think what’s great about this is obviously you can do it on off the shelf infrastructure as well. So if you think about analyzing a bunch of radiology images, I mean as long as it gets done overnight and you can call the patient back the next day or even within hours relative to millisecond latency when you’re talking about running a chatbot at a tokens per second level, right? So there, there’s the ability to use existing infrastructure. And the one thing, again, another thing that’s forgotten about CPUs is their general purpose.

Dave Nicholson: Looking at this from a medical imaging perspective, any surprises here? Any insights?

Mohan Rokkam: Just how scalable and simple a solution is, right? We are talking about something that can work with X-rays, with CT scans, a variety of modalities can be used and put into the system, may have changed the model a little bit, but the overall ecosystem is very simple, very straightforward, and very much something that today’s hospitals have in their environments today.

Dave Nicholson: Does this system detect and flag, “Hey, we think there’s an issue here.” And then a human validates? Is that how this works?

Steen Graham: Yeah, it’s human in the loop, and then there’s kind of a continuous learning element. So if the model does make a mistake, the model can then be refed the updates with that as well. Now for this case, in this reference implementation, we showed just pneumonia because it’s a large publicly available data set. That’s anonymized. So you can use it, but you’re going to want to train a custom model based on different types of medical images. So for example, there’s custom models you can do about bone age prediction, for example, and there’s a good open ecosystem of anonymized medical images actually available in the public domain. And then large repositories of private medical images as well that you can train these customer models. And what we’d normally do in this case is we use the off-the-shelf industry-leading computer vision models, and then we would ultimately use some transfer learning techniques to build custom layers to detect those particular medical images appropriately with high fidelity and high accuracy.

Whether it’s for the data science teams within large hospital networks or research facilities that want to build the models or they want to gain a production-grade deployment, and then Dell’s got a lot of other great leaders in the medical field that are leveraging their hardware as well. For these type of deployments, kind of those big names in the industry you can imagine are all innovating in this category as well.

Dave Nicholson: It seems like it’s a really big opportunity to be efficient and optimize things in a way that is really beneficial.

Mohan Rokkam: I mean, as part of the research for this, I was looking at some numbers. The Radiological Society of North America, they’ve said the same thing. They’re saying that they’re not enough radiologists out there. There’s an explosion of radiological images coming through, and it’s taking longer to come up for them to respond to these images. They are looking at AI solutions as well. I think this is a big problem that definitely needs some help, and this is the kind of innovation that AI brings to the table. We leverage technology to drive human progress. The average state of people has been improved with technology. I think AI is one of the things we should just take to the next level.

Dave Nicholson: On that hopeful, optimistic note, which I agree with by the way. Want to thank you again for joining us here at Dell’s Experience Lounge in Round Rock, Texas. Dave Nicholson, Futurum Group. Thanks again for joining us.

Author Information

David Nicholson is Chief Research Officer at The Futurum Group, a host and contributor for Six Five Media, and an Instructor and Success Coach at Wharton’s CTO and Digital Transformation academies, out of the University of Pennsylvania’s Wharton School of Business’s Arresty Institute for Executive Education.

David interprets the world of Information Technology from the perspective of a Chief Technology Officer mindset, answering the question, “How is the latest technology best leveraged in service of an organization’s mission?” This is the subject of much of his advisory work with clients, as well as his academic focus.

Prior to joining The Futurum Group, David held technical leadership positions at EMC, Oracle, and Dell. He is also the founder of DNA Consulting, providing actionable insights to a wide variety of clients seeking to better understand the intersection of technology and business.

SHARE:

Latest Insights:

All-Day Comfort and Battery Life Help Workers Stay Productive on Meetings and Calls
Keith Kirkpatrick, Research Director with The Futurum Group, reviews HP Poly’s Voyager Free 20 earbuds, covering its features and functionality and assessing the product’s ability to meet the needs of today’s collaboration-focused workers.
Paul Nashawaty, Practice Lead at The Futurum Group, shares his insights on the Aviatrix and Megaport partnership to simplify and secure hybrid and multicloud networking.
Paul Nashawaty, Practice Lead at The Futurum Group, shares his insights on AWS New York Summit 2024 and the democratizing of Generative AI.
Vendor Leverages Amazon Q on AWS to Drive Productivity and Access to Organizational Knowledge
The Futurum Group’s Daniel Newman and Keith Kirkpatrick cover SmartSheet’s use of Amazon Q to power its @AskMe chatbot, and discuss how the implementation should serve as a model for other companies seeking to deploy a gen AI chatbot.