What does NVIDIA do in healthcare?

[00:00:00] Dear listeners, welcome to Faces of Digital Health, a podcast about digital health and

[00:00:05] how healthcare systems around the world adopt technology.

[00:00:08] With me, Tjasa Zajc.

[00:00:19] When it came out, ChatGPT became the number one poster child of how amazing AI can be.

[00:00:27] In less than two years apart from ChatGPT there's probably no one that doesn't know

[00:00:33] or hasn't heard about NVIDIA.

[00:00:36] NVIDIA makes the chips that are crucial to training and operating AI systems and ChatGPT

[00:00:43] also runs on NVIDIA chips.

[00:00:46] The company is known for its graphic processing unit or GPUs and is often mentioned as the

[00:00:53] company that drives advances in AI, gaming, creative design, autonomous vehicles and robotics.

[00:01:00] But actually, the company also has quite a footprint in healthcare.

[00:01:05] And this is the topic of today's episode.

[00:01:09] You will hear from Dr. Chelsea Sumner, a healthcare innovator who currently holds the position

[00:01:15] of the North and Latin America and Africa Healthcare AI startups lead at NVIDIA.

[00:01:21] In the past she worked as a pharmacist and in the pharmaceutical industry but

[00:01:27] today she's going to explain how does NVIDIA work with startups in healthcare, which startups

[00:01:34] might potentially be interested in reaching out.

[00:01:39] What's the broader impact of the company on healthcare?

[00:01:43] What do they invest in?

[00:01:45] Which areas do they develop and much, much more?

[00:01:49] So enjoy the show.

[00:01:51] And as I always say, if you haven't yet, make sure to subscribe to the podcast wherever you listen to your shows.

[00:02:00] Additionally, do check out our newsletter.

[00:02:03] You can find it at fodh.substack.com.

[00:02:07] That's fodh.substack.com.

[00:02:10] The editions include an overview of healthcare in Africa, many countries in Europe,

[00:02:18] an overview of the development of digital therapeutic space and much, much more.

[00:02:26] So go to fodh.substack.com.

[00:02:29] That's fodh.substack.com.

[00:02:33] Now let's dive in today's episode.

[00:02:51] Chelsea, hi and thank you so much for joining me here on Faces of Digital Health.

[00:02:56] To talk a little bit about what NVIDIA is doing in healthcare and your role working with healthcare startups in North America and Latin America.

[00:03:07] And yeah, South America, you actually work with 1,600 companies.

[00:03:13] And I just want to dig into various things here.

[00:03:17] I think the first one would be that NVIDIA is really known as the leading provider of GPU in DEC.

[00:03:25] And I'm not the only one who actually wasn't really aware of the footprint that the company has in healthcare, but it's not really small.

[00:03:33] So can you talk a little bit about that, describe the development of various sectors in the company and healthcare?

[00:03:42] Absolutely. And thank you so much for having me.

[00:03:44] I'm excited to be here and to tell you a little bit more and to let you know that you are not alone.

[00:03:49] If you'd asked me this question four years ago, I wouldn't have been able to answer it.

[00:03:53] As you said, I've been with NVIDIA for about three years now and prior to that was a pharmacist.

[00:04:00] So it was completely shocking to learn.

[00:04:02] But as I dove in and got to learn more about NVIDIA and how we've been in this industry for well over 15 years now,

[00:04:11] it was super, super exciting to just see it all because as NVIDIA created the GPU 30 years ago,

[00:04:18] 30 plus years ago, we're a 31 year old company.

[00:04:21] And after we've created the GPU, just the speed of which technology began to move just vastly increased.

[00:04:29] There, of course, was transformers that changed and revolutionized the industry.

[00:04:34] And then what we call our AI moment or AI revolution moment happened, of course, with chatGPT taking off in 2022.

[00:04:43] But really, we just have been able to just take the industry and just transform it, but not just healthcare,

[00:04:50] but beyond healthcare across all different industries.

[00:04:54] Jensen probably said it the best earlier this year at the JP Morgan Health Conference where he talked about how

[00:05:00] in his fireside chat with the recursion CEO, he talked about how NVIDIA went from computer-aided chip design

[00:05:08] to computer-aided drug design and how everything is going to be done in silicone.

[00:05:14] And if you think about that, it's just, wow, we really can do so many things and get so many things virtualized.

[00:05:22] And he just shared how in his vision every industry is going to be a technology industry,

[00:05:27] which if you think about healthcare, it's like, how can healthcare be a technology industry?

[00:05:32] And he dove into that in detail about how 15 years ago, Mass General published a paper showing how they were using our GPUs

[00:05:42] for CT-imaged reconstruction.

[00:05:44] And this was like back in 2007.

[00:05:47] And in the paper, it talked about how they had an 80-time speedup over a CPU with improved accuracy,

[00:05:54] which is unheard of and especially at that time.

[00:05:57] He shared more about this, what's called NAMD or Nanoscale Molecular Dynamics.

[00:06:05] And this was developed by University of Illinois at Urbana-Champaign.

[00:06:10] And that technology is still being used today.

[00:06:13] They just released the NAMD 2.11 version and it runs seven times faster on NVIDIA GPUs.

[00:06:22] But basically, you can think about it this way.

[00:06:25] Artificial intelligence and technology and every person, I think again something Jensen says all the time is that everyone can be a programmer now.

[00:06:36] And so that's really going to change the way of life, especially within healthcare, within our healthcare specifically at NVIDIA.

[00:06:44] We have our NVIDIA CLARA, which is very adequately named by the founder of the American Red Cross, Claire Barton.

[00:06:54] But we have technologies that spin everything in healthcare tech.

[00:06:59] So we've got technologies for genomics with parabricks, with robotics, like our ISAC technologies.

[00:07:07] We've got MONI for medical imaging, HOLISCAN for medical devices.

[00:07:12] Our BIONEMO platform for drug discovery.

[00:07:15] And most recently, our James, our digital human avatar, plus our NVIDIA inference microservices and our NEMO for generative AI, all for digital health.

[00:07:27] And so how you can use all those models together to really just transform just in general how healthcare and technology are this AI revolution.

[00:07:36] So that kind of makes me wonder, how do you limit yourself in terms of who do you work with?

[00:07:43] Obviously virtual reality in healthcare, things like imaging, that makes total sense.

[00:07:50] It's an obvious fit for the company.

[00:07:53] But when we look at the 1,600 companies and new ones potentially, who does it make sense to reach out to you?

[00:08:04] And what kind of companies are just not in your focus area?

[00:08:09] And I was also, I was listening to one of your previous interviews and I was thinking it would be good to actually start with the definition.

[00:08:16] What a startup is for you.

[00:08:18] How mature companies are we talking about in this batch of 1,600?

[00:08:22] Because a lot of times, I guess in the European sense, a startup would be a company with 5 to 20 people.

[00:08:29] And I remember talking to a friend from the US at one point and he's, yeah, we're a small startup and I'm like, how many of you are there?

[00:08:38] And he's like, oh, 800.

[00:08:39] And I'm like, okay, we've got a bit of a cultural difference here in terms of what's big or what's a small startup.

[00:08:45] So how do you look at that?

[00:08:48] Yeah, that's such a good question.

[00:08:49] And you're so right.

[00:08:50] But luckily at NVIDIA, we use the term startup pretty loosely.

[00:08:54] And I'll even boil your mind even more.

[00:08:56] We have over 3,000 startups, healthcare startups within our inception program.

[00:09:01] So the NVIDIA inception program, I cover the ones regionally focused in North and Latin America and Africa.

[00:09:07] So roughly about 1,500 of those startups.

[00:09:10] And so when I say the term startup and then when I'm referring to our inception program, we can mean two people in a garage as well as two companies.

[00:09:18] Like you've said that are 800, a thousand plus people who are generating millions of revenue.

[00:09:24] So we use that term because we believe that every company as a developer, being that developers are at base and that we're providing the tools for the next generation.

[00:09:35] And so we're working with startups that are very small and very early stage to companies that are in their series C and above rounds.

[00:09:44] It just depends.

[00:09:45] But the goal is that to showcase that NVIDIA is our partners in terms of the entire ecosystem.

[00:09:51] So from the smallest of startups to the largest of startups, we're going to be willing to create tools that can help and support them all.

[00:10:00] And what exactly do you how do you collaborate with startups?

[00:10:05] How focused are you on looking at acquisition targets so you invested or acquired quite a few companies in the last few years?

[00:10:16] So can you talk a little bit about that?

[00:10:18] I think many listeners with their work in healthcare might be interested in.

[00:10:21] If they can work with you or not.

[00:10:25] And also just to add that given that your focus is on the US or just North America and South America, how big of a footprint in healthcare does NVIDIA have globally?

[00:10:40] Yeah.

[00:10:40] So in other continents?

[00:10:42] Yes.

[00:10:42] Still both fantastic questions.

[00:10:44] I'll address the first one more so around like where we are in terms of connecting with the entire ecosystem and just making sure that everyone has access and what how do we invest in kind of what that looks like.

[00:10:58] So for me, working within the NVIDIA inception program, one of our benefits to all of our inception partners is that they have access to our VCA, which is our venture capital alliance.

[00:11:10] And so this is a program for VCs to get access to the innovative startups that NVIDIA is working with.

[00:11:16] So both again, both of these programs are completely free to join an opportunity for NVIDIA to help connect the ecosystem.

[00:11:23] What most people have heard about is our Inventures program, which is another arm to NVIDIA outside of the inception program where they're investing in startups again, not just specific to healthcare, but the broader ecosystem as well.

[00:11:39] And so we work really closely with that team and they have a very much an interest in the inception companies that come through and the ones that we're really excited about working with.

[00:11:48] And then of course, NVIDIA has its traditional corporate development arm that does its own investment as well focused on the strategic initiatives for the company.

[00:11:58] When as it relates to just partnerships beyond the North and Latin America, of course, I have counterparts in AMIA that are based in Belgium and all over Europe and as well as counterparts in China and in the APAC regions.

[00:12:13] And so we all have very similar programming all rolling up within our global healthcare team.

[00:12:18] And so we make sure that what we're doing is very specific to the regions that we're supporting.

[00:12:23] So for instance, I have a lot of startups that are focused in the tech bio space, but there are also a lot of tech bio startups in the European region.

[00:12:32] And so my counterpart Cedric is a fantastic job of engaging that ecosystem as well as my counterpart Ting Jing and China as well.

[00:12:41] And so we make sure that the healthcare technology that NVIDIA is in the tools that NVIDIA is unveiling are available across our entire ecosystem.

[00:12:52] Partners can get access to it.

[00:12:54] We work really closely with our men in the medical imaging space for our MONI.

[00:13:01] So MONI, the Medical Open Network for AI was co-developed and is a project for both NVIDIA and alongside King's College London as well as additional partners.

[00:13:13] And that project is community based to create open source models for the medical imaging community.

[00:13:18] And so we work across all different industries.

[00:13:21] Again, it's not all different.

[00:13:22] It's all different sub verticals within the healthcare space and our partners to expand the entire globe.

[00:13:29] Do you work together a lot with the other counterparts because the focus of faces of digital health is how different healthcare systems adopt technologies?

[00:13:38] What are the cultural specifics, the regulations specifics?

[00:13:41] And my natural thought is to which extent do you exchange information or just experience that you have with different markets?

[00:13:51] And how do you compare them?

[00:13:53] Europe is known to be much more regulated.

[00:13:57] China copies everything.

[00:13:59] The US is a huge market and therefore very potential, but also very competitive.

[00:14:04] So how, yeah, what kind of discussions do you have and to which extent do startups overlap between you three?

[00:14:12] Because oftentimes startups from different continents have ambitions for other markets as well.

[00:14:18] Absolutely.

[00:14:19] And of course with the US being a large market, especially with our regulations from the FDA and software as a medical device and getting FDA clearance for companies.

[00:14:29] So I would say that one of the other ways that we work a lot with our inception partners is by key activations at conferences.

[00:14:37] So for instance, we have the Health US Conference that's coming up this year in October.

[00:14:44] And so NVIDIA will have a presence there.

[00:14:46] And so our opportunity to connect with the startups in this ecosystem, but we realize that those are not just going to be US based companies.

[00:14:54] So working really closely with my counterparts across the globe to make sure that those startups are welcomed as part of the ecosystem.

[00:15:02] And we know what announcements they have in ways that we can support them.

[00:15:06] The Health Europe Conference was earlier this year, as I think about even medical imaging has had a presence in the AI and technology space for the longest really in healthcare.

[00:15:17] And so our RSNA conference is coming up in November and that has a really large global health tech company presence.

[00:15:25] And so just making sure that we're engaging our ecosystem and they're aware of the latest tools and technologies and announcements.

[00:15:33] I always encourage them to connect with us and whichever one of our team is there and that's present.

[00:15:38] We love to interact.

[00:15:40] And so we're always saying, reach out to us if you're going to be there, set up some time.

[00:15:44] And again, it doesn't matter what region you're from at that point, whoever is there, we'll connect with them.

[00:15:50] Make sure they get the latest and greatest announcements from us so they know how to leverage it.

[00:15:55] And so that when they're able to go back to their home countries and regions, they're able to connect.

[00:16:00] And I've helped my counterpart have their technology up to speed.

[00:16:05] And so they're able to just keep moving forward.

[00:16:07] So startups have the greatest innovation and the opportunities to be able to carry that out.

[00:16:12] And they move quickly.

[00:16:14] And so we don't want to hinder that in any capacity and we're trying to enable that with our tools and technology as well as our outreach.

[00:16:22] We mentioned the areas that you cover from imaging to genomics data to drug discovery.

[00:16:31] Can you share any examples of specific AI models or algorithms developed by NVIDIA or on NVIDIA's platform

[00:16:40] that you find more most significant and having most impact on patient outcomes or diagnostic accuracy?

[00:16:49] Which one stands out to you?

[00:16:51] Yeah. So I'll take it less from an algorithm standpoint and more like our partners, like some of our notable partners, just that we've announced because I think that's important to showcase when you think about what NVIDIA is doing, not just at the technology level from like an AI and a model standpoint.

[00:17:10] But for the broader ecosystem and then of course, how does that affect patient outcomes and how does that overall just affect the quality of life that people have within the healthcare system?

[00:17:20] And so when I think about that, my first thought always goes to the world record.

[00:17:24] Do you know who holds the world record for genome sequencing?

[00:17:29] No. Illumina? No.

[00:17:30] No, not quite. So in 2021 NVIDIA along with one of our partners with former inception alumni, Oxford Nanoboard Technologies and researchers at Stanford, Google, Baylor College of Medicine and the University of California at Santa Cruz set the genomics world record for genome sequencing in under six hours.

[00:17:54] So that one always gets me really excited because I think about how close we're getting and how close we're going to become to just personalized medicine and what that ability to do genome sequencing, what used to take weeks into how to be able to get that within days is such an amazing thing.

[00:18:13] But I think about last year at our GTC conference, our partnership with Medtronic.

[00:18:21] So we announced with Medtronic and NVIDIA how they're going to integrate our technology into their GI Genius endoscopy module.

[00:18:31] And their GI Genius is the first FDA cleared AI assistant colonoscopy tool and it helps physicians to detect polyps that can lead to colorectal cancer.

[00:18:44] And so leveraging NVIDIA's Holoscan platform and the NVIDIA IGX, our Holoscan is our real time AI computing software platform.

[00:18:54] And then our IGX is our industrial grade AI hardware platform.

[00:18:59] So the both of them combined will help the GI Genius endoscopy colonoscopy system with AI enhanced diagnostic images.

[00:19:08] So that one's also really cool.

[00:19:10] And then I think about J&J Medtek.

[00:19:14] That was an announcement that we made this year at GTC, GTC is 24 because J&J Medtek is actually an 80% of the world.

[00:19:24] It's operating room. So not just the U.S., but the world.

[00:19:27] And they train more than 140,000 healthcare professionals every year through their education programs.

[00:19:34] So our collaboration with that J&J Medtek allows for NVIDIA and J&J to work together to connect the digital ecosystem for surgery.

[00:19:45] So this is going to help with the delivery of real time insights at scale for to support medical professionals before, after, during surgeries.

[00:19:56] The thing I forgot to mention this earlier when we think about healthcare and the opportunities for healthcare, digital surgery,

[00:20:04] digital biology and digital health are the big areas for generative AI and that NVIDIA can really play a role in.

[00:20:11] So you can see those with our announcements with Medtronic and J&J Medtek.

[00:20:15] And also, as I mentioned earlier in the first question you asked me about where NVIDIA has been.

[00:20:22] And so that conversation that Jensen had with the recursion CEO and where we are in the opportunity for the pharmaceutical industry.

[00:20:30] Recursion actually has one of the largest supercomputers based in Salt Lake City in Utah.

[00:20:37] They're a BioHive 2 supercomputer where they built their foundational models for drug discovery, one of them being called Phenom.

[00:20:46] And so their Phenom Beta is the first third party hosted model on NVIDIA's BioNemo for drug discovery.

[00:20:55] And so I think about all of these different partnerships and as I think about digital health and where we're going in this area too that

[00:21:03] the sky's the limit for partnerships and just opportunities across digital biology and digital health and digital surgery.

[00:21:10] And how do you assess who you're going to go in partnership with or which startups you're going to work with because it's really difficult to

[00:21:24] have a good understanding of what a specific company does until you start the collaboration,

[00:21:31] until you already signed the partnership and NDA and actually tried to work together.

[00:21:36] And then many things turn out to be not as shiny as you thought they were.

[00:21:41] So how do you do that?

[00:21:42] How do you do those assessments and set the companies apart or their offerings because even for drug discovery and working in

[00:21:52] silico, but today there's quite a few companies that are working on it.

[00:21:58] Yeah. So what's your approach to filtering?

[00:22:03] Yeah, you got it all.

[00:22:05] You said it all there.

[00:22:06] You answered the question for me, which is we work with these companies again within our inception program to really make sure they have access to the tools that NVIDIA has developed.

[00:22:17] But we can't know how those tools are being developed until they try them out.

[00:22:20] And there's a feedback loop where companies are able to test out tools in early access or direct access and get the opportunity to test out these models and to provide feedback.

[00:22:33] And I don't think people realize that feedback is taken and given back into our teams to develop better models and to develop better products before it's generally available.

[00:22:44] And when we're working from a startup perspective with companies who are agile and able to take feedback and say, hey, we took the technology that you demonstrated.

[00:22:55] And we'd love to be able to show you a demo of what we've been able to do with it and how we've been able to apply it.

[00:23:02] And so I think that's the main area is how can we get that feedback and create that feedback loop to be able to say, hey, this is a company that is innovative like NVIDIA that is willing to try things out and provide feedback and to make better products to ultimately get to the market to create better patient success.

[00:23:25] And so we're looking at companies all the time and assessing companies all the time, using our technology, seeing how we can better support them and using that feedback to develop better tools.

[00:23:39] Can you think of an example of what surprised you most from those feedbacks that you're getting?

[00:23:47] Because I also, I run a user group of clinicians in the UK for medication management and I just love those groups because they give you an insight into the daily clinical reality.

[00:24:02] That's a lot of different from the announcements of what technology can do.

[00:24:07] You always see where the edge cases are, where the problems are.

[00:24:11] And I'm trying to think of an example that I thought was pretty fascinating, but the only one that I can think of, for example, is not in healthcare.

[00:24:20] So I was just reading how some of the states in the US are trying to limit the use of mobile phones in schools, which I think is great.

[00:24:29] You're not deterred by smartphones during your class.

[00:24:32] And then one of my friends said, you know what?

[00:24:34] They tried to do that at one of the schools here as well.

[00:24:38] But then kids just took some old phones and put them in the lockers where they were supposed to put their phones.

[00:24:45] They actually carried their real phones with them.

[00:24:47] And I thought that's actually really fascinating.

[00:24:50] And we see a lot of that, a lot of those kinds of workarounds or edge cases in healthcare.

[00:24:56] They can always give you an idea on how can you improve your own product and what are the things that you still haven't thought about.

[00:25:05] So is there anything that you can remember from those kinds of thoughts?

[00:25:10] I feel like those things happen so frequently, but the one that's standing out the most clearly right now is we announced the NVIDIA Inference Microservices, our NIMS earlier this year in March.

[00:25:20] And so I was on with the startup because we were announcing our Lama III NIMS.

[00:25:28] And so we were wanting to make sure that there were some healthcare applications for that.

[00:25:34] And of course, this is where we began to see this come up really frequently in the digital health space.

[00:25:39] And so one of our partners, MendoLai, one of our inception partners, were like, hey, we love large language models.

[00:25:47] We want to test this out.

[00:25:48] We want to give it a try.

[00:25:49] And they were able to get down their deployment time.

[00:25:52] It was so crazy.

[00:25:54] They increased their ability to deploy and efficiency by 75%.

[00:25:59] And we were like, wow, that's amazing.

[00:26:02] We knew it was going to be fast, but getting the actual feedback from our partners and really hearing them say, hey, you just saved me so much time from eight hours of deploying to one hour is now this is really changing the people's ability to deploy AI models,

[00:26:18] which can be difficult.

[00:26:19] And especially as we, again, thinking about what Jensen says, everyone being able to have this ability and our NIMS are easily available on AI.NVIDIA.com where you can do it on a smartphone.

[00:26:32] So like kids can do it in the classroom and just deploy AI models.

[00:26:36] That is a phenomenal, again, a revolutionary thing for healthcare.

[00:26:40] And just for vocabulary sake and clarity, can you briefly explain what a NIM is if anyone hasn't heard that term yet?

[00:26:49] Yeah.

[00:26:49] So I like to phrase NIMS or our NVIDIA inference microservices as a new layer within NVIDIA's overall computing stack.

[00:26:59] And so it's the ability to package an AI model and deploy it.

[00:27:03] And so it's a container that you can easily download, have access to standard APIs.

[00:27:10] And so it just enables the AI deployment much smoother and easier for developers to be able to deploy AI models.

[00:27:18] Thanks for that explanation.

[00:27:20] I think it's going to be easier for many people to understand the previous answer.

[00:27:24] We talked a lot about working with companies, working with startups.

[00:27:30] But what about the actual end users, clinicians, healthcare institutions?

[00:27:35] What kind of collaboration do you have with them or how do you get in touch with them to get some of that feedback as well,

[00:27:41] which is crucial for understanding where the opportunities and potential is?

[00:27:47] Yeah, absolutely.

[00:27:48] Like I said, healthcare at NVIDIA is all about our entire ecosystem.

[00:27:52] Again, I'm focused on one tiny portion, one tiny but large portion in the startup space.

[00:27:58] And I think this question goes to where we're really focusing our efforts these days.

[00:28:03] And even if again, if anyone's able to or has been able to see our GTC24 announcement,

[00:28:09] you would see in the keynote from Jensen, him interacting with the digital human,

[00:28:16] their digital avatar, so to speak, with one of our startups, Hippocratic AI.

[00:28:21] And so we made a big announcement with Hippocratic at GTC24 because that's the problem

[00:28:28] that Hippocratic is solving.

[00:28:31] They're using the NVIDIA technology to help with conversational AI and just to be able

[00:28:37] to help with conversational speech for patients and to have empathetic avatars

[00:28:43] to interact with, to help with clinical conversations.

[00:28:47] And Hippocratic is one of our startups and then also a bridge.

[00:28:51] So a bridge is another one of our companies that's building AI powered clinical conversations

[00:28:58] when they can have the ability to generate notes and that ability has been able

[00:29:03] to save clinicians up to three hours a day because they can help you take a conversation

[00:29:10] that a nurse or physician is having in the room and denoise it,

[00:29:14] identify the language that's needed for a soap note and make sure

[00:29:18] that patients have the after visit documentation that was captured

[00:29:23] in that clinical conversation.

[00:29:26] So when I think about how we collaborate, I of course defaulted those two,

[00:29:33] of course, as the big ones that I can think about Hippocratic and a bridge

[00:29:37] within our larger ecosystem.

[00:29:39] NVIDIA does collaborate with all of our academic medical research centers

[00:29:44] and research institutions and medical device companies as well,

[00:29:48] because again, it's a broader ecosystem within the inception program.

[00:29:52] We have what's called the Inception Alliance for Healthcare where we work

[00:29:56] specifically with our partners in these other organizations like health systems

[00:30:01] and pharma companies and we partner with them to get them access

[00:30:06] to some of these companies like a bridge and Hippocratic who are doing

[00:30:10] amazing work in the digital health space.

[00:30:13] And how do you envision the future of digital health and your company's role in it?

[00:30:21] I don't know if you ever think of that.

[00:30:23] It's potentially an ungrateful question.

[00:30:26] If I had a digital ball,

[00:30:30] I would say that

[00:30:35] where the challenge was with just health care in general,

[00:30:40] was how do you incorporate technology?

[00:30:43] The advent of APIs and microservices makes it really easy for startups

[00:30:50] to build solutions and NVIDIA is helping to accelerate

[00:30:56] and scale that ability by being able to call on these APIs in literally five seconds.

[00:31:02] We recently announced our have been talking a lot about NIMS,

[00:31:07] but we announced NAVS last week or our NIM agent blueprints,

[00:31:12] which takes the workflows and makes it even easier to deploy AI models.

[00:31:18] And so I just think about where NVIDIA is positioning itself right now

[00:31:22] of just making sure that startups have this access to build these innovative

[00:31:27] solutions so that companies and health systems can integrate them

[00:31:32] to really improve patient care, enhance patient and clinician interactions

[00:31:39] and just address some of the operational inefficiency.

[00:31:43] There are so many ways that we are going to impact the future

[00:31:48] and we're doing it now.

[00:31:50] And all we're going to do is continue to improve it,

[00:31:53] approve upon it and get better and reiterate.

[00:31:55] Like it's a technology cycle that you build something

[00:31:57] and it's out there for the world to view.

[00:32:00] It comes back and you continue to make it better.

[00:32:03] And so our goal is again, to make health care better for everyone.

[00:32:08] I was just thinking how we really mentioned quite technical details

[00:32:15] around the technology and technology development.

[00:32:17] You are a pharmacist background.

[00:32:19] You worked as a pharmacist in the past.

[00:32:22] So I have two questions related to that.

[00:32:26] One is given that you work as well in the drug discovery space.

[00:32:31] NVIDIA works in that space with technology and how companies use it.

[00:32:36] You must have a specific affinity to that, I assume.

[00:32:40] So how do you look at that sector?

[00:32:42] Is there anything that you would want to mention around that area?

[00:32:47] It's funny you say that because I do have a natural affinity for drug discovery.

[00:32:52] But what I've learned is that all of the industries that we focus on

[00:32:56] are subverticles within health care, all have their part in patient care.

[00:33:01] And so as a pharmacist, I made an oath to make to help make patients better.

[00:33:05] And so not just from a medication standpoint,

[00:33:08] which I think people automatically assume that as a pharmacist,

[00:33:11] my goal is to give you a medication to make you better.

[00:33:14] But ultimately I signed up to save people's lives

[00:33:17] and you can do that in a myriad of different ways.

[00:33:20] And you need imaging to be able to look at x-rays and CT scans

[00:33:25] and ultrasounds to make sure that you have a proper diagnosis.

[00:33:29] When you have a proper diagnosis, then you can find the right drug treatment.

[00:33:34] With the right drug treatment, you can get the right patient,

[00:33:38] the right dose at the right time.

[00:33:40] You can do that even better with genomic sequencing.

[00:33:43] But ultimately you need to have someone that can communicate that.

[00:33:46] So NVIDIA, again, from a digital health standpoint,

[00:33:50] using a digital human avatar that can communicate this information

[00:33:54] to someone in very plain terms in a way that is empathetic

[00:33:58] because we have so much of a shortage and clinician burnout,

[00:34:03] the ability to do all of that.

[00:34:06] And using, of course, we're going to be in a place where multimodal

[00:34:10] models need to be built to do all of it.

[00:34:12] It takes every single sector that we're working in

[00:34:15] to be able to think about how do we improve patient lives as a whole

[00:34:20] through medical imaging, through medical devices,

[00:34:22] through drug discovery, genomics and digital health.

[00:34:25] So it's not one single area.

[00:34:28] And even with my natural affinity to drug discovery,

[00:34:31] I've learned that it takes every single subvertical

[00:34:34] to really look at a patient holistically

[00:34:37] to be able to provide and improve a patient's quality of life.

[00:34:40] And then the second question that I had in that sense was,

[00:34:45] as you mentioned, just giving patients the best care

[00:34:49] that they can receive is really important.

[00:34:52] So is there anything that you you worry about?

[00:34:56] Or is a big challenge in the future development of AI

[00:35:00] and healthcare? Because one of the big questions or topics

[00:35:04] is definitely the diversity in clinical trials,

[00:35:08] inclusion of diverse populations.

[00:35:10] I absolutely love state news reports and investigations

[00:35:14] that they do in the use of AI and technology.

[00:35:16] And just recently, they published a report about clinical tools

[00:35:20] that use rays to steer care.

[00:35:23] And there's a lot of changes that will happen here

[00:35:25] because of the differences in the way that patients are treated

[00:35:29] based on the rays.

[00:35:31] Yeah, when we talk about advancements in technology,

[00:35:34] it's always useful to also think about what are the dangers

[00:35:38] so we can mitigate them already during the development.

[00:35:42] Yeah, that's a really good question.

[00:35:45] And as you think about it and of course being that I

[00:35:50] actually have my degree in clinical research,

[00:35:52] one of the biggest reasons I went into clinical research

[00:35:55] was there weren't enough black populations in clinical trials.

[00:36:01] And so that was my personal reason for wanting to enter

[00:36:05] into this industry and then even broader coming into wanting

[00:36:09] to be a pharmacist and now being able to look at how

[00:36:13] health care and technology can play a role in that.

[00:36:16] Right. And so at NVIDIA, we have what's called our NVIDIA

[00:36:21] trustworthy AI.

[00:36:22] I have a good friend on that team, Michael Boone,

[00:36:25] who probably puts this the best and I think it will make sense to you.

[00:36:29] If you think about it, especially the FDA in the U.S.

[00:36:32] and even in Europe with the C mark or just across the world,

[00:36:36] you think about food labels.

[00:36:37] We have food labels that tells us what's in our food and how what we should eat.

[00:36:42] And we can look at it and say, oh, that's healthy or that's not going to be healthy.

[00:36:45] That's got too much sugar in it.

[00:36:46] And so these food labels give us the transparency to help us make informed

[00:36:50] decisions about what we eat.

[00:36:52] When you think about medications, the same thing is like in order

[00:36:55] to take a medication, you know, when to take it and how to take it,

[00:36:58] how frequently to take it, what's in it and their side effects

[00:37:02] and contraindications.

[00:37:04] And so at NVIDIA, we have a trustworthy AI team that's doing the same thing

[00:37:09] for like model cards.

[00:37:12] So we've been building model cards that help with transparency.

[00:37:16] And I thought that was the greatest way to explain it is like when we think

[00:37:19] about food and medication labels, we've got we think about the transparency

[00:37:23] of being able to make informed decisions.

[00:37:25] And so that same thing works about our AI models.

[00:37:28] And so the tools that NVIDIA is building and the platforms

[00:37:32] that we're building upon, our developers need to know and be able

[00:37:35] to make informed decisions.

[00:37:37] So model cards for definition's sake is a short document that provides

[00:37:42] key information at a glance about our ML models.

[00:37:45] And so this will hopefully as we think about transparency

[00:37:50] and like the standards and regulations that come with being in health care

[00:37:55] that you can find our model cards as a way to learn more about our

[00:37:59] ML models to be able to help reduce some of those biases that can come

[00:38:03] into place and some of the questions about ethical AI in health care

[00:38:08] that come up when you can say, hey, look, here's what's in this model.

[00:38:12] Here's how it was built.

[00:38:13] Here's the data when this is again key information about and for

[00:38:18] transparency purposes to help make informed decisions.

[00:38:22] You've been listening to Faces of Digital Health, a proud member

[00:38:26] of the Health Podcast Network.

[00:38:27] If you enjoyed the show, do leave a rating or a review wherever

[00:38:32] you get your podcast, subscribe to the show or follow us on LinkedIn.

[00:38:37] Additionally, check out our newsletter.

[00:38:40] You can find it at FODH.substack.com.

[00:38:44] That's FODH.substack.com.

[00:38:47] Stay tuned.