Can AI Read Your Heart? Unboxing DeepVessel with Nathan Qin from Keya Medical | #52

Good morning, good afternoon, good evening, depending on where you are in the world. Welcome to episode number 52 of the Unboxing AI webinar series. If you don't know what this is, well, every Thursday, 11:00 AM Eastern time, you will generally see my face talking to some of our friends, partners, collaborators here at Kaapil. And if you don't know what Kaapil is, well, Uh, you know, you've probably heard me— if you don't know what Kaaple is, you've not heard me say this, but for the others, uh, you've heard me say this many times, uh, Kaaple is your friendly neighborhood imaging AI platform. Uh, we help you navigate the relatively complex ecosystem, uh, of AI, uh, solutions out there. Absolutely amazing AI solutions, but how do they all come into your clinical practice? Um, we are a platform that allows radiologists to build, test, deploy, and monitor any kind of clinical radiology AI solution. You can choose from our marketplace of almost 175+ apps, or you can build your own as well. All gets supplied to you through a single user interface, single data channel, and single procurement channel. So seamless self-promotion aside, today I welcome our dear friend and collaborator Nathan Quinn. Thank you, Nathan, for joining in today. It's really an honor and truly an honor to get to know you more in detail beyond the bio and the LinkedIn.

Yes.

So for those of you who don't know Nathan, you know, he's very humble in the bio that he sent to me. So I'm going to modify it a little bit. So Nathan grew up in Germany. Started working at the age of 18, started a business at the age of 18, essentially served as a business translator, mediating deals between China and Europe. Pretty phenomenal, Nathan, like it is very inspirational to hear that and got to know a lot of people in that process, built out another business on selling knives, kitchen knives. Audio, I've got a message that says no audio. Are others, I'm assuming, able to hear?

I can hear. Yeah.

Yeah. So BS, maybe some issue at your end. Apologize. But yeah, so Nathan started selling kitchen knives, and then went into the world of healthcare by selling surgical knives where he, you know, designed scalpels and other surgical products for J&J and many other medtech companies. And fast forward to today, you know, he's the founder CEO of Kinovative. Kinovative is, you know, you can think of them as an investment company, you can think of them as a product development company, and you can think of them as a reseller of medical solutions. And within our world of, you know, AI, you know, he's decided to spend considerable amount in bandwidth on cardiac AI, essentially, you know, taking patients all the way from angina to discharge, and everything in between. So, Nathan, thank you so much for joining. It's truly an honor. My pleasure.

Yeah.

Tell us about being a business translator.

Business translator, okay.

What was that like? What does that mean? Like, you know, first time I'm ever meeting somebody who's done that. So yeah, I meet people who run AI companies a lot, but yeah, tell us what that was about.

I think it's as what we call ourselves, a third culture kid, like growing up in multiple cultures. That we naturally learned how to speak— I learned how to speak German, English, and Chinese. And being a translator is not only means that you're translating the message from one end, A to B, but also translating the culture, the thought process. And that's how I learned how to do business, because at a very young age I've been translating for successful businessmen who try to do business between Europe and China. And with that, I have been functioning as a bridge person, which I still believe I still do today, right? A bridge person that's helping culture worlds connecting. So now we are at Genovative, we are actually a really global company across the world, that bridging US, North America, South America, Middle East, Europe. Obviously China, that helping people doing business and helping the connecting the clinicians, researchers to get together. And so I believe, strongly believe that together we will be able to do more and do better and able to help especially in healthcare, right? Bringing the knowledge across the world together to do better things. Yeah.

So, so again, you know, before before the webinar, we were chatting.

Yes.

You mentioned that, you know, from an investment portfolio perspective, like your asset under management is almost $200 million.

Yes.

And, you know, you're an investor in like Kia Medical, one of the largest out there. We'll talk about that. Cassis, of course, you know, we know them really, really well. And Bata.io as well. How do you decide which company to partner with?

Yeah, so as an engineer turned businessman, and I naturally look at technology first, and especially healthcare, I think it must work. Feasibility is number one. And once a startup company proves itself that the technology works, I think that's a good starting point to look into this business, right? Once you get healthier, once you can solve certain problems and fulfill certain unmet needs of the user or of the patient, there's value to capture. And so even though that as a technologist by background, I understand that it's naturally not easy to turn into a salesman. Process. And then I have experience in my past through my experience in the hard way. And then so I'm here to helping those technologies, I think people out there to turn into where science meets business. That's our slogan at Genovative. So allowing the two worlds collect and helping them to do business, that's one of our mission. And with that, I think it's also how we invest in people. So really, really talented people with good technology but lack of market access. That's how we can add value. And naturally, the first step is actually we never invest first and then try to sell the product. Actually, we start, we always say, okay, let's start together. Let's talk about reselling first. Let's talk about sales first. Let's have a good product. If our team can get it sold to the 1,000, more than 1,000 hospitals out there in the world, then we decide to invest in you.

Ah, nice.

So it's a reverse.

If I am, let's say, if I am, uh, a founder of a very early stage startup company, uh, let's say building, um, I don't know, an echocardiogram.

Yes.

Or, or let's say, you know, building something that doesn't exist, a cath lab AI, right? Yeah, like you know, something that helps me actually do the angioplasty. When do I reach out to you? Do I reach out to you only when I have a product that can be sold? Or is there anything that Kinovative can provide to me? Let's say, like, I'm a 4-people team out of, you know, yes, yes, out of MIT, let's say.

Yes.

You know, and I know that, you know, here's a guy who's building like this entire cardiac AI ecosystem. So when do I reach out to you? How does that process work?

Well, one of the greatest examples is our colleagues at Heart.io, right? Heart.io is a really talented team with Otkar and Michael and Adam. They are like 3 PhDs, like a really young team, really talented. And then they started from Pittsburgh in the US. And when they reached out to me, I think I missed 3 of their emails at the beginning. I didn't know, like, because I got so many emails. But then I look at the 4th email and I realized, oh, That's great. It's an AI EKG software that actually able— it almost before FDA right now, and they're raising their A round. And at that point, when I know them, I realize, oh, because this technology has a future, right? If this works with FDA, then I will able to sell it to so many cardiologists or customer out there. But it has value. So what I mean is like, it doesn't necessarily need to be completely done with the product, but the product needs to show certain feasibility that I can test. In the heart I/O scale, I can, I can, I send them some EKG data and some of our expertise and KOL doctors send them some data and then see it could be validated to a certain degree of feasibility. And then it means, okay, the rest is just go down the pipeline, right? Getting FDA, getting more valuable data and improve it. There's constant improvement of the technology itself as well. But the core technology, core idea must work. And then that's a good point to reach out to me because we are investing most of the time between Andrews round, seed round, and A round maximum. Yeah.

Ah, got it. So that's very insightful. So the key thing first and foremost for all the salespeople who've joined in, at least 7 is the cold email limit that we have set. You don't give up. Before, before 7 emails. So here is case in point, right? So it's my team or anybody else in sales who's joined, uh, you don't give up with them before, before at least 7 emails.

Yeah.

Uh, so, but, um, so that's, that's very good to know, Nathan, that there are companies like yours out there that are taking bets on care areas, right? Because a normal VC fund doesn't go to that level, right? Like, they will have a thesis on, say, healthcare, whereas you have a thesis on cardiovascular disease, right? So I think that is super helpful. Uh, and for everybody who's joined, please ask questions. The whole reason we do this as a webinar and not as a podcast is so that you can ask questions. So use the Q&A box, email me, WhatsApp me, whatever you'd like to do, you know, uh, go for it. Uh, So one of the key tools in your portfolio, Nathan, that gets us most excited is DeepVessel, right? Correct. IKEA, FDA-approved FFRC-T. And it is no secret that HeartFlow has dominated that industry, right? It's a company we all look up to, we respect them a lot. Now, uh, tell us about DeepVessel. Like, how does DeepVessel compare to HeartFlow if we consider HeartFlow to be the gold standard? And, uh, what's the value?

Yes, um, I think HeartFlow has been— did a really great job to educate the market and bringing the technology, uh, to the users. And, uh, it had to do a lot of what we say primary work to get the market understand the value of digital software that can actually solving something that actually originally really invasive, which is FFR invasive procedure in the cath lab. Kea's deep vessel FFR CT and is providing actually the same get the results, FFR value. but using another, uh, technology, um, route. So, uh, I think HardFLOWS, I think most of the people know out there knows that this is a CFD technology base, which is based on a strong complex mathematical calculations. Uh, it's based on supercomputing to get to the FR value. In our case, we leveraging our strengths, which is machine learning and especially deep learning technology. And with deep learning, it's not only about the segmentation part, but actually the calculation simulation part of the FFR value. And whenever we have a CT scan, we first create a 3D model out of CT scan and using CT 3D model run the FFR simulation. And then with that, we can get at any point of the coronary tree, we can get to the FFR value.

And yeah, so not, not, so it's not specific to, uh, to a, to a blockage or to an area or something like that.

So our model is a true anatomy of the patient, first of all. And the second of all is also at any point of the model, you can receive and capture the FFR value. That's the power of deep learning. Second of all, our processing power is really, really less. We take it as a joke that say, hey, any gaming computer out there can run this algorithm. Oh, wow.

It runs on-prem. Does it run on-prem?

Yes, we can do. So that's one differentiator. What we do is like we can do on-premise. Yeah. Wow. Yeah. So especially for when we talk about data privacy, like when sensitive data in the clinics, hospital, like to use our, our, our service because we can just install on their server. And they can execute it on their terms, right? Which is one of the big differentiators, I think, on the third part of what we say, because we are really thankful that Heartfelt did all the market education, but it benefited us a lot because users understand the value of our service. Then on top of that, we have a higher pricing advantage. Because our less processing power, so we are more cost efficient in an overall health economy perspective. You're muted, Vidur.

There's also the whole play of being able to very simply not have to send data to the cloud. I think a lot of CIOs are worried about that. A lot of countries like Singapore, for example, Australia, very reluctant to send data to the cloud. So I think it's pretty, you know, incredible what you've been able to do.

Yeah, thank you. Yeah, so we have a strong collaboration with AWS on the international market. So whenever at each country, whenever a customer demand, okay, we want a local cloud, we can just set it up right away. And then if they have their own like personal cloud or local cloud, we can also do that as well.

Nice, that is quite amazing. Moving on from that, Nathan, you know, prior to starting the webinar, you showed me the entire vision that you have of from angina to treatment. If you don't mind, can you show that to our viewers? You know, since there's no questions that we've got as yet, I think it might be useful.

Oh yeah, yeah. Sure, give me one second. I think it's, It's pretty common that these days when we like, yeah, so basically one aspect of this is like we have been, or I have been really concentrated on cardiology is where my heart beats, right? To offering more because as one of our new initiatives is called Cardiac Intelligence where we really focusing on CAD diseases, which is the leading cause of death. Globally, and then 9 million lives taken every year, and especially 12 million invasive procedures has been done annually worldwide. So there's a huge need to reducing this number by using non-invasive procedures. And long story short, I will say some key facts that we know that through our clinical guidelines, right, there's over 126 million individuals are suffering from this disease and then 9 million deaths related to CAD each year. Um, and our main diagnosis tool is EKG, lab tests, coronary angiography, CT, MRI, MRI, echo. And the one who is making decisions are the general practitioner, cardiologist, interventional cardiologist, and cardio surgeon. And all our jobs should be in every health system, right? Should be moving this last 10% on the right side towards the other 45%, which we can treat maybe with, uh, medicine or even through lifestyle. And I think the biggest trend right now is definitely— doesn't matter where we are in the world, every health system is going to early diagnosis, early treatment, and early prevention, right? And, uh, from the digital standpoint, I think our health system is still suffering for 3 aspects: lack of access, lack of automation, and lack of data integration. So doctors, doctors have no, uh, overview over what happened on this patient. If a patient goes from Hospital A to Hospital B, right, everything needs to be done, redone completely. Um, so there's a lot of risk involved, a lot of misdiagnosis, uh, involved, uh, and false positive rates. Are quite high in that case, and also diagnosis precision. So at Genovative, or our, as our cardiac business, we dedicate ourselves as a mission to offer the best-in-class intelligence out there, AI intelligence out there, to provide to our cardiac health professionals so they can actually worry less about analysis but more focus on diagnosis and treatment. And, um, so with that, our technology will provide you one-click solution from scan to clinical report, basically.

Yeah, nice. That's, that's quite awesome. I think, uh, Nathan, we've got some questions in. Uh, the first is from, from Shruti from our team. Uh, you know, she asks, uh, in your experience, how do most successful radiology AI companies measure and communicate value? And that is beyond the accuracy metrics.

Beyond the accuracy metrics. Yeah, so I mean, one of the definitely important aspect is, do you get reimbursement? So convincing the system and from the health economics standpoint is the major pain point of all AI business today, I think. In any health system across the world. Convincing is not— we need to only convince clinicians, but also convince the payers to pay for that. And then that is actually a huge hurdle, which meaning life and death for the AI companies, right, in healthcare. And the second of the communicating value is definitely that we are not here to replacing healthcare professional, but here to enhancing our healthcare professional to be more efficient. The lack of access, lack of access integrations comes from lack of doctors and healthcare professionals everywhere. And if we can bridge this through AI, that will be the true value that we bring into the system. The third aspect is definitely the health economics standpoint, that how much can we save it's a saving topic system. For US as an example, US spends more than $3 trillion every year on healthcare spending, right? And this is the whole GDP of Germany.

And with that, US healthcare system is the 4th largest economy in the world.

Correct. But the healthcare is still there to saving money, right? Saving it from structural heart disease and how can we capture them earlier. So if AI can capture them earlier, we're actually saving the system a lot of money.

No, in fact, you know, on that point, Nathan, sorry I'm interjecting. So Dr. Suresh Mukherjee has just asked a question. He says, I believe HeartFlow can be reimbursed through a temporary APC payment in the US. Is this based on a type of analysis or is this product specific? Is your company also able to qualify for this payment in the US?

Yes. So the Medicare generally reimbursing this with a CPT code, I think 75530 or 80. So it's not dedicated to HeartFlow, but also that to anybody out there. So it's not only us, right? We see new competitors coming in and we also know clearly Ellucid that trying to launching on Siemens even trying to launching their FFRCT applications. We have also plaque quantifications, right? We have coronary fat analysis. So CADRA 2.0, all of the stuff that the doctors are looking for, we have a whole pipeline behind that. But, uh, we, we need time to get to FDA slowly, but FDA is the first step. Yeah.

Oh, wonderful. Uh, then we have a couple of more questions. You know, we'll quickly go through them. We have way less time. Uh, one is, uh, you know, from, uh, an anonymous attendee. What are the slowest steps in AI evaluation or procurement in the pipeline today? And then we have the ChatGPT double dash. How are companies overcoming those?

Okay, slowest step in the AI evaluation or procurement pipeline today. Yeah, I think procurement is definitely slow. We're reaching end customers because administrative procedure at any hospital is really long.

And Nathan, you know, honestly, I think the next question is also tied to that from Gaurav. He asks, how has the sales function evolved over the last 10-12 years? I think that's a very good insight that you can provide, right? Like 2012 was when you started and today is 2025. Like how has the sales process of selling, say, medical device technology changed over the past, you know, 10, 15 years?

Yes, last 10, 15 years, definitely from technology standpoint changed a lot, right? 10, 15 years ago, we don't have any AI applications. We're still telling minimal invasive, we go from invasive to minimally invasive treatment, right? Cancer, we cut off the whole gut out and then threw it away. And now we just open up 12mm points and then treat it. That was one thing. And from minimally invasive to early diagnosis to digital treatment, that's kind of where it goes. And, but unfortunately, our system worldwide is still catching up quite slow. So from a sales process perspective, a lot of things didn't change.

Yeah. Yeah.

And, but in other side, we see, we see a policy like value-based care. We see volume-based procurements where it put a lot of pressure on manufacturers and hospitals to be more accurate about what they how they prescribing things, uh, which is actually a good thing because efficiency then is more prioritized than, uh, interpersonal relationship, right?

Yeah, yeah.

Which is a positive factor that has been developed in the last 10 to 15 years. But overall, yes, uh, in term— if you was want to sell a product, get into the clinic, clinics, we say, uh, FDA first, then reimbursement, and then business, right? That's kind of a 3-step to get to the—

which is, you know, which is so difficult, right, Nathan? Because today, to be able to build something that you have so much confidence will be valuable after FDA, after reimbursement. And that's why we see such less number of companies succeeding. And those that succeed, you know, turn into like— Kia is a, you know, you mentioned a $600 million valuation company. So very inspirational.

Yeah. So at this point, investors are really, really important, right? So investors with patience, understanding the healthcare business, it's rare. I would say it's under all the investors out there. I think healthcare investors have the most patience, have the most risk, but also at the end of it, have the most return of investment. Exactly. Yeah. And I think we need more and more investors in that field to supporting our healthcare system, healthcare startup to survive.

Oh, I think that's a great shout out to our investors, right? We are very, very honored to be working with, you know, somebody who trusts a slow-moving industry, but then, you know, once it starts moving, it's, you know, there is no going back. So that's awesome. Final question, Nathan. If I am, again, you know, if I am those 3 people starting an AI radiology or AI cardiology company today, what advice would you give me?

First of all, definitely look at the— okay, obviously, hopefully those 3 people are like really talented AI engineers these days.

And everybody is an AI engineer, Nathan, today. Even I am one.

Yeah, so the world needs really talented engineers, and especially biochemists, bioengineers today, because everything else could be replaced by AI in the next 10 years. Second of all, I think looking at unmet needs, I mentioned before in my slides, like lack of integration, lack of data access. And then lack of, um, communication. And I think this is what the healthcare needs. And doesn't matter, you don't need to focus on cardiology or radiology, but overall our health system going towards early diagnosis, early preventative, and, uh, early care. And those are definitely the future needs, um, and any kind of type of, uh, disease states. And focus on one disease state and dive deep into it because healthcare business still need a lot of expertise. Instead of becoming a generalist, become a specialist.

I think, I think that is a great note to end on. Uh, become a specialist, not a generalist. I think that is true for most, uh, like generalists are now, I think, just commoditized with, with the way AI is. And the same probably we're going to see in our world as well. Uh, so thank you, Nathan. Thank you for doing this, means a lot. Thank you to everybody who joined. Uh, yeah, see you same time next week, same place. Take care.

Yeah, thank you very much. Thank you for the invite. Thank you everyone.

Thank you, Johan, for the comment. Thank you.

See you. Bye.