Scaling Diagnostics: From Innovation to Infrastructure | LSI Asia '25

Adrian Chong  0:05  
Right, good morning everyone, and thanks a lot for being on our panel here. Like to welcome all of our panelists with Peter Fischer, Christina Valgrind and Logan. Sorry, your surname. So a diverse panel here, and I'm Adrian Chung, your host. First, we'll start with some introductions. I'm a practicing cardiologist. I work in Hong Kong, senior managing partner at Virtus Medical Group. Our cardiology grouping has nine cardiologists. I specialize in interventional cardiology, meaning valve repair, valve replacement. Complex interventions, complex PCI. I'm also early stage Medtech investor. Over to you. Peter,

Peter Fischer  0:52  
yeah. Peter Fischer, CEO of InkSpace imaging. We are reshaping the future of MI with our products. I'm a physicist by training. Have done the last 25 years product development, both in the existing silicon industry world, but also now this printed electronics. For the past more than a decade, live in the Bay Area. Currently.

Christina Vallgren  1:18  
Good morning, everyone. So my name is Christina. I'm the CEO and the co founder of Tara Pat. So I'm a physicist by training. So before funding Tara Pat, I have been working as a physicist. That's the European Organization for Research Nuclear Research for the last 12 years. So at terrapet, we're developing nuclear imaging devices targeting different areas in radiotherapy and also nuclear imaging fields. Thank you.

Lohendran Baskaran  1:50  
Good morning, everyone. I'm Logan baskrin. I'm a practicing cardiologist at the National Heart Center Singapore, and also have a degree in Medical Physics. I am the director of our AI research lab here at the National Heart Center Singapore, where we innovate to identify coronary artery disease and work towards preventative medicine. The best IP that we have from our AI research lab, we have spun off into tear health, for which I am a co founder. So this is a startup developing and implementing these products across Southeast Asia and the rest of Asia as well. And I'm the co founder for that.

Adrian Chong  2:30  
Well, thanks a lot for your introduction. So diagnostics is a huge subject, so almost don't know where to start. But you know, here we have, we have improving images. Peter is improving quality of images of MRI images. Christina is trying to making, trying to make therapeutics more precise, more personalized, and also creating a new PET scanner for the whole body. And then Logan is trying to make risk prediction more effective. We'll start with the software as a medical device, you know, AI. So I'm a user of AI myself, apart from, you know, bio world, my company investing in AI companies, you know. And in my clinical practice, we have nine doctors of various ages. The senior partner is 70 something years old, and then he grew up in the age where there was no AI, there was just human thought. And you know, there is the youngest one is in his 30s. And we all have different opinions about AI. And you know, CT, coronary angiogram, CT, FFR, etc. The question is, how do you overcome clinician inertia to integrate AI into into practice? You have this young whipper snapper co pilot that's telling you to do stuff. How, as a clinician, as a mature clinician, are you going to accept that? How do you convince clinicians to do that. So I

Lohendran Baskaran  4:03  
think that's an excellent question. Adrian, you know, AI is software from our point of view and what we do, but software is also still technology. So none of this is new in medical healthcare. Technology always evolves and moves forward, whether it's hardware, software, AI. So when doctors have to adapt to new imaging modalities, for example, CT was a relatively new thing, which wasn't present 10 to 15 years ago, or at least not widespread a couple of decades ago. What was interesting is you need to show some level of trust, you need to show that there is accuracy, and you need to show relatability to the current gold standard. So that's what we need to do, in terms of AI and our product, as well as the research that we do. What we try to show is, hey, we can show you that the products that we have are developed by cloud. Clinicians, so by doctors for doctors, and that gives some level of trust. The people who develop it are key opinion leaders, and they are not Junior young people. They are very established in their field, so to speak, and they have overcome their resistance in AI to help develop this product. We show that this is accurate because it reports and derives information to the same level of seniority or accuracy as any expert level. And so we test these cases, blind doctors who are resistant, or so to speak, you know, a bit wary of AI. They bring their images of their CT scanners, the CT scans of their patients, run it through our algorithm, and then they see the result and compare it to themselves or their junior fellows, and that's how we convince them. Now, in terms of integrating, we never take control from the doctors. We give control back to the doctors, but because what we do is we spend all the time doing the tedious work and getting the information and extracting more information so the doctors paid more attention to their patients. So that's what we try to do. And so far, we've had quite a decent success. And

Adrian Chong  6:08  
how do you prove value for money for things like these? I mean, thing is, when you have a piece of hardware, people are willing to pay for it, you know, if it's a blood test, you feel the needle will go into your skin and the tube of blood coming out. But AI is kind of something floating in the internet. And that concept, you know, like, for example, in mainland China, AI reimbursement is a difficult subject. You know, you're not allowed to pay for AI itself. It has to be integrated the system. So how do you convince people that it's value for money?

Lohendran Baskaran  6:46  
So that's a very good point. Ultimately, we don't think that AI should be charged for AI sake itself, as you said, value for money in terms of healthcare outcomes and things like that, these things are what matters the most. So the first thing is, there is a increasing awareness from both the patient and doctor or clinician point of view, that preventative medicine is far more important. Coronary artery disease is the world's number one killer. It is increasing, especially in this part of the world. It's expected to double by the next in the next 25 years, and also, ultimately, it can be detected using CT scans. It's the only non invasive way and the number one way to detect this disease. So by detecting disease, we need to prove to people that you can track and prevent the heart attack, which comes down the line. I think there is a very clear awareness on that part. The next part is, how can AI or the software help? Well, it helps you increase the amount of patients who can go through and have scans by decreasing the bottleneck by six to 10 times. At least. The more people you can scan, the more people you can detect disease and prevent disease from happening.

Adrian Chong  7:56  
Thank you. The other interesting thing about AI is that it's very much in the public's mind. It's very much in the clinicians mind. Everyone's talking about AI, deep sea, chat, GPT, etc, perplexity. You know, how important is actually patient advocacy and creating a buzz, you know, creating publicity in in increasing the acceptance, or pushing the acceptance of of AI diagnostic modalities into actual daily clinical practice. I mean, I experienced it myself, you know, before I didn't use CT, FFR, but then one of my very, very educated patients who invest in Medtech as well. He's like, you know, I just had the CT scan. I can see this. It looks quite severe. LED stenosis, left anterior descending artery stenosis on the CT scan, but I don't have any symptoms. Can you do CT, FFR in it? He actually asked me for it, and that actually initiated my journey into AI assisted diagnosis there, and in the end, we did the CT, FFR, and it was borderline, and we decided that we'd watch it for the time being. So I mean, what do you think of the role of patient advocacy in education? So I really

Lohendran Baskaran  9:18  
like how you're carrying on this conversation and moderating it. Adrian, it sounds like I'm just discussing things with a colleague in clinic, but essentially the advocacy from the patient point of view isn't always 100% there. So you were, you know, so to speak, fortunate, or you you were in a situation where you had a very well educated and typically reasonably wealthy or of a higher socioeconomic status patient inquiring about these things, and we have that the same as well as where we work, but because we're a government hospital, that's where we work, and we have mainly subsidized patients. It's not always there just the simple awareness of compliance or adherence to medication, primary prevention, smoking cessation. Isn't always there. So I think it has to be very much determined by the clinician or the doctor. That's why, ultimately, AI and software needs to give control to the clinician, who then needs give to give control to the patient. So that's how I think it works. It doesn't change that relationship very much at all. Technology gives control to the clinician. Clinician gets control to the patient, and that's how the partnership is made. So that's what I think about it.

Adrian Chong  10:27  
Thank you very much. So over to hardware now with Peter and Christina. So with hardware diagnostics, here you're looking at, especially your two products, your two companies, you're looking at no working with the incumbents, the big three, imaging companies, etc. How do you form partnerships with them, and how do you actually penetrate their markets

Peter Fischer  10:56  
over it? Peter, okay, so it's all about partnership, right? So I mean, you need to do is the same as it's pretty similar to what you just discussed. There needs to be education, right? No false promises. And you promote your product, you try to show that that's an extension of what they already have, right? So in our case with MI, receive coyotes, that's a well established market. Everyone knows that MRI is a good standard in diagnostic imaging, and so there are hundreds of millions of MRI exams performed every year. So there's an install base of scanner, there's an install base of coils, and clinicians have used this for decades, right and now there is this tiny startup coming out of Berkeley telling everyone that the coil performance is so much better in terms of precision, temporal, spatial resolution and acceleration. So actually, the clinicians can see things much quicker. The they can see more details than they could ever see this anymore and in much all the time, which is also important for patients, right? And this also actually promote so working with CES, you call them incumbents, and let's say also smaller MRI vendors, it's all about partnering, right? Not trying to compete with a multi billion dollar company, but offering solution as an extension of their platform. And what we found is that working with the clinicians, with the radiologists, in our case, it helps a lot if you get support from the clinicians in promoting the product. So that's why we have decided to work with a multitude of key opinion leaders first, and so yes, and then we want to expand to Europe and Asia, and that also helps in the collaboration with the OEMs to see the acceptance in the field.

Adrian Chong  12:49  
Thank you. Christina,

Christina Vallgren  12:51  
so obviously, hardware is hard. It's not only hard to build, it's also very hard to raise money for hardware. So normally, often when we talk to investors, they say, Oh no, are you building a such big hardware? Therapet, therapeut, the pet is the PET scanner, and the Tara is big. So what we do is that we build the biggest PET scanner ever. So I thought about the PET scanner. So when investor says, Okay, it's hard, but obviously, but of course, the second question they have is also that, how do you compete with the big players like Siemens GE to be honest? I mean, everyone knows a startup like us, there's no way that we can compete with the big players. If we go to hospital and even we tell them, we sell the best PET scanners in the world, they would never buy it. It's not because our device is not good enough. It's because of the after sales service, because of the reputation, and also most of the hospitals, they are sometimes, let's say, blackmailed by the big players. If a Siemens Hospital has only Siemens brand, and the GE hospital only has GE brands, obviously. So there's many different reasons that's very hard for the startup to enter the market. So therefore, as a startup, you have to have, not only innovative product, you also need to have innovative business model as well. So I think that competing with the big players is not really smart, so you need to have a complimentary method looking to very niche market where the startup startups can bring values, and also that the startups are not head on competing with the big players, but to complement what the big players not doing and the or maybe not willing to do, or not daring to do, because the companies are just too big to be agile. So that's what I think, that the startups can bring the real value, not only to the patients, but also to the whole ecosystem,

Adrian Chong  14:54  
right? So talking about innovative business models, can you. Explain, elaborate, more

Christina Vallgren  15:01  
elaborate. So just to give an idea of what we are doing, so we are building a PET scanner. Talk about the PET scanner. So obviously, the easiest business model that you have is go to the hospital and sell it to the hospital, right? So it's, it's a unit sales. But what we are doing is that we are actually not going to the hospital to directly sell it to the hospital, but together with the hospital, we partner with them so that the instead of selling to them, so we become a device, infrastructure and service provider. So we help in the hospitals that to help the pharmaceutical company, bio biotech companies to do clinical trials. So that's not a direct competition towards the hospital or towards the big players, because the Siemens or GE they are just too big. They will never even become a service provider for anyone. So they sell so well. So what we do is that together, partner the hospitals, so we help someone really need the clinical trials so already now. So together with Karolinska and another fine national hospitals in Sweden, so we are establishing the European, the first, the European national, wide tyrannostic trial centers. So our goal is that to really partner, to put all the partners together to bring the value to the pharmaceutical companies. So make the clinical trials more centralized, more efficient, and also that we can together to bring the drugs really close to the patients faster. So what we do? We think that we have, not only a note, innovative product. We also have a very innovative business models to help us to bring our product to the market quicker,

Adrian Chong  16:50  
right? So targeting the drug pipeline, creation new drugs, rather than just pure clinical service,

Christina Vallgren  16:57  
exactly. So we look at the whole health care as a whole, the complete, the entire healthcare system. So then we looked at, where is the need, need the most need, need to be placed. Where's

Adrian Chong  17:12  
your niche? Great. So, and for Peter, I wanted to ask, because your your MRI coils are very innovative. They're, you know, confirmational. They're very light. You know, who do you choose to partner with? You know, as a startup company, you can't work with everyone, I mean, or can you work with everyone? Do you go broad and fast? Or do you have a deep partnership with one single company and then you can keep having feedback with them and improve your product until you you get to a certain level of advanced maturity, and then you can start working with other partners. I mean, how do you make that choice? I think

Peter Fischer  17:53  
you have to answer it from from two viewing alts. One is, we want to get this technology out into the market, into the healthcare systems around the world. That's our mission, because we truly think that it delivers better health care, both for the patient, but also for the for the clinicians, basically quicker, more precise diagnostics in a shorter time, and for the patient, obviously less exams. So from, from a roll out perspective, we had defined that we are working with certain key opinion leaders in the field to basically let them help us to demonstrate why the product was so much better and which additional use cases one can do, which I couldn't do before this moi. So we have, let's say, two handful of key opinion leaders in the US. We are working this very closely to get, also publications, white papers, out of them, in terms of of the big players in from an investor perspective, our mission is obviously also to have this company at some point of time, acquired, right so my strategy from early on was we work as this as many OEMs as possible, but obviously, as a startup, your resource limited your constraint, right? So we have quickly decided that we are working with a top three first. So we have already products cleared for GE and Siemens. And after that happened, we get a huge spike in interest from Philips because of the number three. And they they are strongly interested in our career portfolio as well. So as the CEO, unfortunately have to say more often, no to things that I can say yes to no pure projects, since you never have unlimited funds, right? And I totally agree it's from my perspective, what I see much harder to raise money for hardware companies, especially in fields where the innovation before us has happened, like 20 years ago, right? So we are very late innovator on a very well established market. So. Never been easy to raise a ton of money, but so far, the strategy worked out. We have products killed for GE and Siemens, and now they are on the field, and people talk about it at conferences. They publish. And so other hospitals see the value at and some hospitals, as you said, they are, let's say, having only Phillips equipment, only Siemens equipment and some also free to buy directly from start apps, so you have to partner with the CEO, and so that's our strategy. We say directly, but you also sell through the OEMs, and that has worked so far for us, but we are still in an early revenue growth phase.

Adrian Chong  20:37  
Thank you. Now pivoting a little bit to the elephant in the room, geopolitics. You know, it applies to hardware and software. So from the software angle, anything to do with AI is sensitive these days. You know, you have to host it on the cloud system. You know, maybe I don't know, in Singapore, you might be on AWS, but then, you know, if you're trying to go into mainland China, for example, then it has to be a different kettle of fish. How do you see this as an obstacle or opportunity in your global rollout of your products? And how do you, how do you mitigate these things?

Lohendran Baskaran  21:19  
So we see this as a huge opportunity. This wasn't brought about from recent geopolitical events, but actually pure clinician need. So there, whilst there have been competitors from, so to speak, China and the US in some of the space and a little bit of what we do, what is important in the healthcare space is patient data and thus clinician trust. So the reason that a lot of these software solution or AI based companies have difficulty coming into this part of the world is not just the revenue model, but also how the data is handled. A lot of them are cloud based, because that is obviously a good way to save money, but also to have rapid throughput. We developed ours from the ground up by clinicians because of the problems that I faced and my colleagues and key opinion leaders in this part of the world face, we don't want to send data up to the cloud. We don't want to share it everywhere else. We don't want it access by other third parties, be it another startup or another company. And because of that, our products are unique in that they can be integrated at multiple levels. So directly pipe from the scanner, such as the CT scanner, to the software to produce an output within a server, within an institution or hospital, within a server that is amongst many hospitals, for example, in private healthcare groups or on the cloud. So we are agile enough to adapt to all of those and they ultimately self train and use federated learning as well. So each hospital or site can use their model and corrections from the clinician to retrain it so that we can avoid drift, and we can keep it up to date to that cohort, to that population, to that institution, because that varies as well. So that's what we see as an opportunity,

Adrian Chong  23:14  
interesting. So in hardware, obviously, you know, with PET CT scanners, we're talking about rare earths, with the MRI coils and things we're talking about rare earth as well, and maybe microchips. How are we going to mitigate these supply chain issues? Christina,

Christina Vallgren  23:32  
yeah, recently, we just had a everyone knows, because the Tali flu issue with the US, between US and China. So in the beginning of April, I think Chinese government just decided, certainly in that they banned all the exports. I mean, you have to apply for the export license all the exports towards any countries, the rare earth materials. So for us, it's a quite a kind of shock, because most of our our, how do you say, scintillators, the earth, materials that we import. We import them from China. But obviously in our case, specific, in our case, we got a lot of support from the Swiss government. So the therapeutic is a switch based companies. So apparently the relationship between Switzerland and China is very good, so in this in this time tunnel period, so it's really good, so that we got the support from the Swiss government. So even the Swiss government had to go inside and tell, tells the Chinese Ministry of Commerce that to release all the export license applications for the Swiss startups. So that was the kind of advantage that we have in our particular case. Obviously, I mean, I think you need to de risk for the supply chain. It's very important if, especially when. Build the hardware. So you need to really to think about, when you start for different parts equipments, that you need to have several suppliers located in different areas in order to de risk the supply chain issue down the road.

Adrian Chong  25:17  
In summary, be aware of your political surroundings. Be friendly with your local politicians. And be aware of the global political situation, because it can throw a span in your works.

Peter Fischer  25:31  
Yeah, and you have to, I mean, as in every industry, you have to manage your supply chain, right? You need to have second sources which qualify them, carefully monitor them, you have to have your own incoming inspection to make sure that whatever you get delivered is according to a specification. So in our in our case, the the more critical, more IP relevant materials come all. They're all sourced in the US directly, and the more standard electronic components, as everyone we are sourcing them from around the world. So we don't have any rare, fancy materials in our our system, at least not rare us and so fast that's that's more difficult, but also in previous jobs, basically, you have to constantly manage your your supply chain. So they are professionals doing that. As much as we have RF engineers on our team, we need to have someone who is a professional and managing and doing the risk management and all of that, and working with your material providers right, and then expand as and also for the manufacturing, for instance, right? Currently, we do all the manufacturing in house, and I plan to continue doing that for quite some time, and then only outsource standard PCB related components, and then decide where to go, depending on the volume, all of that, and then, obviously, especially being located in zbs. Now you have to be carefully monitoring where you, you know, place your bets, and all the interactions we already had five five years ago, trade wars between the US China made everything more more complicated, but that's true for everyone using electronic components around the world,

Adrian Chong  27:10  
right? Thank you. So in summary, for s AMD, it's, you know, localization. Keep things local and personalized, and know for for hardware, be aware of your political surroundings, and keep an eye on on global politics, and also build resilience into your supply chain, to start with, from the very beginning. So finally, I'm going to ask from, from each of you, just a brief, you know, brief talk about, you know, what you envision in the next five years in your particular area, what would be the most exciting developments in diagnostics? So

Lohendran Baskaran  27:53  
I obviously speaking from a bias point of view, but essentially, implementation and integration of workflows which help doctors do their job the best to identify disease and prevent it before it happens. So that's what I want, and I think it's a realistic goal.

Adrian Chong  28:13  
Thank you, Christina, 

Christina Vallgren  28:15  
as both me and Peter agree that it's very hard to raise funds for hardware companies, obviously. So I would like to tell all the investors down there. So I mean, it's nice to invest in a trend AI. I think a lot of companies like AI companies, they raise tons of the money, because as soon as you are AI labor, so you can raise funds. But the hardware is hard. From the idea to the to the market, the path is extremely long, so, but hardware is very tangible. So it's it's less, less. How do you say influence, by the by the bubbles. I mean, it's more tangible for the for the also for all the risk that we have surrounded. So if Yeah, I hope that more and more investors could look into the opportunities in the hardware, so at least you know what you get. It's not a just in the air. Thank you.

Peter Fischer  29:19  
Yeah, in our case for MRI diagnostic imaging, obviously there's a ton of AI startups also working on image recognition improvements and the image quality, less noisy images, all of that, and but I think the key players on the market have understood that AI is making use of the data which come out of the coils, right? So everyone has the 3t or one and a half tesla scanners. The most crucial component in diagnostic imaging to create high image quality. It's the shortest amount of time at the end of the day, is a coil we all bring to the market. So we all mean where? Collaborating with two AI software companies, which have realized that all the data they're getting, if they get our coil this much better data input, their output, will also be significantly different. So they are assess interest from from two AI software companies, both targeting also image quality and speed. That's exactly what we what we deliver with our coils. And I think the combination of this will be very beneficial for the clinicians, but also, most importantly to me, to the patient. So getting a quicker diagnostics and takes less long, and especially for the kids, less sedation required.

Adrian Chong  30:39  
Thank you, lastly. You know, as a clinician, we come back here again, full circle. You know, I look forward to AI diagnostics helping me to make risk predictions more accurate, images easier on the eye and higher spatial resolution, and therapeutics more personalized. With that, I close the session. Thank you.