Building the Internet of Medicine: How AI Could Transform Children's Healthcare | Dr. Timothy Chou - Founder, Pediatric Moonshot

Speaker A: Sam. Mhm. Um,

Speaker B: Welcome to another episode of Progress, Potential and possibilities discussions with fascinating people designing a better tomorrow for all of us. I'm your host, Ira Pastor. Welcome back everybody again to another episode of the show, bringing an awesome guest for you today who uh, has spent more than four decades, uh, the forefront of nearly every major transformation, uh, in enterprise computing. Um, Dr. Timothy Chow earned his PhD in electrical engineering from the University of Illinois before beginning an amazing career start starting at Tandem Computers, one of Silicon Valley's pioneering companies in fault tolerant distributed computing. Went on to serve as president, uh, of Oracle on Demand, leading their early cloud computing initiatives. Uh, and alongside this amazing industrial career, Dr. Chaos taught at Stanford University, uh, since back in 1982, creating the first university's course on cloud computing, inspiring next generations of entrepreneurs and tech leaders. Today he sits on the board of directors of Teradata, chairs the Internet of Things initiative at Alchemist Accelerator, and advises numerous emerging software companies. But ultimately, after what many would consider a complete career, um, Dr. Chow came out of retirement to pursue perhaps uh, the most ambitious mission yet. Inspired by one of his students at Stanford, he founded what is known as the Pediatric Moonshot, which is a nonprofit with an extremely bold vision to ultimately reduce healthcare inequity, enabling privacy, preserving artificial intelligence to secure, learn from more than 1 million medical machines spread across roughly 500 children's hospitals around the world, uh, without requiring patient data to leave those hospitals. And today we're going to be exploring not just the future of pediatric health care and AI really getting into uh, how medicine in this tense is uh, on the verge of its own Internet moment. Distributed intelligence rather than centralized data, ultimately can reshape how care is delivered all around the world. Um, a lot to get into. Really cool guest. Honored to have him. Um, to Tim Chow, welcome to the show.

Speaker A: Thanks for having me, Ira.

Speaker B: Awesome having you. I, I'm looking forward to jumping into all of this, I mean, a couple of minutes just to further tell this story because clearly you've had an extraordinary career that as I mentioned, spanned fault tolerant computing, expert systems, cloud computing, Internet of Things, now, AI, everything we're going to be talking about in a bit when we get to the moonshot. But um, tell us a little bit more about the story. It's, it's really an awesome journey you've been on and just love to hear a little bit more about you.

Speaker A: Yeah, I, um, as you said in the introduction, you know, as, as was already known, I had retired, was teaching class, um, you know, sitting on boards and whatnot and, uh, you know, one of the things I say at the end of class every. I teach this class only once a year. And, uh, we invite CEOs of the, you know, public companies that are movers and shakers. In fact, this year we've got Jensen Huang showing up. Uh, but anyway, at the end of class, I. I always say to the students, I say, well, you know, um, I don't quote scripture that often, but I said, you know, to those who much is given, much is expected. And I say to him, you know, you've been given a lot. The fact that you're sitting in this classroom is you've been given a lot, so I expect a lot of you. And, um, one day I was sitting around thinking to myself, I've been given a lot, so what should be expected of me? And I think the happy coincidence of meeting Dr. Chang and him, I like to say adopting me and bringing me into the world of healthcare was kind of the catalyst that I said, okay, all this stuff I know people I know, you know, this could be, you know, directed at some very useful purpose that, you know, just to say it, no Stanford kids are going to go after this problem. It's not very commercially interesting. It's really something to be done by somebody who has a, you know, let's say, unique set of situations that I said, hey, that's me. So I decided to come out of retirement and, you know, work on this, which has been enormously satisfying. Met m all sorts of interesting people along the way. And, you know, as you've already said, I. I think we can change the game in healthcare, in particular pediatric healthcare.

Speaker B: And let's, you know, let's drill down a bit on sort of the big picture, uh, first from sort of the children's health, and then we'll get into sort of the issues, uh, in 2026 with sort of where tech is. But when we look at the numbers here, um, There are around 2.3 billion people on this planet that are called children are under the age of 18, uh, 1/3 of the population. Um, there are high, you know, disproportionate amounts of, uh, rare diseases, um, in terms of, you know, something like 70% of them starting in childh, a lot of them get cancer. We think of cancer as a disease of aging, but it's not. Um, and many children, um, just never have the opportunity to reach a children's hospital. I'm sitting here in Philadelphia. We got a really great one around the corner, but there's not a lot of them. Ah, distributed around the United States even, and worldwide, of course to this, um, that, um, I saw a figure that in some countries the ratio of pediatrician to children is something like 1 to 100,000. We really have some interesting figures. And I'd love you to just talk about this part of it first, Tim, because I think, um, many of us may be confused that uh, there's a lot of pediatricians around and the children be taken care of, but there's real imbalance here.

Speaker A: Yeah, I think one of the challenges, you know, obviously we spend a lot of time talking to foundations and pediatrics and whatnot. I think one of the challenges is most kids are healthy, which is good. That's a good news. Right. And so we all don't have any personal experience with a lot of these conditions versus as an adult. I mean, we all know somebody who's had cancer, breast cancer, or we all know somebody, uh, who's had, you know, arthritis or whatever. Uh, and so I think one of the challenges is how do we help people understand. Right. The breadth of the challenge here. So one of the things we did about three months ago is we, uh, inspired by the chief science officer at the V Foundation, she made the comment to me, she says, you know, there's 100 pediatric cancers. And I'm sitting there listening to her thinking, I don't know, 55. I had no clue. Right. Which started a process which now I can tell you with reasonable accuracy. There are 104 pediatric cancers. There's 233 known neurological conditions, there's 105 cardiac diseases or defects, and there's 176 kidney diseases that are known to us now. What's interesting is if you go back in time just 25 years, right. The, the dawn of the millennia, it's 2001. If we had run the same exercise, the answer was there were 106 kidney diseases that we knew about. Right. I keep going. That we know about. Obviously this, you know, work in genetics and you're coming out of biopharma, I mean precision medicine. We are getting more precise in our understanding of these diseases. Okay, good news. Now let's turn it around and go, okay. What does that mean to a pediatric nephrologist? Well, by the way, if you were trained 25 years ago, you're kind of in the middle of your career right now. You were trained on 106. There's now 176. And by the way, that means there's going to be 296 in another 25 years. So even for the specialist, right, the breadth, let's call it, the accuracy of our science is going way beyond the ability, our cognitive ability to understand all this and keep it in our brain. Okay. That's the nature of what we know, uh, in the medical world. Now go to the point you were making. You don't have to go to Africa, which, you know, clearly you could point out the deficiencies in pediatric, uh, specialization in the United States in the rural counties, uh, 86% of the rural counties have no pediatric cardiologist. There's nobody there. Uh, in nephrology, kidney diseases, it's 99%, in oncology it's 95%. So as you correctly point out, right, if you're in Philadelphia, or as I like to say, if your mom works for Google and you live in Palo Alto. Well, what we're working on, yeah, it's somewhat useful, but it's really about the children who are not in those advantaged positions. Right. And just throw you another number out there. There's 3,000 pediatric cardiologists in the United States. I just told you, they all live in the major metros. Right. They live In San Francisco, Louisiana, Philadelphia. If you go to India, there are 300. Uh, and if you go to Rwanda, there's one, one guy. There's one guy. So you sit there, I think this is where you go, well, wait a minute, how do you crack this problem? Do you, uh, give people economic incentives to live in rural Montana? Does that work? I don't think so. And then, oh, well, we're going to build tons more medical schools. I mean, and that's when you know, as obviously people understand today you go, this AI technology has a role to play. Not in, so to speak, replacing anybody, but in the idea that there is nobody. And there is nobody in a lot of locations in the United States and obviously rest of world.

Speaker B: Mhm. And you know, continuing along that while there's a lot of that missing, there's another part of this, this issue. And this is core to where we're going to be going with the moonshot. But every children's hospital, as we're sitting here talking, uh, has an immense amount of so called trapped data. So every electronic medical record, every mri, every CAT scan, every ventilator, ICU monitor, whatever the case may be, all this equipment generates staggering amounts of information, staggering amounts of data. And it sort of sits there. Um, and none of these systems communicate in real time with equivalent systems, other hospitals, some of it might get thrown up to the cloud and Sit there and then there may be some AI stuff done up there, but ultimately, um, we're not learning. Children's Hospital Philadelphia isn't learning from, you, uh, know, whatever, you know, the main hospital in LA and so forth and so on. Talk about this. Uh, because again, where we have these islands of knowledge, but when they're not communicating with other, this stuff's going to stagnate. So this is another very important piece of what you're going to engineer us out of, but take us down this path as well.

Speaker A: Um, yeah, let's go down that path. I think maybe your listeners know this, but if you look at why has there been an advancement in AI that we all see today as Claude or chatgpt? Well, the reason is that what they're doing is they're accessing large quantities of training data, uh, New York Times, Reddit textbooks, et cetera, and learning on it, immense amounts of data learning on it, neural networks which are the core technology and basically the only technology where you can demonstrate if you apply more compute and more data, you can get higher degrees of accuracy. And that's what's been going on. Hence all the conversations about building big ass data centers, LA or whatever. Right, okay, well when you look at that and you say, well that's really good for Reddit post and New York Times and you know, learning where you should go on travel, if you go into the world of healthcare and life science we looked at this is never going to work to just bring it all into a giant data center in Louisiana. Number one, the data size are much larger. I mean this is not a bunch of text files. I mean imaging data, uh, C.T. mRI, et cetera, genetic data, uh, the VCF file, which is the difference file, you know this, I mean they're 300 to 500 meg, they're not small things. Right. So number one, number two, security requirements. Right? This is personal data and privacy being the most interesting part of the challenge. And on a global basis, Norway saying we do not want Norwegian data leaving Norway. Right? So when you look at these challenges of data size, uh, security and privacy requirements, you've come to the conclusion that. The conclusion we came to is, well, this is Computer Science 101. Rather than move all the data into an application in Louisiana, let's move the application to the data. And so what we engineered is a distributed AI cloud infrastructure which literally puts a cloud server inside the building of a hospital, research lab, clinic, uh, ambulance or home with. And we had an objective to want to build real time systems because it's like, yeah, there's a lot of good after the fact knowledge that can be derived. But don't we want real time systems? And the only way, and I'm just going to pick on ultrasounds is the only way to get real time data from ultrasounds is you have to be on the network with the ultrasound. There's no other answer. Right. And, and so you could imagine the first time we had a conversation with hospital was followed with nfw. You know, make no mistake about we are putting a server on their network. And uh, we knew this going in. I mean you know all my background basically infrastructure, software, hardware, we knew this going in. So we said we, we engineered over 30 security and privacy features to enable this to happen. As an engineering team we went, you know, let's go do this. And so we implemented, we call them zones in eight sites on three continents because we wanted to prove a global infrastructure. And that really that whole idea of what people will call federated or distributed infrastructure right to me sits at the core. I don't, me personally, somebody else may figure this out as well. I mean that always happens. But at the end of the day I don't see any other answer in healthcare and life science other than moving to a distributed infrastructure.

Speaker B: Mhm. And along those lines. So um, instead of ultimately having sort of one gigantic, you know, pediatric brain out there, ultimately what you're proposing is literally hundreds of, you know, smaller connected pediatric brains and you know, hospital one is running their AI, hospital B, theirs, hospital C. The information never leaves, but insights leave, uh, various outputs. Um, talk about now as part of those A500 connected brains or zones, whatever a subcomponent of that is, so called the one million machines. And so this is um, you know again coming back to every component of the diagnostic equation. Mri, CAT scan, PET scan, the critical care machine, the lab, the clinical, all these represent um, you know, ultimately 1 million connected healthcare devices across the world, children's hospitals. Um, and in this concept, uh, AI is learning from everything simultaneously, um, in real time. And we'll go into some real world applications or some. But say a couple more words about this guy. You said people hear a million machines. They're probably, I don't know what they're thinking about robots or whatever, but we're talking about all the stuff that currently sits there again holding these, you know, this amazing amount of data currently, um, take us into the one million machine world.

Speaker A: Yeah, let me just throw an example out there and what, what it implies. So we did a back of the envelope the 500 children's hospitals in the world in the cardiology department, just in cardiology, in what they refer to as the echo lab, which is where they do the ultrasounds. We back of the envelope. Estimated 6 million terabytes of training data are being created annually every year. 6 million terabytes. To a point you made earlier. Nobody's learning anything about them across that 6 million terabytes. It's just being done. One person does one thing and one person makes one observation. You know, we, we know how you could now learn a whole. I mean this gets into what is the application of this. So can we learn how to do congenital heart defect detection, uh, on ultrasound very early in pregnancy? The answer is yes, we could. I made the comment we're tracking 105 cardiac diseases. A third of those are congenital heart disease issues which are structural. Now could you build AI applications which precisely detect all 35 of them, which given this design, can now run at the point of care, whether that's in Rwanda, you know, rural Canada or in Philadelphia. And now we, we can take, if you will, what we learn across all that 6 million terabytes and now apply it at the point of care. Uh, I think that's, we all know, I say everybody in the middle of this know that that's where it could be. The challenges have been what we think from an infrastructure point of view we have solved, which is the whole data share. How do you share data while preserving privacy?

Speaker B: Mhm.

Speaker A: We think we know the answer to that question and that unlocks everything. Because the amount of training data is crazy. I mean if you think about it, you know, in a matter of a year or two you could be able to do every disease known to man. Literally.

Speaker B: Yeah, let's um, let's continue there. And again for the audience, if we think about the old model, um, um, okay, so we collect this data, we're going to send it up somewhere in the cloud, we'll analyze it later, we'll publish and then we'll get these new human insights. At some point later in the new model, uh, the data appears, uh, and that could be as you're saying, in cardiovascular, um, we do a CAT scan instantly, you know, AI is reviewing it, it's looking across the previous experience, suggesting possibilities. Then we get a human involved and the radiologist comes in and you know, uh, this icu we need to, we see something going on, we probably should keep them here. Um, and right away, ultimately the clinician is warned immediately. Early on in the process, we're not learning stuff months later. Um, and here again, this is, I think, a really nice examp because we hear about AI agents all the time in all fields of technology. I think this is sort of a beautiful example, and there may be many others about what exactly an AI agent can be doing real time in a children's hospital. And that can apply to sepsis monitoring or medication safety or whatever the case may be. But I think that's ultimately, again, bringing it back to an example like that. The difference between what's been going on versus what could be happening.

Speaker A: Yeah, um, it's a good segue into what. What we've been working on over the past couple of months is really perfecting what we call patient digital twins. So. So if you think about it, boy, couldn't I build a digital representation of AIRA from a health perspective based on, okay, the medical record that you have, the last three CT scans you had. Well, I hope you didn't have any CT scans, but the last three, if you had to have them, uh, your genetic, your VCF file, uh, oh, and your wearable data so that we know how well you were sleeping, etc. Right. And if I could privacy protect the version of AIRA at. Ah, and let's assume you're at Children's of Philadelphia for a minute. But, oh, by the way, if. And this happens in children's medicine, obviously a lot, which is the kid is cured of cancer at CHOP and then moves to Montana. Well, there are known side effects of chemotherapies to the heart. So the kid's cured of cancer, but he dies of a heart condition. Well, in today's world, as soon as that kid moves to Montana, he disappears out of the system. Nobody knows who he. Oh, you had cancer in the world. We're building as soon as he moves to Montana and we create a digital twin in Montana of aira. You're now moving to Montana. Uh, the global system can now go, oh, no, wait a minute. The IRA in Philadelphia and the AIRA in Montana are actually the same aira. And now all sorts of analysis can be done. You know, recommendations, care, treatment, et cetera. Because now I know all the data about aira. Okay, now, if I have that, which. Yes, we're already building that, then. And you start to think about, well, what kind of agents would you build? And that's when I go back to this whole thing about atlases. Um, and we started in kidney diseases. I cannot explain why, but we did. Um, and I'm just going to give an example. There is A kidney disease called fsgs. Sure. Okay. Fsgs and your biopharma. Uh, heritage.

Speaker B: I worked. I worked in it a couple years. It's a whole other story.

Speaker A: Go ahead. Okay. And you know, also the fact that, right. Increasingly in. In your old world, people are building, you know, what are called orphan therapeutics, right? Which is 200,000 people or less. Well, then you're at, okay, if you only have 200,000 people that this drug is effective for, and forget about kids, Your first question has to be, well, how do you find them? And, you know, today's technology is go run super bowl ads and do lunch and learns, which seems kind of silly. What if, on the other hand, you could build an FSGS agent, which we call them detection agents, which all it does is looks at the patient, digital twin, ask it a bunch of questions. In fact, 34 questions. Urine analysis, blood analysis, biopsy, et cetera. And every answer to that question gains a certain number of confidence points. So you can imagine now the agent is just grading the patient. Okay, if you grade over 90, that means you really probably have FSGs. And, oh, by the way, here are the two approved therapeutics. If you grade under 50, you can pass. If you're in between, we call it the yellow zone. You could say, well, you might have FSGs. Um, the next most expensive test is to get a urinalysis or a genetic test or whatever, right? Now take that agent. Let's assume the detection agent hits red, right? You grade over 90, you can now invoke a progression agent. Question agents the same fundamental idea, which is, okay, let's go ask a bunch of questions of the patient, digital twin and grade. Whether this patient is moving quickly, slowly, or in between, obviously affecting what treatment decisions you could make. And in the world we are imagining those treatment decisions could be different in different geographies based on resource constraints. Right? That's pretty easy to implement. And then, okay, if the patient in effect is red, red. Well, are there any clinical trials? And you're well aware of the clinical trial problem, which is, well, how do you recruit these patients when you're in small, you know, 200,000 people or less environments today, extremely expensive, takes a crapload of time to do this. All in essence, done with manual labor. We're like, you don't have to do that. A trial agent is just the same idea, which is the questions are now, in essence, the inclusion exclusion criteria. So you can affect grade the patient. Oh, totally ineligible. Oh, totally eligible, or might be eligible if they only got an X ray or whatever it turns out to be. Uh, and now you can also divorce, which is in today's world, you're linked together, meaning where the clinical trial occurs and who I go ask if they have any patients are connected together. There's no, you can't unhook it in this world. You can now say, let's go, let those trial agents run anywhere. We may discover that in, I don't know, in Central Iowa, there's 50 patients. Well then let's go build a clinical trial site there. Right. And then you can also quite simply start to think about, well, why is trial design the way it is? I mean, you could actually start to go, well, if I say brown eyes, I get 50 patients. If I say blue eyes, I get 500. Oh, well, blue, right. Uh, all of this, all these agents are just examples of what is possible, given you can now in essence free up the data and enable compute on it.

Speaker B: Mhm.

Speaker A: Yeah.

Speaker B: I mean, FSGS is an awesome example when we talk about the orphan conditions. But here, you know, you're dealing with 80,000 or so people in the United States, and in a couple thousand cases a year. Um, let's go a little further in terms of, you know, what we, early on we're talking about rare diseases or also the ultra rare or undiagnosed. Because here again, you know, 300 million people worldwide, a lot of them are children. Uh, you know, you got eight to 10,000, uh, classifications here. Um, and you know, in this case, right, I mean, some of these diseases, 100 people in the world, um, it's exciting because you say, okay, um, I have a case over here in Kenya, uh, this rare metabolic disorder. Hey, there's a case in Brazil also, and there's one here in Boston. Um, and maybe this is a, you know, decent chunk of the total population. But nonetheless, here is another example because again, this is, this topic of rare and ultra rare is extremely hot and important nowadays. Um, you're beginning to see sort of like the potential for this integrated vision for all diseases. Uh, you know, whether it's norm, you know, the big stuff, the orphan or the ultra rare. To say whole words about this, because this is where things get, you know, internationalization of this stuff comes into play. And you know, when I, when I have someone in Kenya and someone in Brazil with ultra rare, it might be decades before we're aware of these people existing. You can make it, you know, almost instantaneous with this model.

Speaker A: Oh, absolutely. Uh, I was going to make a comment because it's something we've been starting to work on as well. So I think there's one way to see this problem which is the things I know, right, like okay, I know FSGs, I can go detect it, I can determine progression, etc. But clearly, and I already said, made this point, yeah, we knew 106 kidney diseases 25 years ago, today is 176, tomorrow it's going to be 250, whatever. Well, how do we learn about the stuff that we don't know, right? And here you start to realize that I will say from an infrastructure point of view, the only infrastructure that's been created are these, I'll call them registries, uh, edc, right. And they're all a fairly primitive, I'll call it SQL based technology which demands humans type in data into a fixed schema, which makes them expensive to do because you got to have human labor to do it. Relatively inaccurate because, you know, humans are typing it in. Oh, not very representative because it's only at sites, you know, the big major academic sites that you could possibly have the resources to go do this. Right. And then entirely inflexible. I mean as we've spent a little bit of time with the National Ms. Society, you know, stepping out of pediatrics and um, Tim Ket says, the CEO of it, he said, I want to be able to ask the question that I will ask in three years. I don't know what that question is yet. And you know the SQL fixed SQL schemas leave you with the problem of if you never asked if they had brown eyes or blue eyes, well you're never going to know if they have brown eyes or blue eyes right now on the other hand, and that's what we're starting to build disease precise registries. So all that is is once you have the patient digital twin and once I know that that patient is of this is in this registry, however you want to put them in that registry, you could say I'm going to run the disease precise agent. So I'm going to create an FSGS registry of every kid, uh, with fsgs. Now I can ask a federated query, meaning I can go centrally and ask the question, uh, how many, whatever, we're taking this drug and have blue eyes or whatever it is. Now I can federate that query across every site that has a uh, patient with that condition. We've done the back of the envelope. It would be game changing even in MS, which is funny to think about, to have 100,000 person patient registry, 100,000 representative patient registry, we've done the back of the envelope is between 130 and 160 sites. Because back to the point, you don't want everybody from the major academic. You want some in community, you want some in smaller institution. 130 to 160 sites. Now I have 100,000. Patient registry, which now has enormous data in it that's flexibly accessible, not dependent on humans entering anything. It's because the data is actually being created by the patient. Digital twins. You know, this may be, from an IT perspective, the way that we make the progress towards understanding more about the diseases we don't understand yet, of which even Ms. Would fall in that category. Right. Um, and this. And when you talk about rare and ultra rare, how will we learn about these? I think it'll be, we will have patient registries across every disease that are specific to that disease.

Speaker B: You know, the, um. Yeah, I see the amazing potential here. And then there's sort of one other angle to it, which again, we said we spend a lot of time talking about. And this is this whole, uh, concept of repurposing and this era that we find ourselves in when we ultimately find out what we have with some of these diseases. Um, you know, obviously there aren't going to be drug development programs for a lot of them because of the small patient populations. But learning from, hey, you know, I tried drug X, Y and Z, you know, here and had some decent results. And sort of, sort of the digitalization of sort of clinical experience, um, it really, again, gets you, um, a decent distance from where we are. Um, you know, and just thinking of the time it takes to learn about these patients, let alone, you know, come up with a treatment plan, um, for the ultra rare. So this is, this is very exciting. I, um, you know, obviously see the tremendous potential, um, where, I mean, obviously the, the big picture and you know, people can. We'll put the link to the website and all that. We, we talked about the million machines we got, the 500 children's hospitals. I saw a number out there of, you know, a thousand AI agents across these different indications. Um, where are we now? Talk a little bit about some of the sort of the pilot stuff you're doing and, you know, what should we be expecting in terms of some of the milestones as we get close to 20, 27? Um, take us down.

Speaker A: Um, so I'll, uh, go backward in time a little bit. So really, over the past, I'll say three years, we really kind of wanted to prove to ourselves that this technology could be put in to a major institution. By the way, uh, what we now compute is that it takes between nine and 12 months from the time they say yes till we actually implement. Uh, security review, privacy reviews. Uh, I, uh, will tell you at one point, at one of the major children's hospitals where the CEO said we should do this, uh, we stopped for a month and a half while procurement, and I don't know why procurement was involved, said, oh, that's going to invalid our warranty on our Philips machines, which is like totally Looney Tunes. Anyway, we now go, okay, fine. We know it takes this time, and it's not really technical. In fact, we designed these servers. Remember, our objective was really to push this technology out, not major medical institutions. So you can't depend on there being some sophisticated IT person at the other end. So we had an objective that a server should be able to be installed in less time than it takes to put in a nest thermostat, which we're pretty close to. Right. Uh, in fact, the deployment we did in Rome, in the Vatican, actually occurred during COVID And none of us were there because we couldn't be there. So. So we feel like we know how to go do that, but the next step is really scale. And so we said, okay, uh, let's go build. We call it an AI supercomputer for children's medicine. What is that? Uh, 32 sites, all imaging machine data available in real time. All packs, all EMR data available. Uh, roughly 3,000 servers, roughly 2,000 terabytes of training data flowing through it annually, and roughly 6 million patient digital twins. Okay, so, and we kind of. It's very arbitrary. We said, that's a critical mass. If we could get there, then adding a 33rd, a 34 site would get easier and easier. What we would learn from that scale would be meaningful. So that's the objective. Uh, we did a back of the envelope. We estimated the build cost of this at rough 50 million. Uh, I've been funding all of our work to date. Uh, that was a little bit beyond my pocketbook. So, uh, coming to the conclusion that there is no commercial interest in pediatric medicine, which is true. You just have to say, yes, it's true. Um, we created Pediatric Moonshot as a dot org and really launched a capital campaign. We'll call it early this year. Um, the capital campaign has kind of multiple components to it at this point. The first at the high end, we're saying we'd like people to sponsor a node. So what does that mean? Okay, pick one of four different disciplines. Neurology, uh, nephrology, cardiology or oncology, uh, at a site. One of the sites is $250,000. So we took the $50 million ask down to a much more manageable number, uh, which we are, we are talking to high net worth individuals. Pleased to say we already have one six figure there. Uh, foundations. Right. Uh, and then corporations. One of the things, uh, we haven't done this at scale yet is as you've heard me talk, I think a lot of what we're doing could apply equally into adult medicine. So we want to go to the pharma companies and say to them, guys, look, this could be meaningful in your regular business. I know you really don't care about pediatrics. Okay, that's okay. You're going to want to see a POC anyway of what I'm saying to you. So why don't you fund a node, right? We'll show you the results from the node. And oh, by the way, you did something good for children's medicine. So that's kind of the three high net worth foundations and, and corporates, uh, at that we'll call it the high end. At the retail end, we, we've become students of St. Jude's St. Jude. This is freaking amazing. I don't think people know this. Or maybe some people do. You know, they generate $2.7 billion a year in donations. Wow. I mean it dwarfs everything else by a long shot right? Now obviously they've been added a long time. The Danny Thomas thing, I mean there' a lot of reasons why this is true. But it also said, and by the way, 1 billion of that is occurring retail. The hundred dollars, the five hundred dollars. So we said, okay, we're going to do that. So we are in the middle of pediatricmoonshot.org which you can see right now, uh, probably in about a month is going to be completely new and completely dedicated to, we call it, we think of it as retail fundraising to support the pediatric moonshot. So that's very much the direction we're headed from a funding perspective. Um, there's one. We, we've, we've had some conversations with the ARPA health group where if you go read their mission, you go, there's like complete connection. But I'll tell you, we've, you know, it's just never really worked. I, at the end of the day kind of went, okay, I've never done much with US government. Most of the team hasn't. So we're not really connected into their world. So we kind of walked away from that. Now, the one area we are focused on right now is you may be aware of the Rural Healthcare Transformation Program.

Speaker B: Absolutely.

Speaker A: Uh, yeah, 50 billion, uh, being allocated to the states. The states are getting, I'll call it between 150 and 350 million this year to spend. Um, and so we went, well, we built all this again, not for the kid in Palo Alto, but for the kid in Northern California, which is four hours away from any children's hospital. Um, so we are actively in conversations with, uh, the children's hospital in what we think of as a hub. So I'll use Seattle, Seattle Children's is a hub in Washington to make a proposal to the Washington version of this that they fund. And what would they fund? Being able to deploy our disease precise detection progression trial agents into the community, into the rural hospitals where if you're in the situation, we call it red, red. The kid is red on detection, red on progression. Well, right away you could signal a, you know, connection, a, uh, referral into the pediatric nephrology department at the, at Seattle Children's. So a lot of different combinations, but treating the hub as the hub of expertise and in essence you could think about it as advanced triage out in the rural communities to bring, if you want to think about it, the level of Expertise, knowing all 176 kidney diseases is actually, you could say, conceptually more than any one nephrologist knows. And being able to put that out in the field. Right. So we're, we're actively in that conversation. I'll, I'll tell you, I just had a conversation with the governor of Montana. So. Yeah. Uh, because it dovetails so much with how we, what the mission of the whole thing was is really to get to the children that, you know, are not being, are not being represented, whether it's in the United States or around the world.

Speaker B: Mhm.

Speaker A: So we're actively doing that. Yeah.

Speaker B: Sorry to interrupt you. No, I mean you brought up the part before about, um, it was interesting about it sort of the, the side business, but the, the sort of this aspect of uh, the pharma companies getting involved and sort of, you know, I, I've done a numerous amount of sort of artificial intelligence drug discovery episodes the last couple of years and I never can get away from the whole, you know, how we don't have these high data sets still. I know a lot of people think you just push a button here and the magic pill pops out over there. But we're very far away from that, you know, the same rules that come into play with this data fragmentation that you're solving here, I think. I think there's a really cool side business there that could fund everything else. But putting that aside, Tim, you know, going back to what you were mentioning before about, you know, the 100 pediatric cardiologists in India and the one guy in Rwanda and all that. Yeah, I've had, you know, some of these, you know, folks on. I just did a show with the Minister of Health of Albania, and it's surprising, sort of the degree of digitalization, uh, that some of these countries already have in their health care systems. So the United States, and I'm just. I'm interested, you know, what's happened internationally. I mean, I know you're all over the place with this and discussing these themes, but, um, anything cool happening overseas while we have you today that we should know about, and. And some of those trenches, because clearly they need it, but they also, you know, some of these countries are building up pretty decent digital infrastructure that seems like would jive really nicely what you're doing.

Speaker A: Yeah, um, I'll tell you what we're doing. I mean, obviously there's only so many hours in the day, so we've been very focused in the US but we did deploy in. In Europe and in South America to prove out the technology. Um, I think, you know, everything we're talking about applies, I think, where everything is different, and it's even an area that we are starting to kind of try to unravel is how's the payment mechanism work. Right. And our US System is not like everybody else's.

Speaker B: And.

Speaker A: And I think that causes a whole set of things to be somewhat different, ah. In how this might end up getting deployed. Um, now, are the. The challenges that they have in Italy and England and Brazil all the same as we just stated in the U.S. yes, absolutely. Um, how will it get funded is a curious question. Uh, we. I mean, uh, you know, it's crazy to say this, but given the fact that St. Jude's generates a billion in direct donations and, by the way, is only focused in cancer, really, and pretty much only focused in the US Although they're getting more international as well, it would make an argument that could we not. Could pediatric moonshot not hit a billion in philanthropy philanthropically? Maybe it could. And maybe that is ultimately the answer to how does this end up getting deployed in Brazil? I mean, the flip side is obviously the question which a lot of outside the US Is very government. Right. Um, and how does that we have not had any conversations at the governmental level in any country about this, but you could imagine that being another side of this is somewhat different than in the U.S. um, yeah, but the challenges are all the same. I mean, kids are kids.

Speaker B: That's true. We gotta keep them healthy. They're important.

Speaker A: Well, you know, they're the, they are the future. I mean, like, uh, we like to say it. We don't actually spend like that just to make a point of it.

Speaker B: No, this is, this is, is true. Um, where are we going to find you, Tim? Uh, as we get a little deeper into 2026, is there any other sort of public facing stuff that's coming up for the initiative in terms of talks and conferences? I mean, I've watched several of your presentations online, but, um, anything else happening that's public facing that we should be aware of, um, while we have you today and anything I missed too?

Speaker A: No, no. I mean, I think just to say it. I mean, we, we're constantly trying to be online telling people about what's going on, so. Pediatricmoonshot.org you could sign up for a newsletter. We try to do it once a month. We don't always get it out once a month. Uh, the, uh, pediatric moonshot YouTube channel. It's not beautiful yet, but there is. Interesting.

Speaker B: I watched it. It was okay.

Speaker A: It's good. Uh, we've done our own podcast series, Pediatric Moonshot, on, you know, Spotify, Apple Music. We've had, you know, CEOs of some of the major, uh, hospitals out there, leading cardiologist, oncologist, etc. On it, who, you know, I would say have been supportive of our mission and, you know, bring their unique perspectives. Um, we have a role in all this, which is an advocacy role. And, you know, so we're trying to do that as well. And, and in a way in which, you know, is, let's call it cost efficient, meaning we're not flying anything, anybody around, um, but yet trying to get the word out. So we're very much, um, you know, engaging with people. Um, we have some interesting things on the sidelines that we think we'll be talking about in the next couple months. Um, you know, that. And I think every step along the way, the inflection, I mean, I hate to say it, but the inflection point right now is all money. We know how all this could work. We know it can work. We have m. I'd say very good support from the clinical community, you know, top to bottom. I mean, whether you want to talk about A CEO or the head of cardiology. Uh, I think there's a collective belief that, yeah, here we need to go do this. So we just need to figure out the money thing. And I'm not. We're being pleasantly surprised at the progress we get just one of these things in the rural thing to hit. I mean, Jesus. I mean, you get to a whole nother level. Yeah.

Speaker B: So I think you're on the right path. I mean, clearly, um, this model is required, um, and I look forward to helping you share the word and get the message out there, um, because they are the future. Um, and we have a lot of unmet needs out there medically and, uh, have to solve this. And so, I mean, again, it's an impressive model. I wish you the best with it. Um, again, for everybody out there that is going to be listening to this episode, uh, or will be watching on the YouTube channel. Um, again, you've been spending time with Dr. Timothy Chow, founder of the Pediatric Moonshot. We're going to put links to the website as well as their podcast, uh, in the show notes, but encourage everyone to check it out. Um, Tim, I want to thank you for taking the time out today to come on the show to talk to us for a little while. What your vision is there. Um, thank you for everything you, you look to do for children as well. Say on our show, thanks for helping to create better tomorrow for people. What you're doing, it's a, it's a great story. I'll look forward helping you share it and continue to follow what you're up to.

Speaker A: Well, Ira, you know, we, we like to say it took 40, 000 people to get to the moon. We're a few shy, so I appreciate all your help in, in recruiting more people to the cause. So thank you.

Speaker B: Absolutely awesome having you.

Speaker A: Yeah.