Thank you, ladies and gentlemen, it's a pleasure to be here to a fireside chat without a fire on a rainy morning in California. But it's an honor to be able to sit with a good friend, Eric Topol, for a little bit. And for those that do know Eric, that's great. For those that don't, I'm sorry. But regardless, if you know him, I'm going to introduce him, because it's important, I think, to hear a blend of mentorship, friendship, and collegial connection over the years. Over the four decades, if you will, Eric's been contributing in this space that is really changing. You know, somebody called Eric the father of medicine, which I actually think is confining. He is an ultimate visionary and one of the leaders that you can put on one hand in contributing to medicine in the last four decades. He started at UCSF. We intersected a little bit there. I had two mentors in San Francisco. One was a guy named Kanu Chatterjee. And he didn't really want to be my mentor, but I just bugged him every single day at 5 a.m. and watched him and went on rounds with him. Eric, I was his mentee, at least I thought I was, but he didn't know he was my mentor. So I think I kind of stalked him, if you will. He went on to Johns Hopkins and did his fellowship, then rocketed into a professorship in Michigan, and then became the youngest chairman of medicine at age 35, 36, I can't quite remember, chair of medicine at Cleveland Clinic, where he taught all of us how to take care of acute myocardial infarctions, when to give these blood thinning medications, to who, how, and when. One of the biggest fulcrums of how we take care of people in cardiology. He wrote two books, Cardiovascular Medicine and Interventional Cardiology. I think he let us put a chapter or something in there at that time. There's a guy named Paul Yock and I, and we were putting ultrasound catheters inside a blood vessel, and he was helping us with, that's fibrous tissue and it needs this sort of intervention, or that's thrombus and it needs this sort of intervention, and that really helped with how we democratize our tools to fix some of the coronary arteries. Those two books were the cornerstone of cardiology. Then went to the lighter side of California, where it doesn't rain, at least we told him, on a rainy day, and formed the Translational Research Institute, brought in hundreds of millions of dollars, and did two things, many more, but two things I want to focus on. One, he was talking about digital health before even the word was there. I'll tell you a story. We were at the ACC many years ago, and I think it was in Chicago, and I said, how you doing, Eric? He said, good, I'm just busy. I got a lot of talks, and I'm on a call, and I stopped for a minute, and I said, you're on call in San Diego? You know we're in Chicago, right? He said, yeah, no problem, and he held up his BlackBerry, and he had all the patients in the CCU, their hemodynamics, on his BlackBerry that he was watching, and I said, holy crap. This is something that Eric has always been on the edge of, that is really going to change the way we think about things. We're going to talk a little bit about genetic contribution, but in a minute, 1,200 peer-reviewed papers, I don't know how you sleep. Three more books, The Creative Destruction of Medicine, The Patient Will See You Now, making sure that we appreciate that patients, people, are going to be more consumer-driven as we evolve in the technology that's coming into medicine, and a great book called Deep Medicine, and if you haven't read it, you should. The Audible's okay, but reading it's better, and if you're really lazy, just read chapter 13, because it's very poignant about today what we're doing with telehealth. It talks about deep empathy, and God forbid we get more caring and empathetic in our care, especially as we come out of this pandemic of COVID, so with that, I want to talk a little bit about COVID, because Eric has about been on every broadcast channel I've seen and newspapers talking about some of the things we did right and some of the things we did wrong, and you're very outspoken about remdesivir, very outspoken about convalescent plasma, so tell us what we did wrong during this COVID issue. Well, first, Peter, thank you so much for your intro. I wonder if you're available as an agent, I'd be really appreciative, I mean, really, thank you, really kind of you. It's great to be back with a cardiology crew here, I don't get a chance to do that very much, but as far as the COVID thing, what was interesting is back in the early 2020, one of the guys in our group, Christian Anderson, who is a pathogen genomics guru, he got me stimulated about what this virus might be doing long before it was recognized that it was actually in the US. I said, well, gee, maybe I should take a time out and kind of become a COVIDologist, you know, and so I started learning and obviously talking to a lot of people in the different disciplines, you know, virology, immunology, and whatnot, and basically got to the point where, you know, we obviously have had many issues. We had this momentous, accelerated, unprecedented vaccine production with 95% efficacy, but we let down, that is, we didn't do an Operation Warp Speed number two, where we could have nasal vaccines, which we're now seeing crop up from work in India and China. The work from India was a Washington University intellectual property, and there will be many other successful nasal, oral vaccines, as well as pan-Cervical virus vaccines, but we're not doing that here in the US, and everybody's tired of, you know, all the other things like masking and distancing and want to move on with their life, but we need to be ready to contain the virus. So that's the biggest thing I'd say, Peter, is that we just haven't gone to the next level. I mean, you could do the vaccines, which normally, I think everybody knows here, takes eight to ten years, and that's if you're successful. There's many pathogens we don't have as successful. So the fact that you can do that in ten months, would, you'd think, would say, and it wasn't that expensive to do it, when you look at the big picture, why not reload and do that again and again? So we just don't have the will, and part of that, frankly, is the lack of, the political strife, the lack of support, lack of unanimity in our government to go after this sort of thing. These specifically targeted amino acids wrapped in a butterball that you can generate for different issues, not just viruses, but also potentially cancers, and having that, and Gates has always talked about that, having that armament ready to pull down when these surprising things happen, that's going to be the future, no? Or do we, does the private side need to do this? No, you're talking about something that is going to have a huge impact in our whole approach, because the mRNA story with nanoparticles is much bigger than these vaccines, and it will get us to universal flu vaccines, and it's going to get us to treat pathogens that previously we're not able to do, like RSV, and Zika, and many others. The real issue, though, is that, as you say, with ability to get into cells, and now, you know, to, we're talking about cancer, neurodegenerative diseases, autoimmune diseases, and there's also a really exciting area, and it's really what we call tolerogenic vaccines, which is turning off the immune system, so it's just the opposite of what we're doing now, but basically, you can take the mRNA, and the nanoparticles are going to get much better. All four components of nanoparticles have tremendous amount of up potential, so then you can start to say, you know what, a child has a high risk of autoimmune diabetes, we're going to give a tolerogenic vaccine at, you know, at birth, or, you know, in their first few months, and they'll never get type 1 diabetes. The same thing for people with lupus, and, you know, rheumatoid arthritis, so we're going to see, maybe, now, we may not see it all the way through, but we might not see it all the way through. There's a really exciting thing that this pandemic, the bright side, that we have just starting to recognize how extraordinary that is. And I think that's important, Eric, because, you know, there's lots of books, and lots of info about how bad it was, and how things screwed up, fantasy before science, but we really do need to back up, take a silver lining approach, learn from it, and I do think it's going to be more privatized in those efforts going forward. Let's talk a little bit about medicine. So, you know, training for all of us healthcare providers, nurses, docs, physician's assistants, we tend to have a pretty structured way of how we practice medicine. You train, it's kind of like following a carrot around a track. There's just different tracks, depending upon the subspecialty. It takes a lot to move us out of those tracks. It takes a forcing function. We just live through a forcing function, and as Adam Grant says, if you haven't read the book, Think Again, we need to rethink, redeploy, and readjust how we're practicing medicine. And again, we're not going to do that if we just keep going, but this forcing function that we've just lived through has changed the industry, it's changed medical centers, it's changed the way people, patients, consumers, and the way docs are interacting. How do you see that forcing function changing maybe some of the things you've been talking about but haven't been embraced like they should with digital health, let's say? Well, you know, you touched on this in your early remarks about we still don't have the appreciation for digital health that we should. So it's fascinating that it's still, in the medical community and beyond that, it's still kind of the Rodney Dangerfield. We get no respect, I mean, really. You know, we talked about it while we're getting ready for this session together, and just last week there was a remarkable study out of Germany, right up the alley of this group, where they took several thousand people, randomized them to, their median age was 65, to an app with an ECG recording versus usual care, and they showed in a randomized trial one of the best done in digital health today, more than doubling of picking up people that needed, had atrial fibrillation that needed anticoagulation, so not just diagnosis of asymptomatic atrial fibrillation but serious atrial fibrillation that was a high risk for stroke and needed anticoagulation. So I wish we had more of these trials earlier, but this is now finally cementing the case for digital health. The problem that we've had is we've been able to collect a lot of data, but the rigor of trials and the compelling case is just like AI in medicine today, which builds on digital. If you don't have compelling data, you won't change the, you know, you won't really push the field forward, and that's been part of the problem with digital health, is here you can pick out a recent trial and say, wow, this has been phenomenal, but how many of those do we really have? That's right, that's right. You know, we use loosely the term AI and machine learning, ML, and it's important, but it's as important as the training sets are. So when we take a bunch of data reservoirs from spurious places, that doesn't form a training set to do active reinforcement learning. So we have to get better at not only morselizing the data and making it homogenous for our training sets, but also morselize it so that the healthcare professions can interact with it in a dashboard-like process and not a stream of data, because that's the last thing we want, is a stream of data with five times more patients. So it is important, but the issue that I get excited about is that maybe it leans back on these curves, and rather than being able to deal with what's present, we might be able to predict what might be present, and therefore have an impact on that curve, rather than changing that curve once it's separated. But if you lean back on being able to diagnose things earlier, and I think that's another contribution of digital health that's going to be important. We use this consumerized sort of term loosely, but I do think for the healthcare professionals in the room, we need to think about patients equals people equals consumers. This is brought out by one of your books, The Patient Will See You Now, meaning that democratization, dissemination of care is already happening, but it's not just us that are dictating it, and we do need to embrace that change, because whether you like it or not, it is happening, and it's happening at warp speed. Just look at what has happened recently. We've got Oracle and NLP. We've got Amazon and OneMed. Just this Monday, we have CVS and Signify Health, so that that footprint can extend to the home. The private companies are coming in, and they're changing things. The patient is the consumer. Is that fortifying what you predicted four years, five years, six years ago? It's taking a while to happen, but eventually, the much greater democratization than what we've seen so far is inevitable. As you're bringing out, Peter, the fundamentals, you're seeing them happen. When you see the interest of companies like Amazon, as you mentioned, to CVS, which obviously already had started a business unit of doing research at the home, giving investigational drugs and infusions in the home, and now acquiring perhaps the best way to get further expansion of home to then reduce costs as well. The convenience of the home, the hospital at home, this will be one of the biggest trends over the next decade, is shifting. Again, this is with the tools that we have today, with sensors and AI, the ability to predict decompensation before it happens. Why would you use a hospital room? CVS gets that. They have Aetna that is obviously, its business is to big employers. The whole idea is, why can't we lower costs by keeping people in their home instead of being admitted to the hospital? Here, digital health will be the core of it, but it's the private large titans, these companies. When you see, for example, Best Buy buy current health, a sensor for continuous vital signs, you ought to say, hmm, why would Best Buy buy current health? Because nobody goes to the geek squad anymore. They just repurpose the geek squad to home. These are trends that are happening parallel, many different lines of evidence, to show you how medicine is going to change. It will be very different. Remember, when we were doing our training in medicine, if someone came in for a cardiac cath, they would be overnight, one or two nights. Because you're putting a catheter near the heart and all that. Then that became an outpatient procedure through the wrist in a few minutes. By the way, maybe we'll put in a bunch of stents and we'll send you home in a few minutes. Take that and say, well, we shifted things to outpatient. No, we're shifting things, not procedural, but a lot of things are shifted to home. That is what the patients and their families want. By the way, of the $4 trillion that the U.S. spends each year for healthcare costs, the biggest line item, 1.3, 1.4 trillion, is the hospital. What's the biggest lobbying force in this country, outside of guns and the pharma industry, is the American Hospital Association. In this country, we have some obstacles, but around the world, we'll see this happen relatively quickly. Well, on that curve that spends $4 billion and yet we're number one in maternal mortality, we have four out of five hypertensives out of control. There was a great article the other day in Cirque, it says the doctor's out. He or she is just not doing the episodes of a chronic care-like hypertension effectively. I think what stitches some of that together and maybe allows folks that aren't as skilled is and are the data to help hierarchy what to do during those episodes. I think that's important. Remember, folks, 20 years ago, he showed me hemodynamic monitoring on a BlackBerry. Here 20 years later, we're actually doing it. You just heard Nelson talk about biotronics and a live core. Those sorts of processes, whether we like it or not, whether we're stuck in our own ways of how we've been doing medicine, is completely going to be different. That's what, at least, I get excited about while I keep working. I think that's why you keep working. It's an important issue. Listen, we don't have mics here, I just noticed, so if people have questions, I'm looking at Q&A on this iPad, so if people have questions, please, this should be interactive. You'll get tired of us real quick. Please fill it out if you have some direct questions. Let's talk a little bit about some of the genetic issues that you've been involved. Polyomics, a big term today that combines a lot of the omics. Poly biomarkers that we get from our wrists or necklaces or our patches. How are we going to use some of that for personalized medicine to do the following? Identify the right person, consumer, at the right time for the right technology, leaning back on that curve like we talked about before. How is that, which you've been investigating in these omics and markers, going to bend that curve? Well, what we have right now is a really remarkable situation where we have this enormous body of knowledge that sits in research domain about genomics and a tiny trickle, if any, that goes into the practice of medicine. So a great example would be we have polygenic risk scores that have been validated ad nauseum for coronary disease. And who uses them? Well, you know, for example, in clinic, I was in clinic yesterday, and we're trying to decide this person, should they take a statin, you know, their LDL is not that high, but you know, they maybe have some risk factors, not much. And the person, the patient says, should I take a statin? We say, well, let's get your polygenic risk score, and in fact, our group has it out for free, my gene rank. And right then and there, within a couple of minutes, we upload the data, say, here it is, oh, your risk score is 97 out of 100, yeah, we should start a statin, you know. So we don't use this, it's very, I mean, I don't know how many cardiologists are using this today. Not many. Not many. That's widely available, and the validation that it helps to determine whether statins are useful, rather than just wasting it for a person's lifetime of taking statins. So I think this is the problem we have in general with genomics, is that for 20 years, we've been pounding out data, great papers, but we haven't used it in medical practice. and when are we going to be ready, if not now? When are we going to start to say, you know, it's... Now, another difference is, in this country versus, for example, the UK. The UK, its National Health Service has an education wing, which they put a lot of funds into educating clinicians, physicians, nurses, pharmacists. So they're not only the world leader in genomics, but they're also educating. Their whole medical community. We don't make that investment. So that may be one of the stumbling blocks, but right now, there's many things. I just gave that as an example, what we could do with the data that's in hand, which we don't use. Well, it's also going to probably not be, with all due respect, the healthcare people offering that the next time they have that same question. We all have that question from our friends, from our patients, about statins. But they're going to do it themselves. They're going to get their own risk profile and probably get good RX to give them the statin. So it's just not going to come down the same lane because that consumer that you talked to yesterday in clinic is going to tell four people, and that's more effective than you telling four docs. So I'm not trying to be disrespectful to our profession, but we have not practiced that bouquet, if you will. Let's talk a little bit about this sport that we're here today. There's a lot of ICDs that have been phenomenal, but a lot of them don't go off. In fact, the majority of them don't go off. Chronic persistent AFib, regardless of the mapping technology, regardless of the energy at the end of a catheter, don't solve the problem. They asymptotically creep up. But how do we get better on the patient-specific identification of the right thing for the right consumer, the right tool? I'm glad you brought up the ICD. This is actually, to me, one of the most striking things in cardiology today. How many are implanted, actually appropriately fire during a person's lifetime? What percent? Anybody? What percent of ICDs that are implanted appropriately are needed, appropriately fire in a patient's subsequent lifetime? 30 to 40. 50, I hear. Do I hear 60? 50%? Between five and 10%. Five and 10% of the ICDs appropriately fire, meaning that 90 to 95% of those that are put in by the guidelines are not needed. And then there's all the people that have sudden death that don't have an ICD who would benefit. Are we this stupid? Can't we do better than this? Now, I would submit we sure can. By the way, it takes me a little bit of time to get Eric going, but he's starting to go. So I slowly light the fuse, just to let you know. Yeah, well, no, thanks. But here's what the opportunity that lies ahead of us. Absolutely. So the Johns Hopkins group published about how you could do deep learning from the MRIs, and you could pick much better who is gonna really need the ICD. Then there's all these different lines of evidence. There's a paper recently published in JCC, which just using that coronary disease risk score, not even a sudden cardiac death risk score, which we should have, polygenic, you can markedly improve the discrimination of who would benefit from an ICD. And then there's just looking at the 12 lead with AI, trained with inputs. There's many ways that we could do better than having only five or 10% of people that have a hardware store put into their chest. It's pretty expensive too. And it might even misfire. Can't we do better than this? So that's the challenge I put out there. We have the ways to do it, but we don't have... And that takes, I think, the work of multidisciplinary work. It's not gonna just come. It has to come from the imaging, from the computer science teams, from the genomics group, all working together. And we don't tend to do that. Yeah. Do you think that each one of those, say polygenic scores, you need to randomize to an outcome? You're a trialist all your life. How do we take what you just said and make sure that there's a real world exhibit of that through trials? Yeah, well, I think if we could get samples from all the people at a board sudden death, we could get a specific sudden death risk score, but we haven't done that, which is amazing. It hasn't been done. But no, I think here you don't have to have randomized trials. You just have to have good prospective cohorts. And the other thing is this is another project that we initiated several years ago, which I don't understand is not done everywhere. Somebody young, let's say less than age 50 or less than 45 dies suddenly. There's no apparent cause. Why aren't the family members sequenced, no less the deceased, to do a molecular autopsy? Because through that, we can pick up at least half of the usually ion channel mutations. And this is cases where you've already ruled out aortic rupture or other anatomical reasons. We can then help the family members. Some of them do need an ICD, right? Some of them may need drugs. Some of them just need reassurance that they don't have to go the rest of their life thinking they're gonna drop dead every day. But we don't do that. And it's very inexpensive. It's cheaper to do a sequence now than it is to do a lot of blood tests that we order or scans. And it's gonna get cheaper and more available. The idea is that two and a half billion people are gonna have their sequence, their genetic sequence by 2030. So it is moving towards that. And I still think that it's gonna be private and the consumers that are gonna drive this because waiting for governmental bodies to take a risk I think is gonna be problematic. So I think that answer is somebody that talked about do we need a randomized trial? I'm just reading the questions here for some of this. And I think the answer is not equivocally no, but not for all of them. There's another issue that somebody talked about, data. So much data that it causes physician burnout, nurse burnout, healthcare workers burnout. The answer is yes. There is a little bit of that going on. The EMRs have become much more compatible with fire and HL7. But still, we have to make sure that we don't add to the burnout, especially on the heels of what's happened the last two years. But this is I think where the data come in to stitch our episodes of care together. We are lacking nurses. We are lacking in some cases doctors. You mentioned the hospitals. The hospitals in the last three decades have increased administrators 3,000% and only 12% docs. So we don't have it right yet, to say the least, but having data to help less skilled people take care of that hypertensive or be able to be more proactive in that care continuum I think is gonna be fortified. We can't do all this at once, but it's really important to at least embrace some of these changes. I gave a lecture the other day and somebody got up on the mic and said, you know, I don't want my patients going through all this consumer stuff. And somebody else got up and said, yeah, it's just too early. The way I practice care, and I said, are you guys done? Because they don't care. They'll go to someone else. So we have to change this treadmill that certainly we're around. So. Maybe make a comment about that. So this dealing with the data and also dealing with electronic health records is why we should be pressing on AI as a rescue. All right, so that was really the whole premise of deep medicine, which is, I mean, if you watch, for example, Microsoft buys Nuance, a light bulb should flash. Why would they do that, right? And why are there 25 other companies that are trying to take synthetic notes into mainstream practice? And those notes, by the way, if you look at them, they're phenomenal. Absolutely. And not only that, but the AI tools can review all the chart, everything in there, before you even have a chance to look at it, to help reduce the time. Our problem is, the one that you touched on, the overlords, okay, the administrators that run medicine in this country. So if we have AI and it is making us more efficient, what does that do for the care of our patients? Well, the overlords would say, oh, now that you have these ways to not have to use keyboards and that your voice is doing everything from the notes to the billing, to the orders, to the next visit, to the prescriptions, it's doing everything, you can see patients, more patients, and by the way, the radiologists, you can read more scans and the pathologists, you can look at more slides. No, we have to take medicine back. So here you have, on the one hand, AI that could be making our lives so much better and never having to deal with the data and the electronic health records as we know it today, but we then face the residual problem of that this is a business and that it's not run by doctors or clinicians, it's run by people that are trying to make it more efficient. So we have to get back the gift of time in order for medicine to be successful going forward. The guy that was the CEO, probably some of you know him, of Stanford, was there for many years and he decided, rather than having me wait till they come to me and working on my negative margins, I'm gonna go out and start something called OneMed so that the consumers can make it convenient for their care. They'll do it much easier. And then when an Amazon buys them and they also own nutraceuticals and vitamins and whole food stores, think about the data that are gonna be able to be drived from that interaction and how you can then get better for that particular consumer. For those that listen to Elon Musk, it's hard not to listen to him today, but he watches too many science fiction movies. He scares people about AI. Let me turn it around a little bit, and I was just talking to the FDA yesterday about this. I like to call it IA, intelligent augmentation. You're not gonna take the healthcare providers out of it. You're just gonna augment them with issues that might help them in their decision process for taking care of hypertensive patients, for taking care of liver disease. Because a 30-year-old doesn't have the machine learning that he has in his head over the last 40 years, but we have ways to make that 30-year-old a little bit smarter or suggest without telling them what to do, we augment them. That may be one way to think about it rather than listen to Musk say that we're all gonna be taken over by AI. 300,000 apps in healthcare. Which ones do we tell our consumers, our patients, to use? This is where it gets a little bit tricky because there's a lot of apps and there's a lot of information out there and we live in a misinformation world. So how do you suggest that industry, healthcare professionals, and the consumers use those millions of apps that are out there, hundreds of thousands of apps? I mean, yeah, unfortunately, a very small fraction are useful. Of course. Very small, but it really depends on the situation. But most people with hypertension would benefit by having an app that puts all their measurements that they take onto their phone on one screenshot and they can be sent to a physician and help manage their blood pressure better. But usually they wind up managing themselves better just by looking at their own data, which is great. But what we're seeing, and this is gonna happen more, the first, of course, the front runner was the diagnosis of atrial fibrillation and other arrhythmias. But soon that will be added, the apps for diagnosing skin lesions and rashes across the board, ear infections in children, urinary tract infections, and basically the whole list of things. So if you wanna call it screening, but some people would call it actually a diagnosis, it will be the initial assessment. And so that will decompress a lot of common visits for physicians and it's great. There are also the other part of the AI world, which is really getting more integrated with apps, is the coaching. So this is being now becoming much more popular in people with both type one, type two diabetes, depression, hypertension, and other chronic conditions. So the idea would be having a chat bot support. There have been randomized trials. Some of these are quite good despite John Oliver's recent takedown at least one of these chat bot companies, but also humans in the loop so that if the chat bot isn't helping in a particular coaching, there's people to help. So that's an app that's gonna become much bigger across lots of different chronic conditions or prevention in the years ahead with your data. So the ones that we had up until recently are pretty rudimentary, but they're going into, I think, a more refined, more individualized mode. People worry about data being either plagiarized or stolen. Who owns the data ultimately? Well, that's been a pet peeve of mine for years and it was really the center part of the patient will see you now. And that is people should own all their data. It's your body and there's 27 reasons besides the fact that you paid for it and it may save your life that everyone should have their own data or for parents, they should have the child's data because no one essentially has all their data from womb to the present time because you have many different providers and places that you've lived and on and on. And that data is, even with a portal, whatever you can access is only scratching the surface. Your scans, if you try to get your actual scans and your path slides, good luck. It's not easy. And you beg and grovel and it takes time and it's hassles. So that data should be owned by the person. And so someday other countries that are already done this, like the Estonian model, which is now being mimicked through other countries in Europe, we will get there because once things are digital and eminently portable, that should be a civil right. And that will help because for the AI future, you need all these inputs to make the output as good as it can be. So data ownership is a big deal, but we are compromised here. And I think security is a little bit over dramatized. I mean, I work with this gentleman who runs cloud computing at Amazon. And he does the security for many industries that are much more in need of security or at least real security, such as the, you know, EV, banking. And so we don't necessarily need to worry about this. Well, but now I think there's an issue here. So every year now, tens of millions of Americans are getting their data, medical data hacked. And this is horrendous because- 40% of healthcare. Yeah, yeah. Has been hacked. So, and why? Because cyber thievery, this is so profitable. It's at least 5X what your financial, personal data is worth using it as Medicare fraud and opioid prescription and identity theft and everything else. So yeah, we don't want this data to be subject to this hacking. However, if the data is at units of one or family units, it's not going to be a target for hacking. And the other thing that's really exciting right now is privacy computing is going into high gear, not just federated learning and, you know, the various ways of- Computing at the edge. That's the key, what you're talking about, Peter. So the fact that your phone will never lose, the data will never leave your phone because all the computing would be done in the phone, edge computing. And that would help. And then of course, you know, the countries that have made this shift to people owning their data have set up secure platforms. So between these different modes of, you know, the far better ways to ensure privacy, we can do markedly better so that that won't be a worry, that won't be a stumbling block. So just to be clear, you take the data, you put it in the cloud, you combine it all together and you look for insights, that's vulnerable. To do homomorphic encryption of your data and then do that, that can probably be encrypted. But to keep it at the edge, keep it on your cell phone or on your computer and just take the insights out and put that together, that can be a learned issue that preserves security, yet uses insights from the information to get into real world directives. And that I think, it's like those that know about Waze that started in Israel. They didn't care what you were driving, they didn't care who was in the car, they didn't care where you stopped. They just took the insights from that route and now scaled on Google. So that federated learning that Eric's mentioned is actually one important kernel to keeping us independent with our data and to keeping it secure. We'll just end and I didn't get to all the questions, so maybe you can go see him afterwards. Advice, Eric. There's folks out there at varying ages in their career, some are just starting out. Young scientists, mature scientists, innovators. Is this an exciting time to get into medicine? It seems a little bit confusing right now because it's not the standard track that we all trained in. What would be your suggestion, your advice for the young scientists, the young healthcare worker, the innovator in the next 10 years? What's your advice to them? Well, I mean, I think it is by far the most exciting time. You can see all these things that we've been talking about today have transformative potential, but there's a lot of work to be done to get there. And we're still in the very early phases of implementing AI into medicine, no less bringing these different disciplines together, this kind of whole multimodal approach. So it couldn't be more exciting. I wish I could go backwards and be part of it, but the young folks here have opportunities galore like we'll never have. I think the key here is to not work on small stuff because when you write a paper and the only people that read it are your parents, and they can't even understand it, that doesn't even get you anywhere. So if you look at all the titles of papers that are published in a lot of journals, you say, WTF, what are they doing here? So my advice is swing for the fences, go for big things. You may fail, but you know what? Those are what is worth it. To just work on things that are relatively arcane, it's unlikely to get your passion up. But moreover, even if it does, it's not gonna move the needle. And then we got a lot of needles that need to be moved here. So that's the main thing. No room for small dreams, as Shimon Perez would always say. And I think it's important though to be open to, we've never had the eclectic discipline that is merging into medicine, ever before. We didn't have these tools. So this is not only a little bit confusing and things are a little bit disorganized, but on the edge of that chaos is where big things happen. And that's what I think is a real important message to the folks in the audience. And what else do you wanna talk about? I'm done. No, it's great to be here and it's a really interesting group and mission. I think this HRX and the whole theme of innovation is really important. Hope y'all can keep it up. It's great. I wanna appreciate the organizers for putting together, you know, a different meeting. I'm not used to somebody sitting behind me seeing my bald head. You know, I'm used to having control behind a podium. But I think this integration to different folks is really important. So hats off to the organizers and let's give them a hand. Thank you.

Peter

Eric Topol

fireside chat

medicine

healthcare innovation

digital health

AI technology

personalized medicine