Latest / Targeted Talks / S7 Ep8: Precision Hope: The New Frontier of AI-Driven Oncology

Transcript

S7 Ep8: Precision Hope: The New Frontier of AI-Driven Oncology · Targeted Talks

• Mehta Flora AI - February 23, 2026 00:00:41 Speaker: My name is Doug Flora. I'm a breast oncologist working just outside of Cincinnati, Ohio. Have been here for a couple decades. I run a bunch of cancer centers with my partner, trying scale smarter cancer care. And so I'm excited to talk with you guys. You're working on the same things. Great, great. And, we're going to be talking about AI and oncology today. And as both of us are actually breast oncologists up until a few years ago, AI, I think we know what that meant for us, which was aromatase And now all of a sudden, AI means something very different in our world. So, I'd like to get into actually just give me a sixty to ninety second. How did you get interested in this space of AI coming from oncology, which is a, you know, really kind of, conventional field of medicine where sort of a little more conservative in our approach and AI and the changes are really kind of rapidly, impacting care. So curious how you how you got into it. I took an administrative role. So I still see patients. but, four days a week. I'm an executive medical director and was starting to see problems emerging that I didn't know how to tackle. we don't have enough staff. As you know, the number of cancer patients is doubling by twenty thirty. We've already got almost nineteen million survivors, and there's only fourteen thousand of us. So that doesn't really math, at the same time, we're going to have about a million nurse shortage, in the same time period. And, I started looking for tech solutions to extend us, so that I could replace the mundane with the humane. And, instead of typing in the room with my patient to do my notes, could you use ambient scribes, instead of, fighting with peer to peers or prior auth or rev cycle management to get drugs approved or paid for? Could we automate, could we be smarter about cancer screening, etc.? And so I started reading heavily about it. I'm sort of a nerd, and it doesn't matter if I want to learn chess during Covid or how to be a bartender or whatever, I just like to study. I still study quite a bit. And, AI turned out to be the thing, and the more I got into it, the bigger I realized it was going to be. you wrote this book, Rebooting Cancer Care, which I have now, and I've started to read and, already, very engaging the whole the way you started with this control alt delete the concept of rebooting, which I think people, from our generation remembering the, the early days of the internet and, it is a, I think a really apt analogy for what we need. but how do you think about sort of broadly speaking, rebooting cancer care? I like the imagery for it, for me to understand. So the way I approach the book is, you know, I feel like cancer care is groaning under its own weight. Right now, we have one hundred and forty four thousand new articles indexed to oncology and PubMed last year. you can't possibly keep up. the the number of molecular markers that emerge, the complexity of care, the number of drugs available. I felt like we're working in like a nineteen ninety five windows version of healthcare, and all these tools are right there, ready. And because we're such a conservative specialty, we've been relatively slow to adopt. And so I took on that idea of rebooting as saying, we have to approach medicine fundamentally like our predecessors did back in the old days. You know, first principles of what does this patient need? What are they coming to my office for and not what do I need to get out of this visit? And to do that, there had to be some sort of transition of time to make sure that the tools augmented the physician to do those things again, with curiosity, we've lost a little bit of that. It's become more transactional than at any time in my career. And I know that you were around back in the day to, go back to the the way life used to be when I would come home and carry a case of giant thick paper charts that dictate into a machine, and we can do so much better. And there's still a lot of resistance. I remember I was telling one of the residents earlier this week that, there was a day I still had the record, and I was in a community practice early on. And I think, I saw thirty nine patients in one day. I dictated every patient still went to the gym and still was home in time for dinner. And to think even half those number of patients today, to do those, it would be impossible. it's a completely different world. And obviously documentation has changed. Payers have changed. I can remember an attending physician doctor, bless him, he was a wonderful, wonderful man, teaching me when I was a fellow. And I would be in his lymphoma leukemia clinic on Mondays. And I'm not kidding. Like the charts I would pick up from him, from his life of practice in private practice at the time would be like patient, stable, secret. And that was his entire progress note for the visit. That's great. I've got this nine page diatribe now with, you know, no wonder you could see thirty nine patients. We didn't document anything back then. That's right. It was a few lines of, which were relevant, but it was a few lines. Right? oncology I think when we think about oncology, obviously we're medical oncologists. there's radiation oncology, surgical oncology. I'm curious. And I haven't gotten, far into your book, but what parts of oncology do you think, even in the subspecialty of medical oncology, from things you mentioned, like drugs and, practice workflows, but also, radiation or surgery or anything. What parts do you think are kind of the most ripe for, AI to, be implemented? maybe this year or next year in this sort of short timeline? you work in this field, you know, right now tools like yours, like Remi, tools that can extend my nurse, navigators that can meet patients in their phone to keep them out of the hospital, to do symptom control and IT pros and those sorts of things, those are launched. Those are ready. obviously, you know that better than anyone. I'd say on the other side, the back of the store is robust. So the functionality of things that do rev cycle management, that manage our prior authorization processes, that do documentation for doctors and nurses, those are already pretty slick. And I feel like we can approach those as healthcare systems, as practices with lower trepidation, because it's not clinical decision support. that's where I get nervous that the tools are still somewhat immature. I think there's work to be done on some of those. There are some very, very good tools. but really want to have a handle on that as a, doctor before we roll that out full force, is it fully explainable? What was it trained on? Does it represent the patients in my charge. And so we've leaned more into the image recognition. Slide recognition, ctDNA frame on the clinical side. but a lot of the business of intelligence business of oncology is ripe. something you mentioned these different tools from I'm curious in your practice then, different practices have have adopted this differently. I spoke to, a doc last night. I was at one of those San Antonio reviews, and there was a doc I hadn't seen in a while. He's at a practice at the Jersey shore, and, even the word ai, he said, ah, I don't do all this AI stuff, you know, so clearly he hasn't adopted any of it. But I'm curious, in your practice, when did you first start to kind of have some of these tools and then, what has escalated in the past, x number of years that you started doing this. I would say we're a big at Saint Elizabeth. We're a big screening center. It was one of my highest priorities when I took over in my role. So we built what I think is one of the top lung cancer screening programs in the country. we've done almost sixty thousand scans. at one point, we identified twenty three stage ones in a row, and we have leaned heavily into image recognition software for that. And so that was probably one of our first forays into this, was having AI pattern recognizers overread our scans with the radiologists again augmenting, not replacing, to make sure that we detected every single suspicious nodule, every incidentaloma, every cardiac calcification. And those have really borne great fruit, the number of lives that have been saved. And every week I get a list. our director of lung cancer screening sends personal notes to the referring physician saying your persistence paid off. You know, Sally Ann, on her fifth scan, was found to have a stage one a and has an excellent chance of being cured because you cared enough to bring that up. And Sally Ann's films were overread by AI. Use the same thing in mammography. it's kind of a red yellow green box goes around areas of suspicion or comfort for our reading mammography. we've got a few people using GI genius for colonoscopy, and, I haven't seen the tool myself, but the Endoscopists are quite pleased with how it will say this area has something that's suspicious to me as a pattern recognizer that might escape the human eye, and it ought to be biopsied, and many more. we've leaned in pretty heavily and we are continuing to do so. but those are good early examples of things that we started five or six years ago. that's fascinating. I want to just press on something you said here, this idea of the augmentation of physicians. like you said, I think in radiology, GI I think pathology. You mentioned another place. I'm curious if you've heard from those physicians, are they in any way fearful of their roles and the potential, maybe diminution of of some of their roles, or are they really excited about the ability to kind of just be better clinicians and better physicians? I'd say, the majority of the devices that have been approved by the FDA are are on the radiology side. So their journals and their meetings have been very AI heavy for some time. I'm sure there were some more established legacy sixty year old near-retirees that probably didn't welcome it. but I'd say this current generation, it's think about the stress that you're under if you're reading a microscope slide or one hundred and twenty CT scans a day and you can't ever be wrong. And so I think they've welcomed the backup. And the systems are supposed to really work in the background and not interrupt flow. So it doesn't make them less efficient, if anything. it drives new discovery. And we've picked up a ton of disease in our center that used to be missed. especially things like cardiac calcifications. We may be looking for lung cancer nodule. but now we're getting a calcium score, and our triage sends that patient directly to Heart and Vascular Institute. So it's actually a strength for the institution that we're trying to prevent disease rather than repair it. which is exciting for the physicians. And I wonder, you know, patients. there's this, broad issue that I think every oncologist, maybe every physician in the country is facing right now, which is information overload, misinformation, disinformation, whatever the the terminology is. I wonder if patients have had any fear, skepticism. You know, what's this technology? privacy concerns. Have you had any of those? What's your take on how patients view this progress? Yeah, I think we're in the the first inning of, of this baseball game. And, most of the surveys that are published suggest that patients are concerned, they have continually expressed that they want their physician making decisions. the surveys are pretty consistent that, they're nervous about bias and how the sets are trained because it's still sort of a black box for most of these things. Many of the designers of the AI tools don't even know how they come to decisions, and that's not going to work long term for us in medical decision making. But I will say, man, we've been here before. And, you know, you mentioned my book. I tried to use historical context in each chapter. I write a ton on Substack the same way to say, all right, this isn't the first time we've had this conversation. in ancient Greece, Socrates, was worried that putting things down into books would mean that our memories failed when you and I were carrying around our Washington manual spiral bound thing and eventually on our Palmpilot and BlackBerry, everybody's like, oh, you can't. That peripheral brain is cheating. You know, you're supposed to memorize the fifteen causes of pancreatitis to be a doctor. And I would argue that's dumb. That's just silly. Now we have all of the experience of all the other physicians who've ever lived, and information is immediately available, whether you work in Paducah, Kentucky or Sloan-Kettering. And I think that's a great equalizer that, all patients win there. but in the historical context and there's actually there's a really, really good article in this week's New England Journal of Medicine written by a medical historian talking about how reluctant we have been throughout medical history as doctors to even accept things like that. I'm writing a piece for Sunday for my Substack that's talking about the original thermometer. and it was a foot long thing. And the history of it is fascinating. The physicians were absolutely diametrically opposed to the idea of a thermometer making a diagnosis of a patient because their hand was trained and they thought that it was very, very important that a physician earned the skill to put your hand on a patient and feel how clammy they were and the texture of their skin, and guess the temperature to make a diagnosis. And only a trained physician could do that. And that's just silliness. And so history is rife with stories like that where we push back and said, no, no, we don't want to loom to help make our silk. And then silk got better and and so on. And so in medicine, I think we are slower adopters. But that's part of the reason I'm trying to lead from the front and say, you guys, you've done this a hundred times. You're just not aware that you did this before. But it's the same exact concept. New technology arrives. We can now see inside the human body with a Cat scan. That's progress. And so on. Yeah. I love pulling in those historical ideas. one of my favorite reads, I tell every fellow to read is The Emperor of All Maladies. the book, and he talks about a story of the surgeons during William Halsted's era of the early nineteen hundreds, where they would walk around with a open scalpel tied to their scrub pants. It would fall on the floor. They would kind of wipe it off, put it back, and that's the scalpel they would use. God forbid anyone tell them to use clean tools. we are reluctant to change. And, you know, AI is certainly a big change. you mentioned this issue of do we trust what's going on. Do we trust the technology. how do you trust it. what is it that drives clinicians to trust AI. And what could we be doing or what could AI, companies be doing to try to have us trust the technology? Yeah, that's that's probably the crux of my last two years is, as physicians, we have to have a different level of, of proof before we can accept things. And so I started this journal, AI in Precision oncology about two and a half years ago, to make sure there was a forum for these discussions, to make sure that we published important articles that survive peer review. because I think that's where we start is we really like peer reviewed studies. and so that was one, two, the idea of the book, and I probably speak somewhere about this fifteen weekends a year now, twenty weekends a year, is really having people that are in the know curate the content and make sure that those of us that are a few years ahead communicate clearly the why it's important, and how much our patients and our teams can benefit from, accepting these tools as a crutch, as an assistant, as an extender and I guess where I've fallen down and what I would credit you guys at Reimagined care. And again, this is not a sales pitch by the doc, but your tool is exactly what I talk about, which is meet the patients where they are. We all know it's in their phones. try and prevent calamity. Try and use intelligent digital tools, to get in the way of these diseases and prevent things like unwarranted emergency visits or hospitalizations that we say we don't want to happen, especially the repeat hospitalizations within thirty days. And so a lot of the people are trying to solve these problems, don't understand that there are really good validated solutions already available. So I'm out there banging the drum, whether it's writing or public speaking or keynotes or whatever. Things like this, just to get people curious enough that maybe they pick up an article, maybe they pick up a book, maybe they watch a YouTube video or listen to a podcast. But, this stuff isn't going away. the idea of this, you mentioned sort of the tools that help us as clinicians. This idea of burden is something I think about a lot. I was a full time, ecologist, four days a week. I was seeing sixty patients a week, and I was burning out. And, there's this idea of burnout in healthcare is gaining attention, and particularly in oncology, I think it was ten years ago, there was a survey that said oncologists were out of the thirty. Some specialties were like twenty eight in burnout. And about a year ago we were number two. We were tied for number two. And so clearly something has happened. And you did mention, there's a lot of new drugs and a lot of more patients and there's this burden of care and, this technology sort of has I think some clinicians look at it, one way, others look at it a different way. One is that it can help us. Others look at it as, is this going to increase my workload and how have you incorporated or how do you see incorporating into systems where you might see more resistance and more skepticism to kind of make sure that the clinicians are not going to have more burden to their day. It's a super important point. I'm actually really glad you brought it up, because I think that is where the builders need to listen. so people out there that are designing tools for doctors, some hints. Number one, more clicks. Bad, right? the doctors are overwhelmed. They need to have something that's working in the background. They need to have, something that doesn't interrupt their flow because every minute of their day is scripted and we wake up an hour or two behind. Every day there's thirteen hours worth of work to do in a twelve hour day religiously, and it never, ever goes away. And so I'd say for the builders, the people who are doing this get physicians involved, get clinicians involved, nurse practitioners, nurses, administrators, whomever you're planning to try and develop solutions for. Have them help you in product design. Because I'm approached probably five times a week by a company thinking that I'm easy prey because I run a bunch of cancer centers and I'm this AI advocate saying it's time. and then I hear pitch after pitch after pitch, and they're not solving the right problems, and they're not speaking the language that the clinicians are going to need. And the user interface is clumsy. And, the the programs, the projects are just destined to fail because they didn't understand the voice of the consumer. They're just like in regular manufacturing or design. And so I would say for those of you that are in those spaces, get doctors on board immediately. and ask them really what they want and not what you think they want. Because there's a vast difference between the two. That's really great advice. I think, given I've learned a little bit about this oncology startup space. And you're absolutely right. I think the companies that not only have clinicians on board, but have different levels of different types of clinicians. You have physicians who have different, patient care responsibilities and needs. And then you have nurse practitioners or advanced practice providers. You have nurses that have different, needs. They're all burned out. And, you're right, I think getting all of their input, when we're designing these models is important. I wanted to kind of drill down a little bit now into your sort of day to day, like the clinical workflows, most oncologists, I mean, we'll see anywhere from in a full day, maybe fifteen to maybe as much as thirty. although I don't know if they're doing that these days, just in a, clinical workflow, where do you see AI on a Monday morning when that doc walks in the office? Give me like a couple things, if not more, that you think would help that that physician that day. Well, I mean let's just look real time today, right? The things that are presently available, you can arrive at the office and instead of going through sixty pages of data, the PDFs that are scanned in the media tab for their next generation sequencing, all the other stuff, the note can be pre-generated and pull in everything that happened during their last hospitalization. an AI generated discharge summaries available that has all of the data that you need. Everything's pulled into one note before you sit down in front of that patient at eight fifteen. when you're in the room, the ambient recording software, listens to everything you say and is smart enough to understand that the talk about how was your vacation in California doesn't make it into the progress note, but the talk about the radiating back pain to the left lower extremity does. So you walk out of the room and the note is generated. The bill is ready to drop, and it's appropriately coded for the level of work that you did when you often under Bill and don't know that you're under billing. And then, you move on to write the chemotherapy protocol and, patients going upstairs and the labs are checked, notes already been generated by the electronic medical record to say, Mr. Smith, I pulled in a prescription for potassium. I want you to take twenty milliequivalents tomorrow. and for the next week, and then come back Monday for a recheck. best practice advisory pops up to say you don't want to give this drug in this dose, because we've detected the pharmacogenomics say that she has an enzyme deficiency in DPD? And you need to do a fifty percent dose reduction for your five Fu. Or you can hurt your patient and so on and so on and so on. And we're using these tools in clinics today. And so the people that are out there that don't know these things exist. That's who I'm trying to get to and let them know that there is help. It's ready. And it's just a matter of getting curious enough to learn about it, just like we did immunotherapy, just like we did genomics and, and genomic health. You have to be a lifelong learner in oncology because things are moving quickly. it's amazing how futuristic that that concept sounds. And yet you're right, it's kind of today. It's actually here today that we're we're doing this. And you know you wonder sort of the idea of kind of we know that oncology is unique. And I mean, obviously we're a little biased, but oncology is unique in the the importance of the human role. And, there are certain fields that, technology probably plays a much bigger role. But one of the concerns that I know clinicians have that that the ecosystem in general has is will technology, be replacing the human role. And I'm curious how you think of kind of ensuring that the human connection, which I know you write a lot about. I loved some of your pieces on that, And I think about a lot in my day to day. where do you see it kind of strengthening that human connection. And, do you think in some ways there's a potential for, weakening that link? I think that's my whole point is the current system is not tenable. we're further away from our patients than we've ever been. We're more rushed. We're more stressed. We're more burnt out. And oh, by the way, we're going to double their number. And so the idea that the status quo is this wonderful thing is, a false start anyway. the whole book it's called can AI Make Cancer Care More Human again? And the final couple chapters of the actual rebooting get into what would happen if this all goes right? And what if you could get two hours back a day and never have to type an email again, because the agent knows your voice and can craft the response for you to edit thoughtfully and then hit send. what if every correspondence that you had to send to a patient about an abnormal lab or a test result was automated? What if your note was generated for you? What if you could look the patient in the eye in the room and see that she's getting hives on her chest? Or that her husband's getting vagal because you're sharing too much information at once and you need to back off and you've been in the room, you've seen that visceral response sometimes where you have to actually send the the AMA to get water for the husband, right? We don't see those things anymore because we're looking at our computer screens and you've lost the art of oncology. So if anything, I'd say oncologists should be leaning into this because our primary goal in life is to ease suffering and care for their burden. That's what onkos means. It's a Greek word for burden. And we're sharing patient burdens, and we're not doing it in the manner in which we were trained. We were not doing it in a way that's as rewarding as it used to be, because it's become transactional and we don't have the moments in our day to love on our patients like we're supposed to. And so if these technologies can remove two or three hours of the mundane, letting you be the doctor you swore you were going to be when you were a third year med student on rotations, I think the patients win. I think the nurses win. I think the doctors win. You're right. going back to this issue of like trust and, disinformation. I found that One of the clear ways to to improve that is to really sit there, have the computer, you know, the laptop or the desktop aside, sitting with the patient. I think a lot of us, with the notes that we have to do in all these other clicking that we have to do sometimes forget that. So I think it's an important lesson for everyone. and this, future of this hybrid care. I think about that a lot. at Reimagine Care where, that's our central tenet has always been that we're not technology. We're not just human. we're really a human company that technology can enable. And the future of this kind of model is, it has to be both. We have to, kind of allow both to be at their, optimal levels for patient care to be successful. What you guys are building. What you have built also solves a lot of my problems in the clinic because the support is available to patients right when they need it. Right? And it provides me with an ability to track them in real time. I think it's another anachronism that I give you your chemotherapy. And then twenty one days later, I check in and see. What did I do to you? Right. And here we are walking around with rings and Garmin watches and Apple Watches that are tracking data all day, every day for us. and we're not reliant upon that. We don't use any of the things that happened in that twenty one days before. So I think as we move things like Bispecifics and Car-T cell therapies into the community, you have to have a means by which to check on that patient in four hours on day two and eight hours on day four, and so on. And so tools like Remi and others are helping deliver care for that twenty one days in between. and again, as a hospital administrator, I like that tools can start to, maybe identify sepsis for days before so they don't have a life threatening infection. Eight hours in the E.R. and an expensive hospitalization. as a doctor, man, I love getting a heads up that my patient is suffering out there. maybe day ten on xeloda. And so I know that they're hurting so I can stop the drug and not give them four more days of toxicity. As you know that it's going to get worse after they stop the drug anyway. They still got three or four more days of dosing that hasn't hit them. So I'd love to identify those things. Day eight or day ten rather than day twenty one when I find out that I gave them grade three toxicities. right. Yeah, I can imagine, those every three week visits, even where if a patient shows up, we now have this, AI generated, wearables generated, virtual assistant generated, summary of a day by day or a kind of toxicity graph or it'd be really interesting to have I don't know if that exists right now. I'd love to see companies like yours start to click together with some of these other companies, like Lego blocks. There's a company that makes a device called Temp Shield. the CEO Sam brand is just fantastic. And you stick it on a patient's chest, it's a twenty four hour a day. I think it's one hundred hours of battery charge. temperature check. And, like you guys, there's a clinician on the other end. and so if you spike a temp at, three in the morning and you're not aware that's recorded, that's sent to the medical team, and antibiotics and cultures can be done that next morning instead of in the ER three days later. and so when you have things that intervene and interact with the patients directly through their phone or through their iPads or whatever, tools like yours are even better when you've got objective data to not just sort of responses, you know, how do you feel? Are you tired? Do have any sores in your mouth or whatever? But what if you could integrate with tools like that that the wearables fed your tool. You're the medical officer at your company. Think about the power of objective data flowing in in a continuous pulse all day, every day, and what that could do for triage and prevention and early detection of danger. Yeah, right. I mean, we've tried to especially now, when we think of it, you mentioned these complex therapies like that's something where, I think the country is focused on these newer therapies, not not only because they're really complicated, but because they're also really effective. And, right now, it seems like many of them can only be done in these large centers and academic centers. And I think, there's an unfairness to that, they should be allowed to happen in any oncologist's, purview safely. And you're right, I think these sorts of, tools are necessary, for that the the wearables. One of the things I've realized is that wearables docs don't want more stuff for them. So what you mentioned is really important that information flows to a team that is then managing that patient. Because I think I can't imagine an oncologist saying, sure, you check the temp and the pulse ox, but make sure I get that info at two in the morning. because that's certainly not conducive to their quality of life. But the hybrid, approach is really, really evolving. I'm sure given your interest in history and some of the literature in healthcare, Yuval Noah Harari work, tremendous author and speaker. I think his second book was Homo Deus, where he spoke about his version of healthcare. and this was before AI was, adopted and a lot of that section actually today, medical education and the role how a patient interacts and all of that. So I'm curious if you can play the Yuval Noah Harari here a little bit. give me a I'm going to say one year, five years, ten years. what does this look like? Well, number one, I'm an optimist and I'm probably five to eight years ahead of most people right now because I have the good fortune of a job that lets me read and study. I get to see the journal articles before they're published. I get to sit next to some of the smartest people in the world at all these conferences from OpenAI or Nvidia or Google DeepMind or whatever. And I can tell you that the future is very, very bright. So this is going to be a much bigger deal than the arrival of the interweb, the arrival of cell phones, the arrival of computers. because these tools train themselves. And I think that people don't necessarily understand the difference yet. You don't have to train them anymore on these giant corpus of data. You have to give them the rules, and they can figure out how to play go without memorizing the two hundred million games of go that were played before them. And so it's a completely new paradigm. So one year from now, I think that all the early adopters will be plugged in, that two point five percent of people like me that love toys, that love new tech, will be using these things and publishing heavily on the utility, the outcomes, the changes in mortality, the earlier detection with ctDNA, when those tools finally get good enough to be sensitive and specific. and I think we're about a year away from real use cases appearing in regularity in the medical literature from places like Texas Oncology that are way ahead, that have really, really invested in a digital infrastructure. they've got one hundred virtual care nurse navigators, that are managing triage now. they've got E pros everywhere, and they've published heavily on the success of their Epro program, using digital tools to prevent unwarranted hospitalizations and complications. So fast forward three to five years. I think it's table stakes. I don't think that there will be a patient screen for a clinical trial that's not done with AI. I don't think Crcs are going to be going through charts trying to figure out, is this patient eligible? I think clinical trial conduct and design will be completely different because AI and drug discovery will be here. And tools like DeepMind that won the Nobel Prize last October will be commonplace. And there won't be a pharmaceutical company designing drugs at wet benches anymore. It'll be done through digital twins in silico, and drugs will get to market in half the time and thirty percent of the cost. and we're going to love that as doctors because you get better options. And I think by ten years you're going to have humanoid robots doing the majority of surgeries. You're going to have humanoid robots in our lives, in our daily laundry, folding, cooking, cleaning, unloading the dishwasher for thirty thousand dollars, and the mind blowing things that come with that are that you're also going to have digital assistants in the office. And so we joke about the peripheral brain in the Washington Manual earlier, or in that article in the New England Journal of Medicine. AI I'm quite confident that these tools will be able to intervene and prevent calamity, prevent user error by doctors through process that protects us from the things that we can't possibly know. And that I'll go one step further for purely inflammatory purposes. I think in patients who have access to screening, we will eliminate stage three and stage four cancers by twenty fifty. How's that for a bold close? that is a tremendous goal. And I love the way you outlined that, because that you're right. and I think one of the sort of the full circle moment feels like all of this technology, all of these advances ultimately maybe allow us to be more human and have more time with our humanity and given, everything going on in the world, I think it's something that's really necessary. So, I'm hoping all of what you said comes true, and given what I'm reading and hearing, it certainly sounds like it will. It's all about when, maybe sooner than we think. So just described the last chapter of the book, which is a very, very hopeful look forward at what if it all goes right? we need optimism these days. We'll talk about quantum computing at some point now that, that will play into it. But, uh, the next book. Well, Thank you again. This was, really great conversation. Like I said, I could talk to you forever on these things. Appreciate being invited.