About the Episode
Join us on HealthTech Remedy as we delve into the groundbreaking work of Atropos Health, a company transforming healthcare by making real-world evidence (RWE) accessible to physicians at the point of care. In this insightful interview, Dr. Timothy Showalter and Dr. Paul Gerrard speak with Dr. Brigham Hyde, Atropos Health's co-founder and CEO, to uncover how they're addressing one of medicine's most critical challenges: the pervasive lack of timely, patient-specific evidence for clinical decisions.
Approximately 90% of medical decisions currently lack high-level evidence, leading to potential variability and suboptimal care. Atropos Health provides a revolutionary solution by enabling clinicians to ask natural language questions and receive rapid, rigorous analysis of de-identified electronic health record (EHR) data – delivering insights in days, not months.
Discover how Atropos Health's origins from Stanford University's "Green Button" initiative laid the foundation for their unprecedented speed and accuracy. Learn about their innovative Geneva OS, a full operating system for real-world evidence generation, and the Atropos Evidence Network, which leverages federated data across multiple health systems while ensuring unwavering patient privacy.
Dr. Hyde shares how his extensive background in health data and companies like ConcertAI and Eversana shaped Atropos Health's mission to empower physicians and accelerate research for life sciences companies. We discuss the increasing acceptance of real-world evidence by regulatory bodies like the FDA and CMS, signaling a bright future for data-driven personalized medicine.
This conversation offers powerful insights for anyone interested in health technology, medical innovation, and the future of patient care.
Transcript
Dr. Timothy Showalter:
[0:00] Hey, Paul. Happy Friday.
Dr. Paul Gerrard:
[0:02] Happy Friday, Tim. How are you?
Dr. Timothy Showalter:
[0:04] Doing well. I was in Detroit this week and Springfield, Missouri. So finally back in Charlottesville, Virginia, and just in time for the weekend. How about yourself?
Dr. Paul Gerrard:
[0:16] That sounds like a busy week. I've been in Boston part of the week and now back home in Maine.
Dr. Timothy Showalter:
[0:22] Very nice. We do not have Trevor with us today, so I wish I knew a really embarrassing story about him. Do you know any? We could make one up.
Dr. Paul Gerrard:
[0:30] You know, if it weren't a Friday afternoon, I'd probably have enough energy and brain cells too, but I'm afraid I don't.
Dr. Timothy Showalter:
[0:36] Well, maybe what we can do is just like add it to the show notes. Just track down, either fabricate, like we'll ChatGPT a story. So what phrase should we go with this time?
Dr. Paul Gerrard:
[0:47] I should have captured the list of phrases.
Dr. Timothy Showalter:
[0:49] Okay, we'll go with that. That's fine. We should have captured the list of phrases. All right. Welcome to Health Tech Remedy, the show where three physician leaders in health technology tell the stories of new and established companies and interview leaders from the industry. I'm Tim Showalter, a radiation oncologist and prior medical device entrepreneur who is now focused on bringing AI advances to cancer patients.
Dr. Paul Gerrard:
[1:12] I'm Paul Gerrard. I started off as a physical medicine and rehabilitation physician before focusing on reimbursement policy, molecular diagnostics, and market access for AI products. Notably absent on the roll call today is our third amigo, Trevor Royce, a radiation oncologist and researcher with experience in real-world evidence, informatics, and AI diagnostics.
Dr. Timothy Showalter:
[1:30] We will miss you for this conversation, Trevor, but more power to you serving cancer patients and then coming home to four young kids. So totally respect that. This week, we're talking about Atropos Health. It's a very interesting company that's changing the game and how physicians access real-world evidence to guide clinical decision-making.
Dr. Paul Gerrard:
[1:50] Then we'll speak with Dr. Brigham Hyde, the co-founder and CEO of Atropos Health, about the company's origins, its mission, and how they're using data to bring insights to the point of care.
Dr. Timothy Showalter:
[1:59] So I'm excited about this. It's in the real-world evidence space, but definitely an applied clinical decision story that's like fascinating. And I will tell you, I didn't really think this was possible when thinking about what are the leading edges of real-world evidence. So in the general space, they're focused on taking this sort of evidence that exists in electronic medical records and other data sources to inform clinical decisions for physicians in the cases where clinical trial data don't exist. Right.
Dr. Timothy Showalter:
[2:35] We all know as physicians, that's most of medicine. So generally, when you're getting a recommendation, you do not have a randomized controlled trial or the trial results are so old and haven't been validated in a specific clinical use case applicable to the patient in front of you. They really operate in the growing field as evidence generation as a service, and they enable clinicians on one side and then like biopharma, life sciences companies on the other to get really rapid insights from real-world data. So this is a crucial area where you don't always have an RCT for things. And I think, you know, they've got a great set of use cases for clinicians and also for pharma alike, of course, who may want to investigate important clinical questions like about clinical trial eligibility or understand the patient population that they're aiming to serve. So I first came across Atropos, just like a simple post on LinkedIn about some sort of news article. I was interested. I have a real-world evidence background.
Dr. Timothy Showalter:
[3:42] And I thought this particular application seems so challenging. And there's this story about having the Green Button at Stanford where the original co-founder of Brigham had basically built a way to ask a really rapid question that could be answered through the electronic medical record. And it just stood out to me because it's clearly like starting with the problem first, and asking the questions and delivering the solutions to scale a product like this. I just think that's fascinating.
Dr. Paul Gerrard:
[4:15] So my medical specialty was physical medicine and rehabilitation. So take that 90% of clinical decisions not having evidence and bump it up to, I don't know, 95%, 97%. But what we do have is tons and tons of, well, theoretically, we as a medical community have tons and tons of experience with seeing patients. It's just what's happening with patients in terms of treatment decisions and the outcomes associated with those treatment decisions is not making it back into the medical literature. And so it's really, really great when you have a company that is able to go out and get that data and figure out how to turn it into something usable to really inform us as physicians and turn that existing data into information.
Dr. Timothy Showalter:
[4:59] Yeah, I mean, if you think about it, it's, of course, with NLP and all of the sort of technology that's available now, I can imagine that these problems are solvable, but generally over months and years. I mean, what's remarkable to me is what they're trying to deliver is insights like you would use within the electronic medical record to answer a question for a patient in front of you. And I can just imagine on the care delivery side, figuring out questions related to inpatient care after surgery or pharmacology questions. There are a ton of these scenarios where I think this could be really helpful. Well, let me give a little bit of background on the company based on our research. So first off, in terms of just the basic gist for founding stories, Atropos Health was founded in 2020. It was spun out of research done at Stanford University. The original inspiration was this like early product, the Green Button, which allowed Stanford physicians to submit clinical questions and receive really specific tailored analysis of EHR data within days that would address their clinical question. They've grown a ton since then. They've raised about $33 million based on the reports I could find online in a Series A. And then they've added high-profile strategic partners to their funding.
Dr. Timothy Showalter:
[6:23] And that includes Mayo Clinic and various health systems. I'm sure there are many more. They've got a pretty all-star investor syndicate, Briar Capital, Emerson Collective, Presidio Ventures, and Samsung Next. Based on what I could find on my internet search, I think the company employs between 50 and 100 people, somewhere around that range. And they've got a really strong presence in the academic health system market. The co-founders are Dr. Nigam Shah, who's a Stanford professor and like multiple founder and innovator, and Dr. Brigham Hyde, who's CEO and who we'll interview next. He's the former co-founder of SignalPath, which was acquired by Verily. So he's got clear, real-world evidence in analytics bona fides. He also has an interesting pharmacology background. Their board members are quite impressive, includes the venture capital folks you'd expect, but really prominent figures in healthcare analytics. So they've clearly put that really strong team together. And of course, Brigham Hyde brings significant experience from work building companies directly in the healthcare data ecosystem. So, these include Symphony Health and Relay Health. And I think he's, of course, well positioned to bring all that he's learned in that and in those experiences to bear for Atropos.
Dr. Paul Gerrard:
[7:39] The problem that they solve is the lack of access to timely and patient-specific evidence, because only around 90% of clinical decisions are made without that high-level evidence, and that opens the door to care that is based on judgment, which sometimes may be right, but sometimes may be suboptimal, and of course can be highly variable depending on where you are and who you're being seen by as a patient at that time. The timing is really right for this now because to aggregate data and use data to turn into evidence, you have to have that data somehow digitized. And so the explosion of EHR and claims data is something that has enabled that. Now, of course, we've had EHR and claims data for a long time. So one of the big innovations here is to figure out ways to go in and capture that data, but that was necessary infrastructure that had to be in place.
Dr. Paul Gerrard:
[8:42] The other thing too is that we're seeing increasing acceptance of real-world evidence and decision-making from policymakers like FDA and CMS. I think under FDA, some of the drugs have gotten label expansions with real-world evidence. And FDA has this distinction between real-world data and real-world evidence and really talking about how can we convert real-world data into real-world evidence. And on top of that, the most recent things that have really become important is this growing demand for personalized medicine. And of course... At some level, when we really want to make things personalized, it's tough to have randomized control trials for every possible permutation of a scenario. So we need to have ways, new ways of getting information that can be personalizable. And so from a physician's perspective, this is really about closing the evidence gap at the point of care and giving clinicians more confidence than their decision. And from a patient's view, it's about receiving the best possible care and care that is grounded in data and patients like them being seen in the real world.
Dr. Timothy Showalter:
[9:51] Yeah. I mean, it's kind of interesting because you think about like in the evidence-based medicine paradigm, like for the, for this general approach, I mean, if you're asking questions of your EHR or, or maybe you're asking it more broadly within the Atropos system for a, for a patient population that directly reflects the patient in front of you, it just begs the question of like.
Dr. Timothy Showalter:
[10:13] Is that more, should you base your decision more on the real-world evidence if it's directly representative of the patient that you're trying to help or a randomized controlled trial that may have various biases in the patient population or patients may have been staged in a certain way that may or may not apply to your patient. It's kind of an interesting concept that could potentially have implications for how we approach patients. Well, let's cover like just briefly their solution. So I think what's really cool about it is that they've like built it around how clinicians would actually engage. From reading about their solutions, I have not personally had a chance to try these out, but basically you're able to ask natural language questions. So rather than filling out, you know, very specific study objections or whatever, if you're a clinician, you can ask plain language questions about your clinical dilemma, and then you can receive a structured analysis of de-identified data. So I think what that really opens the door to is maybe just clinicians who are not real-world data experts, asking questions where the data can be used in a very specific way without having to go do a bunch of research training to know how to ask the right question. Their newest launch that they describe on their website is Geneva OS, and it's described as essentially the full operating system for real-world evidence generation.
Dr. Timothy Showalter:
[11:38] So I think, obviously connectivity and interoperability and APIs and all these sort of technical interfaces have been pretty sluggish in the healthcare space. And I think it's really exciting to be able to connect like federated data across multiple health systems and allow really larger scale queries, both for doctors at health systems as well as potential life sciences companies. So I think it really unlocks the full value of what they've built.
Dr. Timothy Showalter:
[12:07] And it can be used to generate custom analytics and to support research as well. So it's been positioned as really the next generation infrastructure for leveraging real-world evidence quickly and efficiently for informing clinical questions. So this is the sort of scenario where they're putting these tools into the scale of days to get answers, not months, and use academic grade methods. So basically, they're really doing running rigorous analyses, presumably that have been scripted or pre-prepared and then presenting the results clearly to physicians. And then they're able to work across various EHRs and partner with different research networks. So thinking about it, I think the real right to win here is that it's so clinically usable, I guess mostly for physicians, but also for research teams at pharma companies if they have important questions to ask. And I think that usability is something that, honestly, if you had asked me if it was possible, I wouldn't have envisioned being able to do research this quickly. But I think if their goals were to support rapid research, they obviously had to find the right solutions for it. And so kudos to them for doing that.
Dr. Paul Gerrard:
[13:24] Yeah. And I just want to take a moment here to point out, I mean, they developed an operating system to provide an infrastructure layer to really support that. And I think that sounds to me like they're really thinking... A step ahead in terms of, not just how we can work with sort of the existing arrangement of IT infrastructure, it's what can we do as a company to improve that and make sure the infrastructure is there to help address these real-world evidence questions.
Dr. Timothy Showalter:
[13:55] I think their solution is novel, but, Paul, can you go into a little bit of what's the rest of the market look like? Like, if you think about their competitors, like, who are they really, who else is in the space?
Dr. Paul Gerrard:
[14:07] Some of these companies include Etion, who's focused on real-world evidence for pharma, Truveta, who's really focused more on population-level analytics, and then some of the other players that probably most people have heard of, such as Health Verity, Flatiron, Komodo Health. In terms of the customers, their main customers are academic health systems, life sciences companies, and then payers and policy groups as well. Yeah.
Dr. Timothy Showalter:
[14:33] And I'm thinking like probably Concert AI, which I think Brigham helped found or was pretty involved is still there in Tempest. But I can't, clearly, I think from how this shows up to clinicians, at least they're probably in a pretty unique position. But I suppose a lot of these companies are really playing in the pharma space as well and trying to accelerate research. So that's really interesting.
Dr. Paul Gerrard:
[15:01] Popular space to be in.
Dr. Timothy Showalter:
[15:02] I did my time at Flatiron. So yeah, it's a really interesting space. Lots of potential clinical impact. And so it's always great to see new entrants in that space.
Dr. Paul Gerrard:
[15:11] So just to get to their business model and strategy, the revenue comes from subscription services for health systems, evidence generation projects, primarily with life sciences companies, and then the potential for government and payer partnerships to understand exactly sort of what is typical and what's happening out there. The policy trends... To be mindful of are that the FDA is using real-world evidence for label expansion. As we discussed a little while ago, there have been a couple of drugs that have had label expansions based on real-world evidence. So it seems like this company is really tapping into that opportunity, and hopefully for them, this opportunity will continue to grow. Additionally, CMS is really focused on wanting to see evidence of effectiveness. When that is present in the form of a randomized control trial. Great. When there isn't a randomized control trial, you still need to show something to CMS. And so this offers an alternative opportunity. And then data sharing infrastructure. That's been very slow to develop in the healthcare field, but it's slowly coming along. And it seems like Atropos is really well positioned to continue to take advantage of that. And as this improves, it should do nothing but benefit them. Now, there's some barriers to entry too, which are, first off.
Dr. Paul Gerrard:
[16:39] They're a company focused on having access to lots and lots of people's data. And you have to be trustworthy to do that. You need to have good infrastructure in place to ensure health privacy. And that infrastructure is not just technological, but also policy infrastructure to make sure that people aren't getting inappropriate access to it.
Dr. Paul Gerrard:
[17:02] And then data curation is going to be huge if you moving through massive amounts of data and want to get useful, abstracted data elements out of it, the ability to curate that massive amount of data accurately is critical as well. And then, of course, getting humans to buy into this. Clinicians are notoriously slow to change. I say that as a clinician who probably has been notoriously slow to change myself. You know, you get a way of doing things. It's efficient. It works well for you. So integrating something new into your workflow can be tough to do.
Dr. Timothy Showalter:
[17:37] I mean, that makes me think, Paul, like there's some doctors that just feel like they already know what they're going to do. Right. So it's pretty interesting. I wonder like how, of course, they'll change over time, but those aren't going to be early adopters. But it is pretty interesting. I'm also thinking, I've seen a couple updates from the FDA already with the new administration that suggests to me maybe there'll be more appetite for considering different data sources and analyses like this in regulatory applications. So I think everyone's been kind of looking to see how it's going to be used by regulators for actual decisions. And it seems to me like maybe there's some tailwinds for that.
Dr. Paul Gerrard:
[18:19] Yeah, I would agree. I don't think we have a lot of examples of that happening yet, but there have been a few examples of like palbocyclib, for example, had a label expansion to men with breast cancer based on real-world evidence. So I think there's we're slowly moving in that direction.
Dr. Timothy Showalter:
[18:36] Yeah, well, I'm looking forward to speaking to the CEO and co-founder up next. So Atropos Health is another great one for us to cover at Health Tech Remedy. Up next is our interview with Dr. Brigham Hyde. Brigham, welcome to Health Tech Remedy.
Dr. Brigham Hyde:
[18:53] Thanks for having me, Tim, Paul. Should be fun.
Dr. Timothy Showalter:
[18:55] Well, let's hop in. So, first off, we'd really like to hear more about your own personal journey. So, how did your background bring you to ultimately founding Atropos Health?
Dr. Brigham Hyde:
[19:06] Yeah, I spent most of my career working in health tech and health data at places like Eversana, Concert AI, Decision Resources Group. I've been part of what I would say is sort of the healthcare data ecosystem during that period. I first saw the technology behind Atropos within Stanford. Dr. Nigam Shah, who's a chief data scientist there, is my co-founder. And he was running something called the Green Button Consult. A simple idea, hard to execute. Could you let physicians at the point of care with a patient in front of them ask for a second opinion from all the data we've created? And in order to make that work, you had to build significant automation and methodology and sort of quality to that in the turnaround time in which you're seeing the patient, meaning at the time, a couple of days, we now do those studies even in minutes. But it's that automation that enabled that point of care use case. I've been doing the real-world evidence studies and that type of work with pharma for many, many years. In my hands, it still take weeks and months to do. And so when I saw what the Green Button was capable of, I thought, wow, really awesome to be able to bring this back to the point of care and actually help physicians treat their patients with great, high-quality evidence, but also that this automation could help speed up research efforts in life sciences and add efficiency to that whole process. So I was super intrigued, got the chance to help them spin it out about four years ago and launched Atropos on the back of that spin out, and we're off and running today.
Dr. Timothy Showalter:
[20:33] It's pretty remarkable. I mean, my impression from following the real-world data industry is that usually it takes months or years even to curate real-world data sets. And so it's pretty remarkable to see what you guys were able to accomplish with the Green Button. Going back to that moment when you saw the Green Button demoed, what was your reaction at that time based on your prior experiences?
Dr. Brigham Hyde:
[20:56] My reaction was, what is this black magic and why am I so stupid? Because... I've been toiling away in the real-world data mines for years. I have a PhD in pharmacology, so as far as a bench and clinical researcher, I've been doing this a long time. And I was sort of amazed that they were able to execute that so rapidly, but without sacrificing on methodology and quality, which was super exciting. As I dug into it, there's a bunch of really amazing things about the way they built it, a lot of real important decisions. I'm generally pretty cynical about tech in healthcare. We kind of all get the same tech at the same time when Amazon releases it to us to build upon and wrapperize.
Dr. Brigham Hyde:
[21:39] But they were sort of forced. This is like, I think, a good way to evaluate technology innovation. Why did they build it the way they built it? They built it the way they built it because they couldn't get like 20% faster. The weeks and months it was taking, all of us were trying to get a little faster what we were doing. But they had to get like 99% faster. And if they didn't, the patient's gone the moment with the doctor's over. So they had this sort of cauldron that forced them to think about things really differently and made a bunch of really terrific decisions. We wrote our own query language called TQL, or temporal query language, that was meant to overcome a lot of the challenges around relational databases and SQL that all of us in this space know so well. So the fundamental flaw of set theory, which belies SQL, is that it's not very good at calculating time. And yet most of what we do in observational research and real evidence is time between things. So the reason to sit down and build your own query language is like, it's never going to work in SQL world. We've got to come up with a different way to do it. And we base TQL and concepts of causal inference, which are at the core of a lot of the statistical approaches done in observational research. A lot of great decisions went into it. I think just forced by that pressure cauldron of the patient's leaving, we have to get the answer back 99% faster than it was.
Dr. Timothy Showalter:
[23:03] So in my understanding is that the use case that they were solving for with the Green Button is primarily like clinical questions, right, that would impact patient care and you have to know about that patient in front of you. Is that correct?
Dr. Brigham Hyde:
[23:15] Yeah, that's right. You can send in a question as simple as a couple sentences of text. This is before LLM world even started. We were text-based input. And the thinking there, which was, I think, an ingenious observation by our other co-founder, Sarp Gumbar, our chief medical officer, was doctors will never do anything outside of what they currently do. You've got to make it feel like they're emailing a colleague, which is what they'd be doing anyway. If they're unsure about patient care, usually they're asking a colleague or doing a peer-to-peer consult. So we needed to make that input feel just the same. And that brought us to text. And the sort of cue to that was converting then that text-based question, maybe a couple sentences, that might include attributes based on like, hey, I've got a patient with a history of diabetes, I've been on this drug, converting that into a good research form. And that's the core of what our technology does. Now, post-LLM world, we have ChatterWD, which is our LLM interface to all of this. And the user can chat with an LLM to send in that text question and structure a well-formed question, run a study now in minutes, which is pretty amazing. I mean, having done this work, again, in the coal mines of RWD to now you just type away and you get a study done on millions of patients, that's pretty amazing that we're here.
Dr. Timothy Showalter:
[24:31] It's remarkable. It's like mind-blowing to me. I mean, you have much more real-world data background than me. I basically was in academics and then spent a year at Flatiron Health. But to me, like what you're capable of doing in this, the speed of it is just pretty mind-boggling. I'm curious though, like you have a ton of experience before co-founding Atropos Health. And so you clearly brought a whole set of accomplishments and skillset and industry knowledge. What did you bring from your time at Concert AI and Eversana that really helped sort of unlock the innovation at Atropos Health too?
Dr. Brigham Hyde:
[25:07] You know, in those companies, I think we always dreamed of being able to bring the value back to the patient, but the economics weren't there for it. The time it took to do these studies, difficult to do. While we'd have academic research collaborations, it really wasn't getting all the way to the patient.
Dr. Brigham Hyde:
[25:23] That's one of the things from that experience that excited me about this was, oh, we can fill that dream. From a business perspective, that also represents a completely untapped TAM, right, that has not been hit by those businesses in a real effective way. So that was always exciting. I think the other side was I've worked a lot with life science companies, pharma companies over the years. I understood their context, and I always felt that this could have a real impact to them. And in fact, that's how we started. We worked with providers primarily for the first two years of the company.
Dr. Brigham Hyde:
[25:52] And about two years in, I went to colleagues at J&J originally, showed them what we had, and I said, I think this is going to help you. How do you see it? And they looked at it, and where we started with J&J was they said, well, why don't you install on the data we're already buying? You know, Health Verity, Truveta, Optum, Flatiron, and all these things that they buy. And they'll use our tools to gain automation on top of that. So instead of having to do big services projects, lots of consulting dollars for that, they could get an answer from me in two days, and I would give them a SLA for that. And so being able to do that, the first six months of a pilot at J&J, they had a team of five, which they gave access to us on those datasets. They did 150 studies in six months, which if you know that space, that's probably 5x what they hope that team would do for the year. Also done faster. Right. And they called us at the end of that. They said, we love it. We're expanding. And oh, by the way, you saved us a bunch of money on consultants during that period. So, the value prop was really clear. And so that was really the base for us was all about automation, take the teams that pharma built, empower them with tools, think about them as users, like what do they want?
Dr. Brigham Hyde:
[27:01] And really focus on that. The other thing that this opened up, we launched the Atropos Evidence Network just about two years ago. And as we've been going along on the provider side, installing on different data sets, we now had significant amount of data on that network. Now, one critical thing about us, we are cloud-based and federated, which is a fancy way of saying we install behind the firewall with the data partner of the health system and no data goes anywhere. So we're not in the data buying and selling business. And that's sort of a new model for pharma. I mean, they classically would buy data assets and then analyze them internally. We, of course, would sit in their firewall on those data sets they would buy, providing the automation layer. But as it sort of evolved, we started to have really interesting data for them to potentially use. And actually, back to J&J, we would have a situation where a user would send in a query to us and we'd say, great, guys, which data set would you like to use of your internal? And we'd get a response that would sound something like, I don't know, the big one, or I've heard of Optum. So, like, let's use that. And, we said to them, like, hey, guys, you want us to be more empirical? Because we could run out on all five of the data sets if you want.
Dr. Brigham Hyde:
[28:08] And they're like, yeah, that'd be cool. So we developed this thing called real-world fitness score. So when you ask a question, we can run that query on multiple data sets, whether it's internal or on our evidence network, and we produce a score, like a credit score that says for this question, which data set is most appropriate? What this really does, it takes a lot of the guesswork out of, good data, bad data, data fit for purpose, and focuses it on your question, which one is the best for that. And that really opened up the evidence network as a cool opportunity because maybe Optum is the best for this, but maybe not. And maybe there's a source out here on the network that would be better fit. So by putting fitness and quality scoring right at the center, it's allowed us to sort of evolve the way that life sciences can even consume data. And for our partners on the health system side.
Dr. Brigham Hyde:
[28:57] They get to participate in that financial opportunity when their data is used, but without losing control of their data, right? So it stays in their firewall. They own it. It stays there. They're just monetizing passively through aggregate analysis. So we think it's a better model. I also think it's something that is better for the users and pharma. They want the right data at the right time. They don't want to get caught up in, well, I bought this data set, but it's not good for this. And so I think that focus on that user experience in pharma is something I took from a decade plus of being in the data aggregation space where it wasn't always the right data at the right time and there was always tension around that. This is a better way to do it, I think, and actually creates more opportunity for everybody because what we found, there are way more questions that pharma has than are going answered. And if we enable through automation and a network approach, them to always use the right data, be able to trust the output and do it quickly, they will do far more analysis. And I think that creates opportunity for every member of R&M.
Dr. Paul Gerrard:
[29:57] Are you able to give sort of examples or conceptual themes of the kinds of questions that you're getting asked and that you're able to answer?
Dr. Brigham Hyde:
[30:05] Yeah, absolutely. I mean, there is no limit on the types of study questions we can answer. I'd say our sweet spot is anything related to observational research where you're comparing outcomes, holding all variables the same except for maybe one. We get on the provider side, we've answered questions from every specialty, but I say there's like a rule of three where if one of these three things is true, we're going to get a question. Either A, there's not enough evidence in the literature or the guidelines, like there just is no trial to answer this question. Number two, if there's high stakes for the patient, right, life or death or critical surgery or something like that. Number three, if there's expensive things. So with those three, you can imagine we do a lot in hemonc. We do a ton in specialty care overall. We do a lot in sort of surgical orthopedics, like anesthesia, like all these areas that there's not a lot of studies out there. One of the interesting areas that we do a lot of, which is sort of interesting, we get a ton of questions from geriatrics and pediatrics. Why is that? Because nobody runs trials on geriatric and pediatric patients or not enough, right? So that sort of evidence gap that's out there drives a lot of the demand for us. In life sciences, we work across both R&D and medical commercial pretty extensively, a bit with brand, a bit with commercial. But primarily it's, we're designing a trial. We want to know what the real-world outcomes look like. We want to know what the patient population looks like. Those types of questions we get all the time.
Dr. Brigham Hyde:
[31:34] HUR studies, those types of things, that's where automation really plays through for those folks.
Dr. Timothy Showalter:
[31:39] I'm curious, just follow up on that. What is your uptake or engagement been with community versus academic providers, like on the clinician side?
Dr. Brigham Hyde:
[31:48] Yeah, I'd say academics are our core. Those users tend to be more research-oriented anyway. And one of the things we see is that like somebody might ask a question, usually the first question they ask is they have a patient situation where they've seen it a couple times and they've tried to figure out what the evidence says and there's nothing out there. They come to us with that question. They see the results. Maybe they decide to publish on it. And many of our requests go on to be published. We have 100% peer review success rate when somebody submits and they'll get a poster at a conference, you know, what have you. Then if it works its way up far enough, a lot of those folks end up wanting to change policy. So like we've actually been used to change policy at institutions, whether that's, formulary, drug coverage side, or even on sort of the guideline quality and sort of pathway care side. So AMC is definitely the core. We have been used in the community setting. I'd say the use cases are a little more rote in the sense that there's a set of things in every care setting that are common problems. Like we're not quite sure what to do here. There's not a clear guideline. And those are the places that will be used. It can empower doctors to practice at the top of their license. I always say, if you give a doctor high quality evidence, they know what to do with it. The problem is we don't have enough evidence, right? The stat we use is only about 14% of daily medical decisions have any high quality evidence behind them at all. So like if we had more evidence than you gave it to the doctor, they know what to do with it. And I think that's the way we're seeing primary care play out of it.
Dr. Timothy Showalter:
[33:16] That's interesting. That feels spot on to me. Like I still, I still practice a day a week and And I would say, when you were laying out your conditions for the ideal use cases, that first scenario of having insufficient data available for randomized controlled trials, even in the oncology space, feels like a very common scenario. So you can imagine how powerful that is.
Dr. Brigham Hyde:
[33:37] I'll just, we've got a million of these. We get, one of the things I would say, if you're having a bad day at Atropos, there's a Slack channel that has all the questions that are being asked that come in. It's cool and inspiring to hear like almost every story. But just like one example, GLP-1s are very active out there right now. We got a request the other day that was, okay, GLP-1 versus metformin. GLP-1s can be more expensive, obviously, and will impact patient economics. And we wanted to know in diabetics, which ones perform better. Now, it wasn't just diabetics. This patient had liver disease. And when you looked at the trial comparing GLP-1 and metformin, they had excluded most liver disease patients. So this doctor was going, well, my patient has liver disease. This is still going to work. And that's a study we can run in minutes and produce an answer. In this case, it showed that GLP-1 is still more effective on BMI reduction in that patient. You could also look at side effect profile if you wanted. And they get a clear piece of evidence. By the way, a study run on like 50,000 patients matched to that patient set. Powered, we run statistics in triplicate. We do unmatched basic match and high dimensional propensity score match to remove confounders exist in real world. The P value was strong in all three cases. And they had a unit of evidence to go, okay, I know what to do. And they could go act on that. So we get questions like that every single day, which is cool and inspiring. There's always amazing sort of patient stories behind all this stuff.
Dr. Timothy Showalter:
[35:00] That's fantastic.
Dr. Paul Gerrard:
[35:01] If a clinician or an academic medical center wants to get involved in this, how do they go about doing that?
Dr. Brigham Hyde:
[35:06] Yeah. So when we work with academic medical centers, we're usually partnering with the broader institution. We just announced a big partnership with Emory, by the way, which we've been in the press about a little bit. And they're partnering with us. They're purchasing access for their physicians. That gets integrated also into their workflow. We're now starting to integrate into the ambient workflow, which is pretty exciting. So like imagine you're a doctor, you record your visit. It produces a note and a prognostogram or study matched to that patient pops right up in workflow. So you know what to do. So lots of cool stuff happening there. And our partners at Emory use this a lot in the pharmacy setting as well. So this is a great tool for pharmacy with like day one ROI, like, what should we use this expensive drug or not? Right. And 340B drugs, as well as those that are covered on inpatient DRGs, like there's real economics behind these things and there's not enough evidence and people are making decisions without really knowing what the clinical evidence should be. We have a day one way to give them evidence to make those decisions, which the pharmacy setting can actually directly drop to the bottom line in terms of cost savings.
Dr. Timothy Showalter:
[36:14] That's so interesting. Is the revenue model then primarily like agreement and subscription based or is there any sort of shared risk approach for savings?
Dr. Brigham Hyde:
[36:22] Not yet. I mean, we'd be open to that in some cases because we can see that what's happening where people are using it that way. Primarily we're doing user based SaaS in those settings based on the number of users who can ask questions across our system. The institutions do get the opportunity to participate monetarily in the evidence network if they want. It's not required. But if they'd like to make some money back by enabling their data on the network, they can do that. They get a ton of transparency, who's using it for what. We do our quality scoring for their data to help them evolve it. And they also, again, don't have to lose possession of their data. They're not sending it off somewhere to be aggregated and sent somewhere else. So there's some security and sort of IT friction reasons why that's also good.
Dr. Paul Gerrard:
[37:05] I guess that sort of plays into questions on things like data use, patient privacy. I mean, even if you're good stewards of that, I guess, are you running into friction or concerns over that? And how are you making potential partners aware that you will be good stewards and gaining their confidence and trust?
Dr. Brigham Hyde:
[37:24] Yeah, I'd say it starts just with the federated setup. It's just a better setup. I mean, and you're talking to somebody who spent a decade aggregating data, like that's the way you had to do it if you wanted to produce these insights. And with the cloud infrastructure, we now have the compute layer, right? Federated compute, which is what we're built on. I'm not sure the data has to move around at all, right? It's just from a data possession side.
Dr. Brigham Hyde:
[37:51] Second, with tokenization and DID now, I mean, most of the time we're sitting on de-identified data entirely. That's our preference. We can sit on identified under BIA. We do that in some cases, but primarily we're on the de-identified data set. And then the last piece of it, Paul, is transparency, right? Like who is using my data for what? Down to the question, who asked it and why? Like having that transparency and enabling a system that allows you to say yes or no based on that basis, as well as getting a clear report from us every quarter, like what was done, so you can track it down. I think what we found in this space, and I go back when I was at Concert, our relationship with ASCO CancerLink, ASCO did a big survey of patients and asked, in a de-identified form, would you be willing to contribute your data to advanced research? And I forget the exact stat, it was in the 80 percentile range. People are willing to do that. They just want to know it's being done correctly. They're not introducing new risk. And that's where I think this, it's like the sunlight disinfectant is what solves this for all of us. More transparency is better. And frankly, it's good for both sides and allows more participation in that market.
Dr. Timothy Showalter:
[39:03] What are you, as we enter 2025, what are you excited about for Atropos? Like what's coming down the pike this year?
Dr. Brigham Hyde:
[39:10] We're doing a lot of work with life sciences. We work with eight of the top 11 right now. We announced our partnership with Merck back at JP Morgan. There'll be more news coming from us. So I'm excited just to, the thing about pharma, like pharma is really in the business of producing a lot of the high quality evidence out there. They do it through clinical trials. Automation like this is going to let them scale their ability to do that. So I'm thrilled about that. On the health system side, I mentioned the Ambient integration. I mean, so cool, right? I mean, and Stanford is one of the places we're doing this. Think about it. It's low friction for the physician, personalized evidence at the point of care. Like we actually might be. Personalized evidence. And this type of integration is super exciting for us. Again, I think you'll see more from us on the automation side, more coming out about ChatterWD, more partners joining the Atropos Evidence Network. What we're finding is not just the health systems, but even the community of data providers are finding value in this because what we're doing is we're unlocking demand. If you can ask and answer a question quickly, you're going to ask more. And this is a way for them to participate in that, monetize as well. So, we feel that we serve the whole community of people and companies that are doing this work and just trying to make it high quality and more automated.
Dr. Timothy Showalter:
[40:26] Well, I know we're coming up on time. A lot of our listeners are either physicians or scientists who have some entrepreneurial bent or they're focused on innovation. And you've had a number of experiences along the way. So since you have a track record, I think you also can expect to be asked for advice over time. So I'm just curious, what would you tell a young scientist or a physician who's interested in getting involved in more innovative or entrepreneurial activities? And how have you sort of approached making an impact?
Dr. Brigham Hyde:
[40:56] Yeah, I think the theme for me has always been curiosity, and that's a good DNA thing to look for. If you're thinking about yourself as an entrepreneur, like, is that what gets you up every day, every morning. I think ultimately that's one part of it. And then the other part of it is just relentlessness and tenacity. Those are key attributes because you're going to hear no a lot. Believe me, like, when we first brought this out, real-world evidence at the point of care, what's happening? The world will come to you eventually if your idea is right. I think the other thing to think about is the problem you're solving and where that fits in the broader ecosystem, right? Where are you creating value for whom, right? How do you really think about user experience? Because I think that's something in health tech we classically haven't gotten right. Like you really got to think about that user experience.
Dr. Brigham Hyde:
[41:47] Another great example of this is like Abridge, like nailed that user experience on, the early days of Ambient. And that's what ultimately leads to adoption. I think the last thing I'll say is we're in a really rapid cycle, innovation cycle right now with AI and LLMs and it is complicated to keep track of so knowing what you're tacking to. Like for us when we're building ChatterWD there was the discussion of like do we need to build our own foundational model here or do we bet on all the models getting better and we figure out how to use them in our process and I think we made a good decision there doing the latter which is every time there's a new ChatGPT release or a new model comes to four that benefits our system so thinking about the technology back and how you want to work into that and around it is another critical part of evaluating these companies. It's more than just, I found a good problem to solve that I'm curious about. Okay, that's good. I've thought about my users and now this will integrate. That's important. And then tech vector, how am I going to align to what's happening in the broader ecosystem?
Dr. Timothy Showalter:
[42:50] Thank you. Well, we could stay and talk to you for much longer. I promised that I would get you out in 30 minutes. So let me go and wrap things up. We're so fortunate to have Dr. Brigham Hyde join us on Health Tech Remedy. Thank you so much for your time.
Dr. Brigham Hyde:
[43:03] Thanks a lot, guys. It was fun.
Dr. Paul Gerrard:
[43:04] Thank you.
