3156: From Siloed Data to Smarter Healthcare: The OMNY Health Approach

[00:00:03] What if we could democratise healthcare, ensure that every breakthrough, every insight and every innovation is powered by data? Data that reflects the diversity of real world patients. Well, my guest today is an emergency medicine physician and also the visionary founder of OmniHealth, a company that's reshaping how healthcare data is shared, structured and then applied.

[00:00:30] And it was all born out of his frustration with the fragmented infrastructure of healthcare data. And as a result, OmniHealth now represents a staggering 80 million patients across the US, serving as a secure and compliant data front door for research and for collaboration. So today I want to learn more about OmniHealth's bold mission to overcome some of the most persistent challenges in healthcare.

[00:00:58] Yep, I'm talking about siloed systems, messy data, complexity of privacy regulations, and also uncover how their platform is not only enabling groundbreaking studies, but also supporting the development of AI technologies that are compliant, secure, and representative of the diverse populations that they serve. So are we on the brink of a new era in healthcare? Where data and technology bridge gaps that once seemed insurmountable?

[00:01:28] Let's find out as I get my guest onto the podcast now. So a massive warm welcome to the show, Mitesh. Can you tell everyone listening a little about who you are and what you do? Yeah, yeah, happy to. And thanks for having me. So I'm Dr. Mitesh Rao. I'm the founder and CEO of OmniHealth. I'm an emergency medicine physician, spent most of my career as a health system executive, prior to starting our company.

[00:01:54] I led safety and quality at Stanford, where I'm still faculty. Also helped lead innovation at Northwestern before that. I've always been in the data space. My whole career has been focused on healthcare data.

[00:02:07] And really, it was out of years of frustration, where I kept having opportunities for data partnerships come forward, whether it was pharma, med device companies, early AI and analytics companies, who all understood that to really transform healthcare, you needed access to that deep and powerful data we were generating on the clinical side. But the question was always, how do you do it in a way that is secure, that is compliant, and that really can fuel the next generation of development of the technology?

[00:02:36] And I couldn't find an infrastructure to really enable it at scale.

[00:02:40] And so we built Omni with that vision, really out of pure frustration, and have now grown it into a national network that today represents a good swath of the country, almost 80 million patients on the platform, and represents, think of us as almost a data front door, really serving as a common language of collaboration between the provider side of healthcare and the pharmaceutical and medical device, the research world.

[00:03:07] So a lot of what we support is really patient-centered research on developing the next generation of therapeutics, on precision medicine, on safety and efficacy studies for pharmaceuticals and medical devices, really focused on supporting health economics outcomes research. That's our purposes. We believe that data can serve as that foundation for the future. And in this day and age, almost everything in healthcare is being driven by data.

[00:03:37] It really is. And I love the story behind your journey here and how it was born out of frustration. So many people will experience frustration like that and just complain about it, but you've gone out there to try and fix that problem. And for people listening outside of the healthcare industry, I was doing a little research on you and I was reading about healthcare data, how it often ends up siloed, which limits its use. Can you just tell me a little bit more about that problem just to bring it to life? No, happy to.

[00:04:05] I always say that it's probably the biggest Achilles heel for the healthcare industry is the fact that our data and our infrastructure on the technology side has traditionally been set up to really limit our ability to collaborate. I like to say that we're data rich in healthcare, but information poor. And that's a combination of a few different factors.

[00:04:31] One, the underlying tech stack is not really designed to enable secure and compliant data sharing and data collaboration. It's designed to create these silos. And that's a historical context. That is what it is, but it's always been a challenge. The second thing is that healthcare data itself is incredibly messy and complicated. And it's not easily usable. And you see a ton of variation.

[00:04:59] Even within one hospital system between the individual hospitals, you can see data being coded different or labeled different. There's tons of heterogeneity in the data, and that makes it very difficult to use on a broader scale, like a national level. So you've got sort of those two inherent challenges. And then the third challenge, which is a really important and critical one, is that our industry is very highly regulated when it comes to data.

[00:05:24] And that's an important piece because we have called a sacred duty to protect patients, protect patient privacy. And that compliance and regulatory hurdle is an important piece. So we now talk about things like HIPAA and limitations and restrictions and regulations. And they're purposeful because we deal with very sensitive information. So all those elements combined make for a lot of challenges in healthcare. And that's why historically, this has been a very difficult piece to bridge.

[00:05:52] Other industries move very quickly on data. And they use it as a foundation for a lot of their advancement. And in healthcare, we're kind of playing catch up. And after experiencing that frustration, you created Omni there, which aims to open up that data to establish a unified network that connects healthcare and life science companies, to name a few. But can you tell me a bit more about that solution and how you're able to do that? Yeah. Yeah.

[00:06:20] So Omni partners directly with provider organizations all over the country. Large IDNs, nonprofit health systems, community health systems, academic medical centers, large specialty groups in dermatology, GI, orthopedics. And we effectively help them get their hands around their data. So we help them take the vast repositories of information that they've generated. We clean it, structure it. We de-identify it to protect patients.

[00:06:50] We turn it into research and regulatory grade real world evidence. So think of us as refining that data. And then we create the secure, compliant pipes, if you will, for them to be able to build collaborations on that data.

[00:07:05] So now when you have a medical device company that's looking to do a safety and advocacy study on their device and make it safer, not only can we build a partnership data between that medical device company and the health system, but we can also do it in a national scale where now that safety data represents every geography, every ethnicity, every archetype, every region. So it truly makes that safety and efficacy study now valid for the entire nation.

[00:07:34] And that's a really important piece is it's not just about picking data and making it usable, but doing it on a national scale so that you can improve health care for every American. And democratizing health care through the power of data is incredibly ambitious. So to bring that to life, is it, is there anywhere, do you get to hear any use cases or, or feedback stories of how this network facilitates that, that you take it?

[00:08:03] I mean, communication and collaboration, et cetera. That's a great question. So everything from studies that have helped improve the safety of medications and reduce errors to pharmacovigilance, to understanding why patients are failing therapies or why certain care patterns have led to clinical improvements in things like rare diseases and oncology. I mean, there's just so much nuance in this data.

[00:08:32] It's truly the deepest and most comprehensive data layer in automation. And that's, what's really allowing us to, to step forward and do things like bring new, new, new treatment patterns to market, help patients with rare diseases who have suffered for generations now approach new care, new treatments that are life changing, life altering for them. You know, for example, we do a ton of work in the autoimmune and inflammatory space.

[00:09:00] Just today, we announced another partnership now in the autoimmune space where we're helping bring in precision medicine testing and pairing that with our clinical data to be able to create comprehensive journeys for these autoimmune patients, rheumatoid arthritis, helping them tap into the power of precision medicine to hopefully change and transform their long-term outcomes, right? Lead to improvements, let them live better lives, let them reduce the suffering that they have from this disease, which is a chronic lifelong illness.

[00:09:29] That's an important piece of the work that we do. Our North Star is always the patient. We're always centered on what research work, what development work can we support that's going to transform the lives of patients going forward. And to answer your question about like, this is a lot to take on, it's a big mission. And the reason we did this was sort of twofold. One, we said, well, if we don't build this, who's going to build it? And how are we going to move forward without this type of data? And I couldn't come up with that answer.

[00:09:58] So I felt that obligation to sort of step forward. But then two, it's the idea that if you're going to change healthcare, you've got to do it for every American. We want everyone to be represented. And that's an important piece because traditionally research has happened in silos. It's not really the representative of the nation. We talk about efforts now to improve diversity and inclusion in clinical trials and research initiatives. But the foundational piece of that is data, right? Like you need everyone represented in the data.

[00:10:27] And that's where you, when you do something like early stage research or you do something like safety and efficacy studies, then you can guarantee that it's going to make something safer for every American. And that's an important mission, I think, on our side, particularly as healthcare evolves for the country. And we tackle more and more challenges on a unified national level. And as this is a tech podcast, it almost feels like the law that I've got to ask about AI.

[00:10:56] So how do you see AI bridging the gaps in the healthcare system too? Neil, that's a conversation on its own. But I love it. I'm glad you dived into it because we see AI as a really important aspect of why healthcare needs to be data-driven going forward. So a few things to consider. And allow me to monologue here for a brief second because I'm passionate about this topic. First off, one of the biggest challenges for AI in healthcare is access to data.

[00:11:25] It's not like other industries where you can get access to information and train things. You can train a public-facing GPT model on publicly available data and it'll answer questions for you like, tell me about the Roman Empire or help me find a restaurant to go to today. And that's great. But if you ask a GPT, why is a patient with rheumatoid arthritis likely to fail therapy within the first 30 days? It can't answer that question for you because it hasn't been trained on the deep clinical information.

[00:11:54] So you've got that problem. And then the second problem is, well, how do you get access to that data, right? And then the third problem that you end up with is even if you do get access to that data or you train it, how do you ensure that the LLM is safe and it's representative, right? So we're tackling this from a few different angles. First, we're working with a lot of AI companies now providing best-in-class clinical data to be able to help train the large language models.

[00:12:21] And I'm talking, we have over a billion encounters, over 4 billion unstructured notes on the platform that serve as the largest repository of healthcare LLM training material now. And that's allowing companies like, for example, we have a partnership with QuantHelp where they're transforming the clinical trial space, right? And being able to help them transform the clinical trial space with Jarin of AI using Omni's data layer, right?

[00:12:47] We have other partnerships now with Aeros Global and the pharmacovigilance and signal detection side where we're helping them build next-generation large language models to help improve safety in medications, right, for the country. And many, many more AI partnerships now that are kicking off where companies have realized that, look, in order to get past that initial hype cycle, everyone's excited about AI. Now you get to the unsexy part of, okay, well, we're all excited. We want to build. Where do we get the foundational building blocks?

[00:13:17] And Omni's data layer represents that piece. The other important aspect of this is that, again, going back to the representation question, in the last year, a bunch of the LLMs got flagged in the public press because what they were putting out for outputs weren't necessarily representative of the country. And that's great. Like, that may upset people, but it doesn't necessarily physically harm people.

[00:13:39] And the challenge with AI in healthcare is that the outputs, they need to represent everyone because you need to have something that's actually going to change technology that's going to work for every American. But the second piece is that you need to be able to trace that output, right? So this goes back to the black box question of AI. If I ask a GPT a question on my phone and it gives me an answer, that's great. Like, I don't, I make questions like where did that answer come from, but I could probably trust it.

[00:14:07] And if something is off, I'm probably not going to get harmed. But if a large language model in healthcare gives you an answer, tells you, hey, Neil should be on drug A instead of drug E because drug A is likely going to be more effective for him. And then you put him on drug A and he has an adverse reaction or it turns out it doesn't work and he gets, he loses time in being able to get to an actual cure. Like, that's bad, right? And we don't want, we want AI to be safe. We want it to be compliant. We also want it to be effective. And we have to be able to trust it.

[00:14:36] And so one of the beauties of Omni's data layer is that it is from the source. You can actually trace down to the line of sight of the individual encounter, which means you could take a GPT output and you can understand that, hey, the GPT now has been trained on Omni data. It says that Neil's likely to fail therapy in the first 30 days because of these five reasons. And here are the thousand encounters, just like Neil, that this is based off of. So now you have almost a regulatory grade output in the GPT where you can trust the output. You can say, okay, I understand that.

[00:15:06] And that's, that's a big differentiator in healthcare because in healthcare, we need to be able to trust the technology. Right. And that's, that's a big part. We need to be able to trust it and we need to be able to ensure safety and not just being able to protect things like privacy, but also ensure that a GPT is not going to harm a patient. And that, that's rooted in data. You raise so many important points there about being able to access that data, the challenges around it, and also being able to trust the technology.

[00:15:36] And of course, anybody in this industry or on the periphery will, whenever you talk about healthcare data, they immediately think of cybersecurity, protecting patients data within healthcare data, etc. How do you navigate these challenges? Is it something that's, again, probably another episode on its own, right? Yeah. The security and compliance piece is a really important part because here's the thing. If you expose PHI to a large language model, it's likely to incorporate that and carry it.

[00:16:06] And in the future, there's a risk that it may spit that PHI out. So there's an inherent risk in what you train a large language model on. And that's a risk that nobody wants to take. And rightfully so, right? Because nobody wants to be responsible for sharing protected information. Nobody wants to be responsible for that piece.

[00:16:23] And so one of the things on OmniSci is that we not only securely and compliantly de-identified that data, but we get it certified, expert determined, so that you can feel confident that this is data that you can train down a large language model that is not going to expose sensitive information down the road. That is not going to have risk down the road. And it's not. So we talk about how we reduce harm and likelihood of harm from an output by having lattice-like truth. How do we ensure protecting patient privacy?

[00:16:53] By ensuring that this data is certified, expert determined. Like all of these boxes need to be checked. And that's our central focus. So we are the unsexy piece behind the flashy healthcare AI, right? We are the rails. We're the pipes. We're ensuring security, compliance. Like these are all the important parts that we don't often talk about but that are so critical. Because if you don't have these, AI is going to go nowhere in healthcare.

[00:17:21] There's just too many hurdles and there's too much risk. And we've spoke a lot around US healthcare today, which is obviously where you're based. And we do have people listening all around the world. And when I was doing a little research on you, I saw that I think you were attending a conference in Berlin in August. Do you have an international focus too? Because I see you're on the road a lot. Do you think about this too? No, it's a great question. It's starting now. And that's because of a few things.

[00:17:50] One, a lot of the partners we work with are multinational, right? And they want to be able to think about populations outside. Now we are, when we were early in our development, like there's just so much to do within the US. We were so focused here. But we are. We are starting to look now at partnerships in Europe and South America and Asia. But we can start to expand the data because now we want this data. The next phase of this company is the data not just representing the US, but representing the entire planet, right? Every geography, every country, right?

[00:18:20] Like we want to be able to expand that piece. But there's nuance there, right? Each country has variations in populations, in language, in healthcare systems, right? Some have nationalized systems with central repositories of data. Others are incredibly diverse with multiple different variants of EMRs and a lot of heterogeneity in their information. So there's no one size fits all. We will start to grow now internationally. And that's sort of the next phase of this business.

[00:18:47] And I think the most important piece here is that we've built the foundation within one of the most complicated healthcare systems on the planet in the US, which has a lot of inherent challenges that we've managed to overcome in the line. And so now the next stages of the growth of this company, I feel very confident. And on this continuing mission to connect data, to transform lives, et cetera, what's your big focus this year? Where do you go from here?

[00:19:17] What are you going to be focusing on? Yeah, this year is a big year of growth for us. We ended last year at over 75 million patients on the platform with living data, right? This year, we'll double that. We'll go to over 150 million patients, which when you look at the US population, now we're talking going from to one in three to one in two Americans on the platform. So that is truly a best-in-class representation of the nation. And that's our big mission for this coming year.

[00:19:45] The second piece is continue to expand now into areas like precision medicine, continue to support the AI space. I think last year was a big breakout year for healthcare AI, and now reality has set in. And so every day where more and more companies are coming to us looking to get access to this type of information that meets all those qualifications, right? That one is representative that could truly be building blocks for them. Two is done securely and compliantly, right? We check all those boxes.

[00:20:13] And on the flip side, we're helping all of our health system partners, all of our provider partners engage in AI and do it in a secure, compliant way. Because the last thing any of them want is to take risks with their patients, to expose sensitive information. So it's sort of two parts, right? Like this common language of data serves as a collaboration piece in research, but now it can also serve as a collaboration piece in AI. And that's really where we're growing this year. And I'm curious.

[00:20:39] Obviously, you said at the very beginning of our conversation you still work at Stanford University. So many creative minds, diversity of thought surrounded by so much greatness. Is there any moments of serendipity still happening now where you're meeting people that might be able to help on your journey, etc.? I feel like there's got to be a few stories there that might happen on a daily or weekly basis. I spend, I have a rule, which is that I never say no to an initial meeting.

[00:21:09] Even if it's like a 10, 15 minute initial call, anybody who reaches out, I connect with them. Because what I found is that everyone has different experiences in healthcare and they're tied into different problems. And the one question I always ask people is really, how can this data help you? Or how can we help you on your journey? And the answers I've gotten are what have allowed us to do things like expand, understand how to expand the genomic and precision medicine space. Understand how to expand into the AI space.

[00:21:37] Because what I found is that there's a lot of people tackling the multitude of angles when it comes to healthcare challenges. And everyone's fighting these really interesting battles. And there's ways that we can support it. So I do. I network a lot. I spend a lot of time. I travel almost every week. Whether it's at conferences or it's speaking on panels or meeting with partners. I do that specifically, not because I love being on airplanes, but because healthcare is a group sport.

[00:22:06] It's a movement that we're all trying to do because we all really believe in what can happen. When we advance this industry. And look, there are many other easier places to work. I always say that healthcare is the hardest vertical. It requires a certain level of masochism to commit yourself to when it comes to the tech space and innovation. But we're all here because we're true believers. We really believe that there's opportunities to improve the health of the nation.

[00:22:36] And that's why we commit to this, right? And so that's an important piece for me is going around having these conversations. And it's not just on the academic and university side, which I have a lot of those, but it's also some of the best ideas that we've tackled. Some of the biggest challenges are going to have come out of conversations within communities. Within folks who are frontline primary care doctors who are understanding and describing challenges that they're facing and limitations where data can make a big difference. So I welcome every conversation. I'm easy to get a hold of.

[00:23:05] And I look forward to how we're going to be able to support transforming healthcare together this coming year. Fantastic. And I think that every single way, if I look back at your own personal journey here for a moment, right at the very beginning, it was born out of frustration. You've created Omni here. But in that time, technology has continued to evolve at a rapid rate. The last two, three years alone is breathtaking.

[00:23:29] I think there's an almost pressure on every single person, including everybody listening, to be in a pressure of continuous learning. How do you keep up to speed? How do you self-educate and keep up to speed with those tech trends as they evolve? Any tips or advice you'd offer anyone? Yeah. So I am a voracious reader, and that has paid dividends. It used to be that I could read for fun and pleasure, which was amazing. Now I sadly don't have that much time. But I do.

[00:23:55] I get up early every morning, and I dedicate the first hour of my day. Once I've done morning exercise and gotten ready and the kids are out of the house, I lock off about an hour. In that hour, critical emails get answered, but I go and I read through. I read through multiple blogs and newsletters. I catch up on the news of the day, and I try to make sure that I cover as much as I can in that first hour before I dive into my day.

[00:24:22] And that lets me get a good grasp on kind of what's being talked about today, what's the latest news. I try to do the same thing at the end of my day, too, because inevitably there's a lot to catch up on. And I'm in the throes of my day. I'm focused on the business. I'm focused on clients. I'm focused on partnerships. And then there's a lot to catch up as well. So this is going to sound crazy, but I try to zero out my inbox every day, too. It's a little bit crazy to think about, considering the volume of emails I get.

[00:24:51] But I try. I have an internal rule that if somebody emails me, I try to respond within 48 hours. If it's not an email that I need to immediately respond to, but I actually connect somebody in, I try to forward that email to the right person. And I try to get to a clean inbox every day. And that ensures that not only am I grabbing all the information I need to grab, I'm catching up on everything I need to catch up on, but also that I'm keeping communication with everyone. So it's a tall order, but I've created that level of personal discipline.

[00:25:19] And that's allowed me to sort of keep a focus here because it's busy. I'm on the road every day and I work 12, 14-hour days sometimes. But that commitment to communication, that commitment to information, I think makes a big difference. A hundred percent with you. I'm completely with you on inbox zero. It's going to be incredibly hot. I think I achieved it over the holiday season, but it's starting to get difficult again. And blue light blocking glasses, which I'm wearing now.

[00:25:45] I know the audience can't see this, but I'm about to turn 44 and I can feel, I don't need reading glasses, thankfully, but I can feel the strain in my eyes, staring at screens and being on Zoom calls all day and running around. So the little things of making sure I keep hydrated, wearing blue light glasses, having a standing desk, staying active. And I take, even in my internal meetings, my team knows this, I'll take those on the, maybe an internal Zoom, but I'll be taking a walk when I take that internal Zoom.

[00:26:15] Or I've got a treadmill next to my desk and I'll get on a standing treadmill. Because that level of activity, I treat running a company like a performance sport. And it's not just the mental piece, but it's the physical performance component that lets me actually be engaged and present in the day of the day. Wow. A powerful moment to end on. But before I do let you go, for anybody listening that is inspired by you, your ethic, your work, and everything you're doing at OmniHealth, maybe they want to start a conversation or just find out more details.

[00:26:45] Where would you like to point everyone listening? Yeah. Look, I am very active on LinkedIn. I'm easy to get a hold of. You can go to OmniHealth.com and you can send us a message and we're happy to set up a time to talk. I'm not hard to find. If it's just to say hi or want to connect or want to talk about an idea, I have early budding entrepreneurs all the time who reach out. Who want to talk about an idea that they've got and, hey, is this valid for a company? Do you think I should actually go do this? So I'm always happy to talk data.

[00:27:14] I'm happy to talk healthcare. I'm happy to talk entrepreneurship. I'm happy to talk technology. Reach out. And I tend to respond relatively quickly. Sometimes it's a short response, but I try to respond to everybody who messages me. Well, we covered so much there from democratizing healthcare through the power of data. Learning more about how AI can bridge the gaps in the healthcare system and cybersecurity, protecting patient data within healthcare data. Even how to get to inbox zero. I don't think we could have covered that anymore, but thanks so much for talking about that today.

[00:27:44] And thanks for having me. It's been a pleasure. As we wrap up our conversation today, I think it's clear that OmniHealth isn't just building a platform. It's creating a pathway to a future where data and AI truly democratize healthcare. Whether it be addressing the complexities of siloed systems and messy data to enabling secure collaboration on a national scale, maybe even global one day.

[00:28:10] Because OmniHealth's mission is as ambitious as it is impactful. So what are your thoughts on the role of data and AI in shaping the future of healthcare? Could these technologies finally bridge that gap in access, quality and equity? Love to hear your perspectives on this. Please email me, techblogrider at outlook.com, LinkedIn, Instagram, just nice and easy, at Neil C. Hughes on any platform. Let me know your thoughts.

[00:28:40] But until next time, stay curious. Keep exploring the way that technology can make a real impact on our world for the good. We hear enough about the bad. Here on this podcast, we're trying to change that narrative. And with that, hopefully you will agree to join me again tomorrow. Speak with you then. Bye for now.