Episode 29: Mika Newton, CEO at XCures

Chris O'Brien Hello, and welcome to Biorasi's few and far between podcast. I'm your host, Chris O'Brien. My next guest has stated that cancer is complicated, even for medical professionals, and this is certainly an understatement. Fortunately, Mika Newton has become a steadfast navigator for cancer patients around the world, seeking more information on their condition. Mika is the CEO of xCures, a company which has built an AI solution for cancer patients designed to organize their medical data, connect them with physicians, and facilitate treatment options. Starting his story from the idea on a napkin stage four years ago, Mika and I had a chance to discuss the development of xCure's data pipeline, the importance of patients owning their own medical records, and the need to generate answers quickly and efficiently in an incredibly complex healthcare industry. We also talked about the continued and surprising importance of that age old relic, the fax machine, in building and organizing his medical records database. It was great talking with Mika, and I hope you enjoy this episode. Okay, let's start the podcast. Chris O'Brien Mika Newton: You are the CEO of xCures. Welcome to few and far between I'm really excited about this conversation as we learn a little bit more about the impact of advanced analytics and artificial intelligence on cancer. Welcome. Mika Newton Chris, I'm delighted to be here. Thank you for having me. Chris O'Brien So tell us a little bit about xCures to get started. Maybe you can give us the origin story first. What's the creation myth around the company? Mika Newton Yeah, thank you. I'd be delighted to. So at xCures, we brought the company essentially out of a nonprofit called Cancer Commons. So Cancer Commons had been helping advanced cancer patients with navigation advocacy. I always think of it as like, I have cancer, what should do and how can I get it? And that's been going on for about ten years. Helped about 10,000 patients, about 1,000 patients per year. And then we realized that if you were going to help people, you wanted to know whether whatever you informed them about, was that a good thing or not? So you needed to capture outcomes, right? Like what happened to these people? And then we wanted to do it at scale. And the original service really just provided direct human contact with PhD level scientists. So we thought, let's develop a technology platform to try to do those two things, capture outcomes and try and help a lot more people. And so we raised venture capital funding and did all the things you do to get a company starting it going, and now it's up and running. And we work with hundreds of patients a month, helping them, first of all, understand their own medical history, and then tracking what happens to them in the future and providing them and their physicians with decision support and alternative thinking about ways to treat their cancer. Chris O'Brien That's great. I love this question about origin stories because it's sometimes exciting and interesting stuff emerges that might not otherwise be obvious, and I think you had told me the Cancer Commons story when we spoke before a little bit, but would you double click on that a little bit? How did you get involved there? God knows that if you get a cancer diagnosis or someone in your family does, it's terrifying. And that addressing that urgent need for people for guidance and advice. I'm sure that's incredibly valuable. Mika Newton So Cancer Commons was founded, as was xCures, actually, by a gentleman by the name of Marty Tenenbaum. So Marty is an Internet pioneer and AI scientist and a cancer survivor. So he spent probably the last 25 years working on technology and just helping cancer patients directly. I always think about the advanced cancer problem, I've been through this myself, I lost both my parents to cancer, is none of us really know anything and probably shouldn't know anything about cancer until it happens to you or a loved one or you happen to work in the field. And the problem with that, it's a very complicated subject, even medically. It's very complicated, even if you're a medical professional. And so you suddenly find out you have this deadly disease and you have to learn an enormous amount about it and there's no right answer. It's not simple in the sense. I think we cataloged over 12,000 things that are potentially being done clinically in the United States alone. So how do you even know where to start in that world? And so Cancer Commons itself focused on getting PhD level scientists to help you understand the biology of your cancer and then do the research on those treatment options and what would they be? Mika Newton And we took a technology approach that it executes and said, can we read your medical records, basically, and identify from the hard data that exists in your medical record what are the most important features? And then we built a catalog of those 12,000 things. And so that whole idea of just help me understand what's out there. And from a technical perspective, I think of this as information retrieval. Like, I have to write a query, right, of some sort, and then I have to bring the right information back, which is what we do when we do Google, but it's hard to get it done that way. So we've been working on that process now, I said, for four years technologically, but going back over a decade with this nonprofit Cancer Commons. Chris O'Brien Yeah, I think everyone who has ever Googled a medical condition of their own, and probably most people listening have done that, have seen the horrifying range of possibilities that usually goes from this is no big deal to this will probably kill you. So Google is wonderful for many things, but not necessarily here. And my mom got a lung cancer diagnosis and I was working in healthcare, doing this kind of stuff. I still found it hard to navigate that at the time. So I can certainly see the value. So, okay, you guys then decide, hey, we think we've got something potentially really interesting here that can be developed more that needs to be a company in order to be developed more fully. We talked first a little bit about that decision. What did you need that you couldn't do within the confines of a nonprofit or cancer commons that made this make sense? Mika Newton So the reason to spin out a for profit is twofold. Number one, to raise capital, right? If you want to raise investor capital versus donations to pursue a project, it needs to be in something that can generate a return, a for profit company. Investors invest to make money. The other trend we were leveraging there was what we think of as impact investment. There are a lot of people in the cancer space who not only want to make charitable donations, but actually want to make investments into a sustainable business that's then going to support their cause as well. So there's a pretty nice overlap in that thinking. The other thing is to retain the staff. So we wanted to hire the best AI, NLP engineers, machine learning folks, data science team, engineering team, product team, clinical operations team. And to bring those individuals in, we also wanted to create a company where they would have appropriate compensation in a long term future. Chris O'Brien Such a great point. It opens up a different talent pool, I guess. I'm sure you got as often as the case in nonprofits, lots of really talented folks who are excited about working in a nonprofit organization. But yeah, for profit does bring a different talent pool, I guess. You were competing well. Tell us more. So where are you based? Who are the competitors? When you're hiring those initial staff, what are their other options? Mika Newton So it's compensation also an ability to provide equity, right. If you think about the startup world, a big part of the reason people go to work at startups, and I'll just tell anyone who's thinking about isn't because of the fabulous salary at the small company, right. It's because you get to participate in the long term value generation and you get to work on really cool projects very early on. Whereas in more established businesses, right, the products pretty much figured out. When we started this was basically an idea on a napkin four years ago. So we are based with a physical location in North Carolina, and that location is there because of the great talent from the CRO industry that you can hire. And so a backbone of what we do is essentially the clinical operations team who manages a lot of the data and final data validation pieces. Right. After all of the technology is done, you got to have humans still confirm and check things. And then when we originally started, we also had offices here in the Bay Area where I'm located. But as the pandemic wore on, we actually realized that a lot of our kind of senior management and engineering team, we could really virtualize. Mika Newton So now I think we have employees in almost five states across the US. And we've really, over the last few years, we live and breathe, zoom and slack, and virtual meetings are probably more so than I ever imagined possible, but it's actually a very effective way to work. Chris O'Brien Take us back, Mika, to that. So you said this is kind of almost like a cocktail napkin. When you guys initially form this, what would you have described the company or the opportunity, like, to be then and what is it like now? Usually we find that reality means we shift a little bit. Sometimes it's a big pivot, sometimes it's small. Talk to us a little bit about that, if you would, Chris. Mika Newton I love that question. So the very first thing we focused on was this idea of decision making, right? So when we started the company, what we said is, what's the decision and was it a good decision or not? What are the outcomes from deciding to pursue this therapy or that therapy? And part of the problem we were really focused on at that time and what continues to be an important question for us was in precision medicine, where you are making individual decisions for a person and potentially assembling a cocktail or a combination or novel combination of drugs and therapies. Could we learn from that? So it's like, could we not only make the individual decisions, but then aggregate these essentially n of one experiments where every person's being treated on their own and then develop a statistical model to understand why and when you should pursue basically one strategy versus another. That remains really important to us. But what we very quickly discovered as we started to set that up was in order to do what we were talking about, you needed really high quality, longitudinal data. And so we went on a mission. We shopped every single company we could find out there. Mika Newton In fact, we continue to do it, just trying to buy what we think of as real time, regulatory grade clinical data. Meaning I can't wait a month for it because I'm trying to make a decision with it. I need it now, right? I need it to be verifiable, ideally to the point where I could bring it to a regulator like the FDA and they would agree that it's really high quality data. And I need the clinical stuff. It's not claims or billing data. I need to know, did you end up in the ER? What was the size of your tumor? What was the interpretation? What were the lab results, the clinical part of it? And so we ended up, I wouldn't say pivoting, but really focusing our energy on that data pipeline. We use a relationship that we formed directly with patients under HIPAA. So the patient's right of access, and we leverage that to get access to their medical records. We can now get them electronically in the United States from, like, 90% of institutions. But when we started, even three years ago, this has been changing. We were getting everything via, like, mail and fax. Mika Newton So how do you get these faxes in the door and process, basically, images and try to do that at speed? And so we built an entire kind of technology based pipeline for requesting, receiving, aggregating, organizing, and then structuring those medical records. And then it really took off on us, I would say in the last, like, six to nine months as we started completing the integrations we were doing with health information exchanges in the United States, which has just opened up a whole new door for us. But if we hadn't focused, I think, on that pipeline early on, just getting access to electronic data wouldn't have been useful because instead of receiving kind of the slow mail, we would have been drinking from the fire hose of disorganized stuff. So it was that whole organization piece that ended up being a much bigger challenge, I think, than we thought on day one. We thought we could just source that and it wasn't possible. Chris O'Brien This is the kind of thing that is probably surprising to folks who have had a major medical condition themselves or in their family and tried to get access to records. The idea that fax machines still play an important role in our healthcare economy, it makes me crazy that this is a truth, but it is a truth. So am I understanding you correctly? In the initial days, it was faxes and printouts and you had a way to digitize that. Is that what you're talking about, that you built? Mika Newton Yeah, Chris, that's exactly right. And to this day, fax has not gone away. And I agree with you. It's just shocking, right? It's like fax, like fax machines. Like, who has those anymore? Chris O'Brien But welcome to the 80s. Mika Newton Yeah. So we get data in basically four different ways. And I'll start with the first way, which is, number one, you might just have some of your medical records right at home or on your computer as a file or something like that. So we originally started with, hey, send us what you have about yourself or what you can get from your doctor and just, like, email it to us and we'll process it. And sometimes that still works. The problem with that is, like, the provenance, like, where did that data come from and is it the right stuff? Is really hard to determine overall, as is, what is the latest data? Because now we're relying on, let's say, a patient or a family member to actually go and ask and if you've ever tried to get your records from your doctor's office, even, that's not an easy process. Chris O'Brien It's not easy. Yeah, right. Mika Newton Number two is we asked and talked with patients about getting their login credentials for their My charts or access to their medical records system. And often you have like with your physician, a way to just create your own username and password and you can see some of your charts. And then we could go in and essentially copy that data out of the background. The problem with that is you miss a lot of the underlying information. There's a lot of information in the medical record that's very hard to surface through those interfaces. So there's like data underneath that you can't really see. That's important. And it makes sense, right? Like even those interfaces are designed to help the patient, the doctor, understand what's going on. They're not meant to expose the entire system record itself. Chris O'Brien Yeah, that makes sense. Mika Newton So then the third method is the one we talked about, which is fax, right? Which is we send a records request to the records department. Then you get a return that could be like 1000 pages of fax. Or we got like a FedEx box right in the mail sometimes, right, and we have like the high speed scanner. Literally, my head of ops, I walked in the office one day in North Carolina and he's standing there feeding paper into the high speed scanner and then shredding it on the other side. Chris O'Brien Bananas. Mika Newton And that's still, by the way, a valid method. I will say it's now required, or will be in October of this year, for providers to provide electronic access to medical records. That's a change that's ongoing in the regulatory and then the way that we're most excited about are these health information exchanges, right? So, on a national level, hospital systems and provider systems, not just all hospitals have set up these networks that are mostly nonprofit entities, some for profit, but mostly nonprofit that exchange the data in common formats. So HL Seven and Fire for kind of the latest, those are just the jargon around them. But those records, like, if you can picture 1000 pages in a fax, you might get 1000 files. Right. And everybody implemented their medical record system slightly differently, meaning they have different fields or a different way of organizing it. So there's some standardization. The language is standard, but the organization is not standard overall. So that's kind of the landscape. Chris O'Brien And talk to us a little bit more about the health information network. So at that level, the data is standardized, or you still see variation. Mika Newton There's some standardization. So certain structured fields I'll use an example like your name and address or blood pressure or labs tend to be fairly standardized when you get them, they don't always show up in the same order in the document, right, or where they are, but you know what it is when it comes across. So I would think of that as structured data. Then there's what I would call semi-structured data. So this is machine readable in the sense that it's notes. So it could be like your doctor typed in some free form text or a nurse who was providing infusion care to you kind of listed out the infusion therapies that were being administered at the time. And that's actually a little bit harder, right? Because it's human language, but you know where it is because it's in a specific field like the treatment plan or the notes. And so you got a topic that's there and then it's actual text characters and then the last piece right, is there's images. So even in these medical records and when they transfer electronically, you will still see like a scanned copy. Let's take in cancer, Mika Newton Genomics reports are really important. So sometimes there will just be a PDF or a TIFF image of the report stuck into the file. So you kind of have to work through all of that. So when I talked originally, and we were talking a little bit earlier about being able to consume the facts, that technology capability exists or is required, even when you have the electronic feed capability. Chris O'Brien That you built early on, actually translates really nicely into the current, you still need all that. Mika Newton Yeah, and I don't think it's going away, particularly anytime soon. The health information changes are growing. I think about years or so ago there were about 400 nodes on the health information exchanges. Now it's over 40,000. Chris O'Brien Wow, that's crazy. I didn't realize that they had increased to that degree. So, okay, now we understand how you get the data. So now you have this data. You want to talk a little bit about how you read the data. How big of a task, how big of a problem was it? Let me frame it this way. When you think about the challenges that the company faces or face around data, are the biggest ones getting it, standardizing it, or interpreting it or something else? And I'm sure there's some degree of all the above, but how do you think about this range? Mika Newton Yeah, I was going to say Chris. Chris O'Brien Yes is the answer, right? Mika Newton Yes is the answer. So I think the getting your hands on it problem is getting solved, right? It's being solved by these networks, being solved by having this mixed modality where we can go out and get the data in different ways. There's clearly legal precedent now for it. And that's changed even three years ago. There was a lot of providers saying we're not sending the data even if the patient requests it. And they've just realized that that's not the case. So I think that getting your hands on it is an engineering problem now and that the data flow is out there. So that leaves kind of organizing it and structuring it as becoming the important part. And those two things are really important for each other. So we built a series of machine learning based algorithms, classifiers essentially. And this goes back to the fact story again. If I get 1000 pages of facts, what did I get? Like, was a lab report in there? Was it a notes report? What was the section? Was this one of these PDF files of a genomics report? And so one of the first things you do if anybody hands you a stack of paper, right, as you flip through and you say, what's in here? Mika Newton Like, and put some markers in it, what did I get? And so this document classifier or information classifier set of tools we built is really important. It just tells us what came over in this package. And the same thing applies to getting 500 XML files. What is in those XML files? And you look at the name, the date, the time, right? Then you actually look at the content and you can make some determinations around it. The second part of it is Natural Language Processing and Named Entity Recognition, which has been around for a long time. Just to be really clear, this type of technology has been developed in a lot of places, but it's highly specialized, right, in terms of oncology or a particular disease type. And so you want to train to it. And the key of having those two components together is you can start to take a Bayesian approach, right? I'll give you a simple example here of the letters VA. So VA could stand for the Veterans Association, it could stand for valproic acid or the State of Virginia, right? In which document and where it appears tells you an awful lot about which one of those three things it is. Mika Newton And so those types of conceptual model become really, really important and you build those. That's a big part of our proprietary approach is the way that we build those models, right? And it allows us to extract now that data and automatically database it without any humans into where we think the right place to go is. Now we'll call that real time clinical data. The regulatory grade piece that I mentioned before really happens if, let's say, we have a specific high value research question or we see something in the data that makes us think this might be submissable, let's say, for a label change. We worked with Critical Path initiative on a drug repurposing project. In that case, now we want to deploy the humans to actually go in and source verify the data. Meaning somebody's going to have to check each one of those data points, right? And frankly, it makes sense, right? Like the technology is there, it's great. But at the end of the day, we want to be absolutely certain. And the way to do that is still to do it the kind of human way. What's important, important here is we want to use technology to defray the cost essentially of human intervention until we know what we want to do. Mika Newton A big part of my career was spent trying to build the perfect database and I built a lot of data sources that were always, say, 80% perfect and 100% useless. They just did work, right? And the problem was, you never know what the question is that somebody's going to ask. And every research question, every drug development project, every whatever it is, everyone's different and has its own kind of idiosyncrasy. And so what we want to do is build the catalog of all the information we have, verify and organize it as much as possible using technology, and then deploy our most valuable and most expensive assets, the humans, when we know exactly what we want and where to get it. And that just creates operational efficiency. That's incredible. We're talking at minimally ten times human force factor multiplier and up to 20 to 25 is where we're seeing it going. Chris O'Brien Is one of them a harder problem than the others? One could say, hey, the thing that's a differentiating capability for us is blank, and that is ingesting that data. It is categorizing it, it is a human piece of it, or no, they're sort of equally weighted in your mind for difficulty of task. Mika Newton There's different complexities right throughout this thing, but it's the process. It's intertwining all of these things into a functional process. And you bring me to a really interesting point. I get asked all the time, Chris, like, lots of people have tried to solve this problem, right? But most have kind of not been successful, have really struggled to kind of get it to work at the end. So why are you guys different? What is it that happened? And I always say that it's and I thought about this a lot, it's because we had a directed reason to do this. Meaning we were trying to help a real person, a real patient, who had to make a critical decision with their doctor in a very concise period of time, meaning we couldn't take a month to figure it out. We needed the answer quickly. And in order to generate answers, we needed to build algorithms. So we had to build tools on top of the data, right? And because any tool that we're going to build has some specific set of inputs, like, we need to know some really specific pieces of data, we got a really clear answer. Mika Newton So going back to that, until you know what you want, it's really hard to kind of picture how to get it. But we needed it for ourselves. And that was actually our big problem when we realized we would have to build this ourselves, is nobody could give us just we're like...We need these 20 things right, from the data. Can you just give us that? Nobody could. And so that led us up into like, okay, we're going to build these algorithms. They're going to help people find the right clinical research programs or other disease management programs to participate in, or we're going to give recommendations to people around you and your doctor should consider these five combination therapies first, but we couldn't build those tools without the data piece. And again, I think there's kind of that whole thing together becomes the challenge versus any kind of single element. Chris O'Brien Hi, this is Chris O'Brien, host of Few and Far between. We'll be right back with this episode in a moment. I personally want to thank you for listening to our podcast. Now in our third season, it continues to be an amazing opportunity to speak with some of the top thought leaders in the clinical trials industry. If you're enjoying this episode, please leave us a review on Apple podcasts. It really helps people discover the podcast. And don't forget to subscribe to Few and Far Between so that you never miss an episode. One last request. Know someone with a great story you'd like to hear me interview. Reach out to us at [email protected]. Chris O'Brien Thank you. And now back to the podcast. Chris O'Brien Yeah, that makes a lot of sense to me. My question is a little bit unfair. I have certainly seen the sort of health technology and data landscape is littered with the corpses of smart, analytical people who thought this, how hard can this be? That truth is pretty hard for the kinds of reasons that you just laid out. You used an analogy with me. Can we talk about clinical trial matching? You use a playing card analogy there. I think that was interesting. Can you talk a little bit about what that service is and how you guys do it? Mika Newton Yeah, Chris, that's great. So enrolling patients or recruiting patients to clinical trials is a major issue. There's lots of ways to think about this. Number one, there are often lots of trials looking for the same patients, the same rare patients, and so there's competition from the trials to attract the right patients. There's another viewpoint as a patient, which is saying, like, of all these trials that I might be able to participate in, which I do yeah. Which one is the best one for me? And then there's also what I would say a mismatch between the I would say the inclusion exclusion criteria and there's some socioeconomic and geographic disparities. So oftentimes there's like, what I would call structural barriers to participation as well. And so we were looking at that and then thinking about... Chris O'Brien Will you double click on that? I think I know what you mean because of my day job when you talk about structural bearers and inclusion exclusion criteria challenges. But can you give us an example of that? What's a practical way that that manifests? Mika Newton Yeah, so the trial is going to open at this site that I'm at, but it isn't done opening yet. And it's in the IRB stages. Right. It's going through the paperwork process, and I just showed up too early right in the process, except I have cancer and I'm dying. And so really, the hospital's process for. Going through their diligence, which has to happen, is just not conducive to me participating in the study. I'll give you another one that I always find super interesting. I'm in one system and they're offering a clinical study in the hospital system literally across the street, let's say. But my insurance coverage is in network at the hospital I'm in and is out of network across the street. Now, sure, the study is going to be paid for, but the standard of care treatments and my copay associated with it makes it financially impossible for me to cross the street and participate in that study. Chris O'Brien Ridiculous and real, right? Mika Newton And there's lots of those, right? I would say too sick, too healthy, wrong place, wrong time. It kind of goes on and it's really unfair, honestly, if you look across the country. So kind of going back to then the matching, a lot of companies have focused on clinical trial matching. And when I think of matching, I kind of think of a card game like Go Fish. Like I have a queen, you have a queen. Go fish. Right. But in reality that's not the case because we know that while we're looking for, let's say this card, a queen, for the right patient, in reality, as we start to get to know the patient, even in the hospital, whatever, we don't have all the information yet. We haven't done what we were talking about earlier, aggregate and organize it. And so what we want to do is think about kind of two things. Number one, the trajectory or the amount of data that we have at the current time, right? And then the way that the data is changing over time as being kind of two dimensions of the patient's understanding. So the first one says basically, if I only know 5% of the things that I need to know, what is the probability that after I look at the other 95% that it's going to be a good fit or not? Mika Newton And we can build models that are essentially probabilistic search agents. So as we go from five to six to 7%, we're going to rescore the other patient. Chris O'Brien That's really interesting. So instead of disqualifying that person because we don't know enough about them, you're saying...At a certain point we can say, we have an acceptable probability that this person is going to qualify and maybe it makes more sense to then gather the rest of that data or target this person. Is that right? Mika Newton Yeah, you got it. Exactly. And some of that is us going to get the data and some of that is actually the data maturing. So that leads to kind of the second piece, which is if I run that score, let's say I come up with a score, it's between one and 100, right. It's not a percentage. So let's just think of it as a score. And right now it's like a 40. Right. And then next week, it's a 45, and the week after, it's a 50, and the week after, it's a 55. Now not only can I take that score, but I can take the trajectory of that score over time and say it looks like this person or patient is on their way towards becoming eligible for a study. So now we can start to address some of that structural issue which is like, hey, maybe you need to start looking or talking to your insurance company about this other hospital system because you can negotiate maybe some of those payment options. You might look at the fact that in order to be eligible for the study, I may need to discontinue one of my medications, right? Mika Newton There may be something I can do to make myself more or less eligible. And so starting to have that type of insight really lets us reach out to patients and try to solve the kind of last mile problem of the clinical trial space. The other thing is, what I just described is actually very expensive, right? And you know this Chris, from your day job, right, like getting the patient over that part and actually into the study and everything ready to go, that's really expensive. So why deploy those human resources again until you've consumed enough of the information from a machine basis? So essentially replacing all of the human kind of notetaking and full screen by prescreening the data on a continuous basis with these algorithms and tools. We're really excited about that. We're seeing right now when we apply that technology double to triple the rate of enrollment that you would get in the marketplace traditionally. Chris O'Brien Yeah, that's super interesting. And Mika, am I understanding it also correctly that the patient has some agency here so he or she can see that if I do stop this medicine or if I get into this other hospital, I could potentially join this trial? Something like that? Mika Newton Yeah. So number one, we give the patients back their data. So when the patients are on our platform, part of what we do, right, and we'll go all the way back to the discussion, right. Getting my own data is hard, right. So we just want to give it back to patients and the doctors and produce summaries of it so that it's more useful. Right. It's not thousands of pages. It's just like the most pertinent information at the top. So that information along with recommendations that are essentially machine generated. Here's some things to be thinking about. But I will say the final point is we still make that contact. There is actually a human contact to the patient in our relationship with patients and doctors. What we want to do is have this kind of informed consent and right to contact, which is if I find something that's really suitable for you, can I call you right, and tell you about it. And the reason it's important to talk to the patient is most of the doctors are so busy and most of the relationships between industry is with the provider system that you're trying to call the doctor's office and get someone to pick up the phone. Mika Newton And you're like, hey, I think this patient might match the study. Or can we talk to the patient? But you're talking to the provider and they just don't have the bandwidth. It's not their fault. They're just too busy. But the patient is at home or their loved ones are at home waiting. Chris O'Brien I want to get to decentralized trials, decentralized strategy. I know you've got a bunch of stuff to say on that, but before we go there, tell us what the business model is for the company. So you said, hey, it was important to make a for profit kind of business. So what's the path on that? How does the business work? Mika Newton Yeah, it's really a marketplace. So one of the things that we recognized early, right, was with the changes in patients right, of access to data, that the ownership of the data. And this is happening, by the way, all across society. The ownership is moving from institutions to individuals, meaning as patients and as participants in the healthcare system, we own our own data. And so we started scratching our head and say, okay, well, what does that mean? So great, I own my own data. What can I do with it? How is it useful? Chris O'Brien Congratulations. Now what? Mika Newton That's right, now what? And so we essentially allow patients who work with us and their doctors, they are buying our services, which is all of this organization aggregation in return for data rights. So think of it as, I can participate in this platform, but I'm going to have to put my data onto it and then execute. The company now has research and commercial rights to use that in appropriate full transparency, IRB approved, right, release and consent forms to commercialize that data. So we now provide that data in a variety of ways. Number one, as just insights so we can anonymize the data completely and provide it as data for all sorts of research studying. This type of data is consumed across the industry on a regular basis. Our data is a little different because it crosses multiple provider systems, right. All because of the patients at the center. So it's not just a single silo, it's all of the providers tied into a single longitudinal piece. And we still have contact with the patients. So if you needed to go get something else or talk to them, we could go act as your agent on the behalf. Mika Newton That's one piece. The second piece you started to talk about a little bit, which is actually a decentralized study. So if you think about one of the big things in any research project is the data, right? The product of any research study is the database that gets created at the end and so that means that somewhere out there are a study coordinator sitting in a research site, typing the data out of the medical record into a CRF form or electronic data capture system. And then somebody else is probably flying out a CRA from the CRO or whoever it is, to verify that. So we can centralize that part of the process. And that's a scarce resource right now. It's hard to hire study coordinators. It's hard to get the Cras out in the field. And we centralize that data and keep the source documents from itself. And that means we can provide the backbone for these very decentralized studies. Right. And we've got, I think, almost probably ten or 15 of them in one form or another now up and running. And they range from like a biomarker study where, let's say we want to do serial genomics on a series of patients with a certain category. Mika Newton So we can actually deploy either sending test kits or mobile phlebotomy to the patient's home and collect the data, and then they can continue being seen in whatever system they are. They don't need to go to a different site and we just collect the data, but then we can verify all of it to fully decentralize. Chris O'Brien Yeah, that's super cool. I think when people are thinking about Oncology trials, there's a tendency to assume you have to be at major institutions, and for some trials you do. But I think your point about that we can take a mobile phlebotomy, can come visit you at home. I'm sure that's a pretty big selling point in some trials. Mika Newton You know, to make life better, you don't have to do all of it, Chris. Right. Like, there may be some important visits, right? Or key points where you need to go to the specialized center, but why do you need to fly there for every single blood draw? And sometimes it's like flying. Like, traditionally you would have to fly to the site often for something that could easily be delivered to you in a home or community setting. Right. And the reason you were going there was to ensure data integrity and quality and oversight of the process. Again, these are really important topics. But the question becomes, can we use technology and a different approach and reduce the burden on both the patients and maybe allow their physicians, who they're seeing anyway, to have a greater role in participation in oncology where we have thousands of studies and most of them are desperately trying to recruit patients? I would say desperate is probably a pretty accurate statement for that's, right? Chris O'Brien Yeah. Mika Newton This ability to open up and interact with a larger number of patients means a lot. It means a lot for the patients who many times their best option is research. Right. And they're coming there for a care modality. They're not just purely altruistic. Right. They're hoping that this trial will help them with their disease. Right. And do these drugs actually work or not. The reason we're doing the research is we don't yet know whether these are medications or not. And the sooner we can eliminate the ones that aren't going to work and identify those that are most promising, the faster we're going to be able to address, treating and eventually, hopefully, curing cancer, if that's possible. And so we all benefit from accelerating this drug development pipeline. Chris O'Brien So it broadens the reach. It reduces some burden on the patient. And I would assume it also probably reduces cost in some ways, because if you can do some of these things remotely, you ought to be driving some cost out of the system. Mika Newton That's true. So I think the cost on the per patient, I actually think the value is there's some in the cost savings of the individual, how you get there, but the costs are spread to the patient, the insurer, right? The sponsor, that's kind of all mixed together. There's some cost. I actually think the big advantage here is it's actually accretive to everybody. Right. So I think of, like, the positive effect of this. Right. So the sponsor is getting the answer to their data faster. Right. And they're moving their program along quickly, which is what they've usually funded. Right. They've got funding for or they need to know whether the drug acid is going to move forward or not as quickly as possible. We think of the major academic centers where a lot of the trials are run, the sites themselves. Well, they for the most part, have exhausted their patient supply. They do a really good job with the patients they have in these clinical research institutions. But that study they have is still open. Right. And so if they could get more patients into the study and had a way to reach outside of their own walls, their bricks and mortar, they could actually run the study faster. Mika Newton Right. And that's a revenue generator for those sites. That's why they provide research services and then the actual community. So the patients who are outside of that network, their physicians, are now getting a new set of services they could offer that they weren't able to offer before. So it's actually good for that community practice. So this is one of those rare situations, I see, where there's like, it's win win, right. Versus usually there's a loser in this whole thing. Chris O'Brien Usually, yeah. Mika Newton But so far I haven't found one. When I think about where we are, this is all possible technically. Now, I do think as an industry, though, we're still kind of hesitant to really take the full leap. Right. This is new for many people. An idea of thinking this way is and is it going to work? And am I putting my drug development program at risk? Is the FDA going to come back and say, you did it wrong, start again? Right. That sort of stuff. I think there's lots of indications, none of that's true. And so I think we're just at the early stages. This is going to be a revolution that takes us through, I would think, the next five plus years here. Chris O'Brien Let's talk a little bit about digital twins and synthetic control arms. Many of our listeners may be familiar with these, but will you tell us what they are and why this is something that you're excited about as things move forward? Mika Newton Yeah, so I think we've often thought right, I think of particularly synthetic controls. Right. Or if you're a cancer patient and you're going to enroll in a study, one of the things you really worry about is randomization, right. When you still used to have these, either standard of a care or placebo controlled, not really placebos anymore in oncology, certainly. But the idea here is I'm going to participate in the study, but I'm not going to get the new thing. I'm going to get the old thing. Chris O'Brien I got the old thing. Mika Newton I got the old thing. Chris O'Brien I don't know what I got. It may turn out that I got the old thing. Mika Newton That's right. And so why would I do that? Because I already know my cancer is probably going to kill me. Right. So this is not an option. I don't want to go and do something where I'm kind of rolling the dice on that. And so because of the difficulty in enrolling patients and the ethical considerations of that of right. I'm actually going to give somebody nothing on purpose when I know that they're dying. It is also an ethical issue for many researchers, the idea of these. And we see a lot of these single arm studies that are coming through. And so anytime you run a single arm study, of course the question is, well, what do I compare it to? So everybody got one thing, and what's the comparison? And so one of the things we're starting to explore is the idea of, like, can we capture from the patients who didn't want to participate in the study or couldn't or just weren't in again, the other folks who are not part of the study, can we match the same or do a match control, essentially from the general population? And a lot of people have looked at this right, in terms of, let's say real world data, which I think is a little bit different for the most part right, of anonymized records, and can we create a statistical match from large data sets? Mika Newton And there's actually been approval in that space. Right. I think it was I remember like Ibrance for male breast cancer, I think the first one that was done that way. And so you can do it statistically. The problem with a lot of that kind of anonymized data is it's from either individual providers, so you don't have the whole history of the patient or it's not fully longitudinal right. In the sense that you truly know what happened to individuals and can go back and kind of source verify and validate. You just don't have the source documentation. So a lot of what we think there is and I think this has been in the agency's latest guidance, even we want to look at this data like it's clinical trial data, meaning you really need to structure, you need to have a CRF. Right. You need to have a management process for extracting the data. I think that's really exciting overall. And if we do that right, by the way, and we get all the data, and I think this is a real opportunity for us, the comparator can be other data that we have. So we try something new, and we have a repository of comparators right, that we can assemble appropriately to do that. Mika Newton And inherent in that is this idea of a digital twin. So I get excited about digital twins. I was just talking to a colleague of mine who's got a great company that works with millions of patients on almost a social network sort of basis. And the thing that they try to answer, which I think is really common question, right? And I do this every time I go to a restaurant, right? Which is, what's the review? So what did somebody like me do and did it work or not? I mean, I don't buy anything online now without reading the reviews. Why would I? But we don't have that in healthcare, right? And we don't have that around treatment. And so when I think of this idea of a twin, I want to know what everyone who was like me, which is a definition, right? My twins, what did they do? Did it work or not? And then it'll help me make my decision. And it's exactly like the match control in the study, but it's actually now like the real life, everyday application of this idea. Yeah. Chris O'Brien Look, it's a lot easier for me to get data on the quality of the Tacos at six different Taco restaurants in my neighborhood than it is for me to assess different medicines that I could take. Mika Newton Right? Yeah, absolutely. And we've gotten all very used to it. Right. So it's coming to healthcare. I really believe that, Chris. This is going to change. Now it's just a question of how long and how are we going to do it. And there's a trust issue also in this. Right. So we talked a little bit earlier about this idea that data is going from institutions to individuals. The amount of misinformation in the marketplace is also explained. Just look at what we went through with vaccines, right? Chris O'Brien Yes. Mika Newton And everything that's happened here. So as a consumer of data, I love your example of the Tacos, right? You're going to read who wrote this review, and then you're trying to figure out, was this the owner of the restaurant or their competitor, right? Like all this sort of stuff and I don't have an answer to this part. I think it's something we're going to have to solve and we should think really carefully about like how do we separate misinformation from real knowledge as this data becomes more and more widely utilized and applicable to individuals. And it goes all the way back down to clinical research, which is if I'm going to say this is the best trial for me, how do I know whose interest is it? The sponsors telling me I should definitely do this study? Is it the doctor? Is it because the doctor can offer that study? But across the street they have a different like all of these kind of conflicts need to be kind of worked through. Again, don't have an answer, but I think we're going to be in a very interesting time here as a lot of this data starts to flow around these decisions and who influences them. Chris O'Brien Super exciting. Okay, that's going to bring us to our sort of final section here. Perfect. Segue, what's next for xCures and kind of where do you think the application of AI in oncology is headed? So get your crystal ball out, tell us what's coming. Mika Newton So I can tell you what's next for xCures because I think about all day and then I'll speculate on some of the stuff that's going on in AI. So we spent basically four years building the platform. We are now putting it to work with partners. So we are working with biopharma sponsors, right, with CROs, with hospital provider systems. And actually I should do that providers. Because sometimes providers aren't who you think they are, right? They're not necessarily a hospital or a doctor's office. There's all sorts of providers of healthcare services that are covered entities out there. And basically anyone who interacts with patients is an opportunity for us to basically data enable. So just think of us like you're doing whatever you're doing, providing services to healthcare space. You want to make it have real time regulatory grade data, stick X cures on the side. And we need to figure out a business relationship. We've got a lot of that. I think that we've built enough now that we are putting it to work and putting it to work at scale. And that's actually going to lead us to a lot of learning around the AI. Mika Newton We'll continue to refine the algorithms that I talked about overall and they'll get better and better. I'm sure just knowing the people that I work with, give them a month and they'll think of some new, even more exciting way to apply that. So in terms of the future of AI and oncology, there is no way that oncologists are going to treat patients without access to modern decision making tools. And the issue that it's faced, right, is this information retrieval issue that patients have, the doctors have it too, right? Which is if there are new drugs that are coming out all the time and being approved and new combinations and new research studies, et cetera. There is no way that a physician can keep track of that. I did like a basic pubmed search, right, on cancer treatment as the search. How many articles? Like 150,000 plus articles per year. Even if you said only 10% of those were even worth reading, you still can't read them, right? You still can't read them. So the physicians need the help not necessarily to make the decision, but to identify the relevant knowledge. So it's this information retrieval that I think is really, really important. Mika Newton And in order to do that well, we need to really move the data interoperability which is coming. That's why the health information exchanges are coming together. So thing that I'm super excited about on the exchange side. So currently, I think you remember when we first got email used to have to pull down on your phone, right? Or click refresh and you would go and check your email, right? And now your email just shows up, right? Chris O'Brien Yes. Mika Newton So the exchanges are currently working on the like retrieve it. So we have to go say, hey, is there anything new? Tell me, is there anything new today? Chris O'Brien Got it. Mika Newton And they're moving towards a world where if something happens to you, it gets pushed out to you can basically set an alert and that gets really important. So like, let's say I'm here in the Bay Area, right, and I'm under care, and then decide to go down to San Diego, right? I want to go to the beach and go surf or something, right? And something happens to me down there and I end up in the emergency room. How does my team and care team up here know what happened to me? Or if I'm in a study, right? This could be an adverse event or something. So that type of push notification and how to utilize it in all of these things to alert care. So that set of tools is going to be, I think, really exciting for treatment and care. I think that's where the future of healthcare is, right? So it's retrospective know what happened in the past. It's prospective like let's make decisions and learn from them and then it's real time. Meaning when something needs to happen, it's not because somebody happens to go and look for it, it's because the system itself is surfacing the data. Mika Newton And I think it's a huge opportunity to just transform the way care is delivered and frankly, to make the humans more efficient. We don't have enough in oncology, we're short 2,000 plus doctors by 2025. Where are they going to come from? And I know that's true across most complex rare diseases, there simply aren't enough experts and providers and so they need this tool set, which is why, frankly, there's so much interest in it across the industry. I think a lot of people see this coming. Chris O'Brien That's fascinating. Mika Newton, thank you so much for sharing what's happening at xCures with us today. Really exciting. And I certainly leave this powered up and motivated. How can people get in touch if they want to learn more about what you guys are doing? Mika Newton Sure. Thanks, Chris. Check out our website, so www.xcures.com. You'll also find us on Facebook, LinkedIn, TikTok, Instagram, we're everywhere. And you can always reach me directly at [email protected]. Chris O'Brien Fantastic. Mika Newton Okay. Chris O'Brien Mika Newton. Thank you so much. It's a great one. Mika Newton Thank you, Chris. Really appreciate it. Speaker 1 Thank you for listening to the latest episode of Few and Far Between. Our podcast is now available on Apple podcasts and other major streaming services. Please take a moment and leave us a user review and rating today. It really helps people discover the podcast, and we read all the comments. Those comments help us to make Few and Far Between better and better. Also, be sure to subscribe to Few and Far Between so that you don't miss a single episode. Got an idea for a future episode? Email us at [email protected] or contact us on our [email protected]. I'm your host, Chris O'Brien. See you next time.