BẢN CHÉP ÂM THANH 🔗
David: Welcome back to Beyond the Scope. Our guest today works at the point where digital pathology, AI, and precision oncology meet. Karin Aurora is Senior Vice President of Advanced Assays, AI, and Pharma Services at Leica Biosystems. He leads a global business unit focused on bringing digital pathology and AI into the drug development and companion diagnostic space. Karin joins the show. Take a deep dive into how digital pathology is accelerating drug development. Beyond the scope is brought to you by the Digital Pathology Association. You can visit www.digitalpathologyassociation.org for everything you need to get connected to the digital pathology world. This episode is co-hosted by Ale Kalam and is produced by David Romedios. It's been quite some time since we've had a guest from the drug development space join the show, and I was fascinated by Karin's insights on digital pathology's role in big pharma. We also take an interesting direction relating to this idea and how it could impact rural healthcare. And finally, if you like the business side of digital pathology, then this episode is for you. So let's get to our guest, Karin Aurora.
David T: All right. Hey Karin, how are you doing today?
Karan Arora: Great, David. Nice to meet you.
David T: Yeah, thank you so much for coming on. where are you calling from today?
Karan Arora: I'm actually at my home in in in the north of Chicago. So yeah, l excited to be on this conversation.
David T: Yeah, thank you. We're we're gonna cover a lot of interesting topics. You're a you're a healthcare investor, innovator, have some really interesting perspectives on companion diagnostics. But first, tell us about your background. How did and how how did you end up in this space? What attracted you to the digital pathology world?
Karan Arora: Yeah, I know. Thanks for the question. Look, my career has really sat on the intersection of like diagnostics, biopharma, technology, and commercialization. And I've had the opportunity to work across diagnostics and pharma. So I've seen Precision medicine from both sides, developing therapies, and then confronting the very practical question of how you actually identify the right patient for the right therapy. This is ultimately what drew me towards digital and computational pathology. you know, pathologists sit at this sort of extraordinary intersection. you know, you have the physical tissue, the biology, the morphology, the biomarkers, and increasingly a very rich digital data. Asset all coming together in one place. What has changed for me over the last several years is that you know I no longer think of digital pathology as replacing the microscope with a screen. You know, digitization is really the infrastructure layer. The real transformation begins when tissue becomes computable. You know, and at Leica, that is increasingly how we're thinking about the opportunity, connecting the physical tissue. workflow with advanced assays, digital pathology, AI and companion diagnostics, so that an insight discovered during drug development can ultimately be delivered reproducibly to a patient anywhere in the world. And you know, I think to sum it up, like to your point, what really excites me and what drew me here is the slides becoming becoming digital was sort of chapter one. And the slide becoming computable is chapter two. And you know that that's really The exciting part about being in this role now, you're writing that chapter two. and you're right at that inflection point. So here I am.
David T: Yeah. When we when we get to the end of this, I wanna I wanna get your thoughts on what do you think chapter three is. Cause it seems like we're in the middle we're in the middle of chapter two and it's rapidly evolving. But yeah,
Karan Arora: Yeah. We're in the middle of it, yes.
David T: on on your on on your career. It sounds like you didn't start your career like in in the pathology world. You started it kind of somewhere else and then it let and then just because of the exciting things that are going on in pathology, it kind of your expertise that you learned in in that early part of your career drew you to this field.
Karan Arora: Yeah, great question. Look, I started as a consultant. You know, I spent you know, if I break my career down, I spent seven years in consulting between CapCrim and I and McKinsey. I spent another decade in pharma and then another decade in diagnostics. And, you know, what happened naturally is as a consultant back then they used to call it technology enablement. And I was working on strategy projects in McKinsey where I had a background in biochemistry, I had a master's in technology and an MBA, and they were like, Well, you understand the vernacular, so we're gonna bring you into these technology projects. And I was not a technologist, I understood the vernacular and through those consulting engagements did technology enablement work. And then when I got into pharma and slowly they were going into clinical diagnostics and algorithms, doing dissemination of therapy at pumps. they were like, and we need somebody un who understands computers and technology. So I got put in those roles. And that was the early beginning of digital. And subsequently, you know, those roles led to me leading digital businesses and then my last role, AstraZeneca, leading precision therapeutics. And I sort of became the intersection point of drug development and the biology, understanding digital, and then sitting in the middle of running those businesses. So that's how those three things came together. So I think Just interesting roles that built on one another and and got me to this point of really at the intersection of digital and and biology.
David T: we we in in pathology with what's going on with digitization and and the next steps, AI, computing, and and the future after that with like all leading towards like very advanced precision medicine. I I can tell you firsthand, we we really need
Load full transcript
people like you with business acumen and experience to be in this community because I I think there's a lot of places that are they want to adopt digital technologies. And we're just talking about we're just talking about slide scanning and maybe some AI stuff right now. But it seems like we still haven't figured out the business model behind it.
Karan Arora: That's right.
David T: So like from your from your perspective, what what are what what are the what are the challenges that that that you're seeing and maybe if you have any insight on how we can overcome
Karan Arora: Yeah.
David T: them, that'd be interesting to learn.
Karan Arora: it's it's it's it's a good question. I think look, people look at digitalization in in many different ways, right? Some say, look, it's an operational efficiency play, I can do more with less. some look at it as a data play and they're like, Well, if you can get all this data, we can do so much with it. And do you see platform players out there that are that are building, data assets around it. some see it as like, Okay, this is where the modern world's headed and we want, not to be in the dark ages, we want to Be computerized, want to be digitized. So they do it as a capital investment, thinking that this is what they need to do really without understanding what the benefits are. So I I think you have people at different points of
David T: Yeah.
Karan Arora: of the graph. And you know, in in my mind, you know, personally, I I view it a little bit differently. You know, I think, look, the world's shifting from, you know, where you're just thinking about digitization of, you know, generating a slide. That you can see. I think what you have to shift your mindset to is a slide computable? And can I now connect this with other data sets that generates information that lets me get therapy to the right patient at the right time in the fastest way possible? And and I think to do that the right way, you have to think about the entire workflow of the system and not just a digitization element. I think it starts all the way in the beginning. of drug development and doing clinical trial enrollments, to also thinking about, you know, how you actually digitize slides, you know, like, you know, are you using glass slides? Are you putting tape? You know, what is your pre-analytic conditions look like? What is your staining conditions? How do you calibrate the scanner? So there's so many components involved. So if you truly want to digitize and you want to extract the value to the patient, and you know that should be our North Star always. Then you have to look at it end to end. And I think as you start to piece that workflow together, that's when you realize that the value of digitization and how it can reduce variability, how it can empower people to deliver better care in any place in the world. you know, how you can actually democratize access to the best medicine, how you can find patients for the right therapy the first time. So th those are the benefits and you know, why I think. that matters for us to focus on digitalization and and and how we need to think about it differently.
Alae: Karan, Alet here. I wanted to ask
Karan Arora: Yeah.
Alae: you a quick question on, basically what a lot of what you're saying. What do you feel like is are or are barriers to this digitization I guess project or you know, it's basically the future. you know, so I'm a pathologist and I find that there's a couple patterns that get in the way, but I'm very curious to hear what your perspectives are when it comes to barriers to digitization.
Karan Arora: Well, I think the simple one is capital investment, right? most pathology labs are under resourced, and you know, and I think it's a huge cost to build in digital infrastructure, right? The cost of the scanners, the cost of the IMS system, you know, so that's barrier number one. I think barrier number two is reimbursement. if you look at pathology labs, you're still trapped in an era where manual reads are reimbursed and you get a minuscule you know increment in reimbursement for digitization, and then there's nothing for computational. So there's no incentive, you know, for the hospital to basically say, why am I investing in this? I don't make more money. and by the way, that's true for any IVD therapy, right? If you have a class one. you know, assay in the US versus a CDX, the reimbursement for the pathologist is the same. And you know, some people are like surprised by that, but that's the reality. you
Alae: Yeah.
Karan Arora: know, even though you're talking about a class three assay that may be computational versus a class one manual read assay, you know. So I think that's your second issue. I think the third is, you know, boils down to training. you know I I think you have a lot of technicians, you have a lot of experience that has accumulated over the years. And, you know, if you look at the pathology schools or you look at the training curriculums, they just haven't moved fast enough to incorporate like, okay, how do I train, you know, staff on digitization? What do they have to look for in a workflow perspective? What causes drift in a system, you know, that may lead to errors when you digitize the site and use computation to read it. I think those are probably your three biggest barriers. we have to solve for those first. you know, I would say the first two are mostly addressable by the business economics, and you can work with the diagnostic companies and the hospitals to figure out what are the right business models to alleviate the capital constraints. I think the reimbursement is gonna come as clinical utility around these domains develop. And peers and the CMS starts to see value in it. I think the third one is a little bit of an industry calling. you know, one, I think young pathologists and technicians have to draw a pull towards digitization and see the value in it and how it frees up time for them to actually
focus on cases that really matter and helping those patients. And and then two, the industry I think collectively has to help. bring the industry to the next level of training and education and and I don't think we're doing enough there. So
Alae: Yeah, it's interesting 'cause from my perspective as a pathologist, I always tell people like other pathologists, D digital pathology is the future. there's going to be a point in the future where the cases that you send out to other organizations are no longer gonna accept glass. They're gonna say, Okay, send us the slide, we should have it within, by the end of the day or whatever, right? and from my perspective, a lot of people really do cite what you said, which is the reimbursement. It's like, well, I'm not really making any more money off of this if I'm digitizing, like so what? so I you know, from a pathologist perspective, it's definitely something to address. I don't know how much traction we can bring when it comes to the reimbursement, or you know, how we can change that. Maybe that's a unmoving factor. But another thing is you know, like there is A lot of pathologists do want to go digital, but I think you mentioned this briefly is like they're overwhelmed with what they already have and they're severely understaffed. So it's so difficult to take on another project and to validate another thing and to have to go shopping essentially for a scanner and figure out th the the actual physical being of going digital. So it you know, as somebody who's in the trenches, right? Yeah, it definitely I'm glad that we have like some shared perspective on that.
Karan Arora: Yeah, and I I think what I'd leave you with there and and the community is like, look, any change is difficult, right? And you saw that in in the automotive industry. You went from gas cars to electric cars and you know, everyone was like, my god, like how will I charge a car? How will I go across country? You know, like what do you mean there's no key to start the car? You know, like it's like yes, there are concerns, they're legitimate, but it's change. And you know, I I think you Have to put it in perspective. And for me, again, bringing it back to the patients is really where the value lies. it's like, imagine today, right, with all the variability in PDL1 testing, you still miss five to 10% of patients that should be getting Katruda or any different therapy that relies on PDL one. And you're like, what if I was able to find those patients, get that therapy to them the first time? Imagine the impact you make to their lifestyle, right? And and that is what you have to keep front and center and say. Yes, I'm not getting reimbursed, but it's the right thing to do. And you gotta
Alae: Yeah.
Karan Arora: bring it back to first principles.
Alae: Right. You're right. And and it's interesting because it's as a pathologist we recognize our limitations. And I think that from the, bird's eye view, the leadership perspective, this is like they think that we function at a hundred percent capacity all the time, meaning
Karan Arora: Yeah.
Alae: that we're that there's no room for error, which isn't true. Not we try to minimize error as pathologists, but there's always gonna be something that maybe clinically won't make a difference in, Maybe like you might call something a certain diagnosis that might be similar to something else, but clinically will make a difference. But the reality is is that when you have very powerful tools that you can integrate with digital, like AI, it becomes very it it it standardizes a practice that's not necessarily standardized, to be honest with you. And and it really is like an extra layer of protection, especially when it comes to the things like PDL1. So yeah, I I'm I'm I'm glad that you bring that up.
Karan Arora: Yeah, and it's not actually it's not even the experience, right? It's just the load and the volume and the repetition that leads to it. Because it's so easy when you're so well versed in something and you've been doing it for so long to not see the subtle changes. And I think that's where with volume, like computation really helps. I I think it catches it it it doesn't get tired. That's one. I think two is that it applies the same consistent rules. in a standardized way on an ongoing basis. So it reduces that variability. So in the case of PDL1, it's not looking at binary cutoffs. It's looking at is there any presence? And that should trigger therapy. It's that simple.
David T: Karan, I I wanna I wanna close out this reimbursement conundrum talk with with something very important, very current. So we've had in our space, we've recently had some some major acquisitions, right? We've had
Karan Arora: Yeah.
David T: Path AI, Page, and others get acquired by by major companies. And they did that without they they developed AI for pathology. They didn't get an an added reimbursement for it.
Karan Arora: Yeah.
David T: So what what are these large companies? What do they actually buy?
Karan Arora: look, on the from the outside you would say, look, data is really what they went after, right? And and you know, the play was really around appealing to biopharma for early discovery and you know, biomarker selection. But but I think, you know, three things have happened, right? And and we've lost sight of it. I think one is that digital pathology infrastructure has finally reached sufficient maturity that you can deploy it at meaningful clinical enterprise scale. And they they never looked at that equation in most cases, right? It was like, okay, we're gonna use it for discovery and translational medicine, but the clinical equation will figure out when it comes. Well, the reality is it's come, it's here. I think second AI has moved from demonstrating interesting algorithms to, you know, sort of potential utility across workflow. biomarker quantification, drug development and clinical decision support. So it's no longer, a siloed application, connecting the
dots through the continuum become important. And I think that was missed. you know, and and now people are thinking about it. I think third and probably most important is companies have realized that the value isn't the algorithm in isolation. It is the ecosystem around the algorithm. Right. So what I think people are buying back to the question you're asking is like I would say is three or four buckets, right? one is data. Second is workflow. I think third is evidence. Fourth is distribution. And I think the players that are going to be successful are the ones that are open, best of breed that can bring those components together. There'll be no single player that will win it all. There is too much you know I think depth required by modality by disease that you know want It's almost like you want to democratize the ability for millions of people to work on the problem, but you want to aggregate all that know-how and knowledge in a unified workflow and ecosystem on which it can be deployed. So I think companies that go with that mindset of bringing the community and the ecosystem together and being able to deploy it in a unified manner, you know, to scale clinically and drive value for patients are the ones that are going to win. And I think that, you know, you're asking me what chapter. three looks like. that's sort of where chapter three is going, right? You're going from the data to the compute to the ecosystem.
David T: Okay, so you're in this companion diagnostics space, and
Karan Arora: Yeah.
David T: we're hoping we're hoping that, you know, these acquisitions probably gonna lead to a lot of new companion diagnostics, right?
Karan Arora: Yeah.
David T: Okay. Why don't we start from the beginning and just define what a companion diagnostic is for the audience?
Karan Arora: Yeah, look, at its simplest, a companion diagnostics answers one of the most important questions in precision medicine. is the therapy appropriate for the patient? That's what it is. But, we are increasingly sort of designing medicines that act on very specific biology. you know, if you cannot reliably identify the patients whose tumors have That biology, the promise of the medicine can fully translate into clinical practice. So, I see it a little bit differently. A companion diagnostic is not there for just a simple lab test, it is part of the therapeutic strategy. for some therapies, it identifies patients likely to benefit. In other situations, it can identify patients at particular risk for treatment. And the diagnostic and therapy become interconnected in how the medicine is used. So my tagline on this one is, precision medicine is only as precise as our ability to identify the patient. So, you it's like part of the flywheel. You cannot do without compatibility diagnostics anymore.
David T: Ale, what what do companion diagnostics mean to you? How do how do you how do you use them as a pathologist in your practice?
Alae: I was actually gonna ask Karan to go over that.
Cause I think it means different things. I honestly have not you you know, I
David T: Well
Alae: I'm I'm not familiar.
David T: Well the the then the point is is okay, how do we bridge the companion diagnostic world to pathologists, right? How do we do that?
Alae: Y yeah.
Karan Arora: that's so interesting, Ali, that you you you said that you're not familiar, So I I think it's think of a needle in a haystack. It's right, how would you go about finding it, right? And and that's the analogy I give to companion diagnostics. you have amazing precision therapies launching, ADCs, bi-specifics, multi-specifics, car T therapies, and for many cases, if you look at large cancers like lung and breast, you have several different therapies available for patients within that same category. So then the question really becomes, what is the best therapy for that patient? So you know if you boil that down, that's really what companion diagnostic is. So as as farmers developing those therapies, they require a test to enroll patients in onto those clinical trials. And they usually develop tests that identify the presence of those biomarkers because that's the single marker that says that that therapy will be beneficial to them. So all the way from therapy development in identifying patients where the trials actually show success with that therapy, now you're bringing that same test in a clinical setting to say, okay, well, the clinical trials, which were in a controlled environment, proved that this therapy worked. How do I identify those same patient types in a clinical setting? And that's what
Alae: So
Karan Arora: companion diagnostics is. It's it's the diagnostic that is specifically tied to that therapy.
Alae: So in a sense that I'm familiar with like PDL one and K Truda. That's that's okay. So I know what it is.
Karan Arora: Yeah. That's a companion diagnostics.
Alae: Okay, so I know what it is. I just didn't know it was called companion diagnostics. Okay.
Karan Arora: Yeah.
David T: but but but Ali the c the question is is okay, so you so you do you do a PDL one read. And I wanna know, d do you what your your read on that determines if a patient is gonna get prescribed PDL one, right? So Ali,
Alae: Essentially.
David T: yeah, do you do you want to have a say? I mean this is your chance as a pathologist to have a a say in in tr in treating in what treatments patients get.
Alae: Yes.
David T: Is that something that you want?
Alae: Honestly, if an AI algorithm could do it on my behalf and make it extremely reproducible and accurate and very precise, that would make, my life and other pathologists' life a lot easier. Because if you call something like negative and it completely excludes a patient from getting treatment, then that's a problem. I think in the breast world there was similar things like the ultra low expression.
Karan Arora: Yes.
Alae: So that that's that's you know so one pathologist might call it negative, another person might call it like focally
weak positive, right? That determines whether that person gets just by that decision. and the reality is is like there is interobserver variability where one person might call it one thing and another person might call it something else. And that actually, you know, depending on that pathologist or where that, you know, person or the hospital or whatever it is, that either determines whether the person gets it or doesn't get that treatment.
Karan Arora: Yeah. Yeah. And you're
Alae: So
Karan Arora: you're pointing to something very interesting. You're talking about the HER2 biomarker. And it's you know, I think where things started, people were like, Look, you know, you have to identify the concentration of the biomarker at a certain level before you're eligible for therapy. Then as you know,
Alae: Okay, okay.
Karan Arora: the signs grew and the evidence base grew, it became like, hey, any biomarker present presence. and that morphology actually should dictate that the person get therapy because it's positive for them. and the systems today and the workflows today is aren't designed for that. because now it's like im imagine, a collage that is all you know gray and there is gray dots made of gray dots and you have one black dot. And you're being asked in a matter of like 45 seconds to find that black dot. And if you miss it, there's no therapy.
Alae: Yeah.
Karan Arora: That's what's happening.
Alae: Exactly.
David T: Al Ally, I'm also thinking about really complex cancer cases. And I'm thinking about your go back to your academic days. How much how much time did you spend in tumor board with oncologists, surgeons trying to figure these things out? Right. And what if we
Alae: We try to
David T: had more what if we had more specific companion diagnostics that could reduce some of that burden?
Alae: So this is interesting because it's a funny question because everyone tries to keep tumor board to an hour. So sometimes it'll be an hour and a half because there are some very complex cases that we actually have to discuss. I think from a pathologist standpoint is to get to as accurate a diagnosis, you know, as soon as possible not as soon as possible, but try try to get to a very accurate diagnosis. Now if something is very obscured or the morphology is funky and we do a bunch of ancillary testing and it's like, okay, this is this vague thing. sometimes we kind of let the surgeons and the patient decide what the best treatment of therapy is, but that doesn't mean that we didn't miss something, right? If if it got had gone for like more testing or whatever it may be, you know, like some genetic mutation that I I mean usually we'll do it anyway, right? We'll usually just be very thorough if it's something very weird and doesn't look like it read the textbooks. It's it's it's an odd morphology. We'll do our best to diagnose this and be like, this is what it most likely is. and then it becomes a clinical decision on the surgeon or the oncologist and even the patient has to be included in that conversation.
Karan Arora: Yeah. And and David, just think about what Ali said, right? so for these complex cases, you're not just looking at one modality. we were talking about IEC for the last 15 minutes. Now you're adding molecular, right? You may have HE, you know, you have flow. So, you know, the the pathologist being asked and that tumor board's been asked to take, you know, multiple data inputs, you know, and then figure out a diagnosis. So the complexity is only growing. And the variability comes because the experience levels around these tumor boards vary. It's not the same people making the decisions, right? And then so you have that's one challenge. And then second challenge is like if you just think about the number of cases. So it's estimated by 2050, there's I don't know, like what, 35, 40 million additional cancer cases that are going present themselves. Just look at the curves going up, right? So cases are going like this, pathology ramp up is not going like this. So you just don't have the people. So so you run into two challenges. Complexity of of you know decision making is becoming harder because you're trying to find these obscure patients because these therapies are getting so personalized. And then two, the number of cases are going up. Like, look, if you're not putting digital and you're not thinking computation, it's like you can't do it. It's that simple.
David T: I wanted to I wanted to get to this major global problem. Rising rising cancer,
Karan Arora: Yeah.
David T: declining physicians, specifically pathologists. What do we do?
Karan Arora: Yeah.
Alae: you invest in digital and and and integrate AI, that's what you do. Come on, David. We already know the answer.
David T: Okay. No. I
Karan Arora: I mean
David T: I wanna hear what Karin has to think.
Karan Arora: I mean it it certainly makes a very strong argument, I would say. You know, I think and you know, I I I think also in fairness, like look capacity is just one of the issues, right? You're the other two issues we talked about, Ali pointed to one, which is the complexity of the diagnosis. I think AI can assist in that, just dealing with the computation of the varied data sets and being quickly being able to focus that tumor board on the actual synthesis rather than combing together the data is a more valuable brain exercise. And then I think the other piece is also removing disparity, right? today you have patients in rural care settings in the US, like a first world nation where, those patients don't present themselves. Why? Because drive time so long to a hub, you know, where the expert oncologists and pathologists sit, they're like, I'm just not gonna go. And then if you go to the rural hospital, they don't have the experts. So just imagine for a second, if there was a technician there, all they did was take tissue and scan it. And that image went through a hub, let's call it like a, any big academic clinic where you
had an expert looking at it and going, I made that diagnosis and it goes back to the rural hospital and the patient didn't have to go anywhere. that's great service to the patient and that's advancing healthcare, w with with less resources actually.
David T: I've Karna, I I've I've been doing a lot of work in this area. I've been going around to rural hospitals and and and trying to understand these challenges. So I want to talk about this for a little bit longer because I've
Karan Arora: Yeah, yeah.
David T: I've I've learned I've I've learned some things and then I have some questions for you about how we how we solve them. So specific to pathology. Pathologists have pulled out a lot of have basically pulled out of on-site services from rural pathology places because the volume doesn't the volume doesn't support the business and a lot of pathologists don't want to live in rural areas, unfortunately.
Karan Arora: Yep. Yeah.
David T: Okay. So that's I so I I understand why pathology companies have done that. On the other side, we have we have digital technology that's available. But these rural clinics, hospitals, while they can perform the procedures, it's difficult for them to afford new technologies, specifically in pathology, when they were just told, well, we're not serving you anymore. You're just gonna have to send everything to us, and that's that's all we're gonna do for you. So how do we how do we bridge that gap? How do we deliver technologies that seem expensive to these relative to these rural places so that they can get more pathology services there?
Karan Arora: Yeah, look, before I answer that question, you know, I I I think you have to remember like, you know, we talk about precision medicine. If but if precision medicine is only left to the richest zip codes in the US, it's not really precision medicine, right? You're not you're not uniformly helping people. So I think with with that in mind, like if you work backwards to the question you know you're asking, I I think
David T: can I interrupt for a second too? It's not it's not just it should serve the richest zip codes. Or sorry, it should serve you know, not just the richest zip, it should serve everybody, right?
Karan Arora: Yeah.
David T: it's it needs to be developed outside of the richest zip codes. The only way you can
Alae: Yeah.
Karan Arora: Hundred percent.
David T: get personalized medicine is if you get a diversity in your data set, right? So
Karan Arora: You got it. You you got it. You got it. And and
Alae: Yeah.
Karan Arora: and so it's, and and that's exactly what I was alluding to. it starts all the way from the beginning, right? It's like all the way from clinical trial design and the diversity of patient population there to ensure that you actually had represented populations and in clinical settings. And you know, I think your question around you know what needs to change. Well, there are a few things, you know. I think look, one is diagnostic companies, even
like ourselves, we have to realize that the business models in these rural settings are different, right? Like you can't go to a rural hospital and expect them to pay large capital expenses to procure the investment, right? It's almost like, you know, in the software business, you have a SaaS business, right? Like you have to think about moving to platform as a service or software as a service kind of models for digitization. Do you say, hey, listen, we will make the investment for you, put the capital in there, and then you pay as you start to utilize, right? It's a tokenized system. yeah, that could work even in our industry. we haven't done it, but it can it can work in our industry. I think there's a huge revenue opportunity for academic hospitals, that are around those rural clinics, to say, hey, instead of doing send outs, why don't I take high school graduates, where there's a job shortage in those communities, just train them as technicians. That all they need to do is basically be able to take tissue, put it on a slide and scan it. just that job. Just teach them in that one job really, really well. So that that's, that image can go to someone like Ali who has a digital setup anywhere in the world and can read the slide and with their experience make a good diagnosis and it goes back to the rule setting. It saves costs and it actually unlocks additional revenue opportunity for these academic clinics. So there's an opportunity there. I think third is government, all of these rural governments. Are trying to deliver better healthcare to their rural communities. And I think there's an appeal to those governments to say, hey, listen, this is the fastest way to deliver better care, especially in a domain like oncology, which is like the fastest growing expense, you know, after like cardiovascular and diabetes, you know, like in terms of healthcare costs in the US ecosystem and around the world, frankly. Like most of the major economies have oncology rated one or two. in terms of the, most focused places that they have to try and figure out how to democratize as well as reduce cost. So I think those are probably the three ways that, we can take crack at rural.
David T: So we gotta start at the we gotta s we gotta think about this from the the beginning. So the question is I'm an in I'm an innovator. I have a lot of friends and colleagues that are developing biomarkers that can turn into companion diagnostics. So what what's your message to them? They're developing these things that could have really good specificity. At what point in their development should they partner with someone like you or other companies that are developing therapies to to to to work together to advance their companion diagnostic together. When do they start?
Karan Arora: Yeah, I use this with some of our pharma partners and I and I always tell them, look, a drug cannot help a patient you cannot find. it's and it and that that's your that's your message, right? It's like, guys, it's great you're building fantastic algorithms or you're building phenomenal therapies. If those things don't tie together and you're not deploying these diagnostics at scale, and you cannot find those patients, like that therapy is not relevant. You're not helping anybody.
Alae: Yeah. I agree with that. I
David T: And the the the other side of it too is you can you can develop a a wonderful algorithm or a really specific biomarker, but if nothing ever g if it doesn't lead to an effective treatment or improving some kind of patient care,
Karan Arora: Yeah.
David T: then that that cool thing that you built also doesn't mean anything.
Karan Arora: Hundred percent. And that that's really what it is.
Alae: Yeah, I feel like from my perspective, this is like if I make a diagnosis of a certain type of cancer, right? And it's a rare type of cancer, and it is not very well known that it's associated with certain types of mutations. Like there's some big ones that we know, right? But there's ones that are, I guess, hide in hiding or they're not very emphasized. And you as a pathologist can only retain so much information, right?
Karan Arora: Yes.
Alae: you get that one case every ten years. And it's very hard to remember what, okay, that mutation. yeah, there's a treatment for that thing associated with that mutation.
Karan Arora: Yeah.
Alae: nearly impossible. So yeah, we definitely need stuff like that where it A recognizes it, B, obviously you need the marker, but it's like, well, how would I know to test for it in the first place, right?
Karan Arora: Exactly.
Alae: because it's one of these rare entities. so how would I know how to test for it? And okay. So if I test for it, what does that mean? What is what is the implication for the patient in terms of treatment? And then, well, what is the treatment, right? So
Karan Arora: Yeah.
Alae: yeah, you as a pathologist, I think practicing the modern day, I feel like we can do our work, but we definitely need to enhance what we already do with tools that we may not be aware of. And and this is, yeah.
Karan Arora: Yeah, and I think that's such a great way to explain it. You know, I look at pathologists as like ultra athletes, and I and I this is why I make this comparison, right? Like think about it this way, right? It's like you are doing something consistently, repeatedly, at scale, that and requires a lot of concentration, a lot of focus, endurance, and has meaningful impact to the resulting patient. And and now you're doing that without data. Now, for a second, compare yourself like this analogy of an ultra athlete, right? Now an ultra athlete today like has a crap load of wearables that they put on, right? And each insight from that data makes them better, whether it's like understanding their sugar levels or understanding their sodium levels or understanding, you know, like temperature and all those things collectively. gets them 30 seconds more, a minute more, and able to drive longer and
go faster. And you know, I think that that is how we need to view data, right? It is not, it is not a replacement. And you know, I think a lot of people, I speak to pathologists are like, my God, if I bring AI in, like what'll be my role? I'm like, guys, you're looking at the wrong way. It's like there there is in no scenario today that you yourself With the onslaught of different therapies, with different data coming in to your point, the case variation load that happens, the obscurity and the one in a million, you can keep all that information in your head. No human being can. So this is basically like your aid that's collecting all this information and basically piling it on your experience to make you more effective and efficient in that run, right? Which is why I equate you to an ultra athlete, right? It's like you need the data.
Alae: I appreciate being equated to an ultra athlete. I don't feel f like I could physically like that, but
Karan Arora: You will, you will. It's that's why we're talking about it.
Alae: I don't maybe well here's the thing. I feel like maybe in the medicine world we can be like the ultra athletes of the medicine world because we do so much and it's so
Karan Arora: Yes.
Alae: you really do need I mean, in a certain sense in the pathology lab, everything is highly regulated, right? And everything
Karan Arora: Yes.
Alae: like where your slide is, you can check the logistics are on point. we have internal lot of quality control things to know where your slides, when they're gonna get delivered, where they are, are they in process, are they in transport? we do regulate that stuff. So maybe in a certain sense, yes, we have our own version of wearables.
Karan Arora: Yeah.
Alae: but I think, you know, one thing that I noticed when it came to pathologists who were not aware of like companion diagnostics is one thing, but like when it comes to digital and like their lack of awareness of why digital is important in the AI world, like if you have a hundred cases, right? And a hundred c let's it be reasonable. Thirty cases in one day. Okay. And AI scans them and says, like, well, I picked out these four and these are probably malignant. They should get your attention first. I think a lot of pot pathologists don't have the they don't know that the AI that AI can actually do simple simple, you know, quote unquote simple tasks or like triaging. They don't like there's a lack of awareness in our field about what digital and hence AI can do for us. And I always talk about decision-making capacity. You only have A, so much knowledge, B so much making, so much decision-making capacity before your decisions aren't as sharp.
Karan Arora: Yeah.
Alae: so there's just I feel like a lot of people that just don't pay attention to other things that AI can enhance us with.
Karan Arora: Yeah.
Alae: Not just diagnostically but for pati in and in terms of patients, but really in terms of our workflow as well.
Karan Arora: Yeah, and you I I think that's a that's an excellent point. Look, CDX and compan and diagnostic driven therapies are less than fifteen percent of your entire workflow, right? In most cases. So you're you're talking about eighty-five percent of stuff being outside of CDX. So I I think that's such an important point, is like workflow optimization efficiency is where digital plays an excellent role as well, right? It's like just getting routine stuff out of the way. So you're actually spending time on things that require your attention and experience versus right now, you don't know you know till you actually examine a case if this requires the same level of attention as the next case. Whereas we know with AI and some computation involved, you'll be like, no, no, no, these to your point, these, you know, 26 cases I can let go, I can just do a review and say, yeah, it makes sense and check it out. Whereas these four cases I really need to spend time and examine further. So yeah, well said.
Alae: Yeah, yeah, exactly. It definitely helps when you want to spend more time on a certain type of case or a certain case. but you have a bunch of other cases in your queue that need to get signed out. there there's kind of like a I wish I had more time in the day and then by the end of the day you're just like, Okay, maybe I'll just sign this out tomorrow because I want to sleep on this. That was advice I was given as a resident. Is like if you're not a hundred percent there, don't sign out a cancer case. Make sure you review it with a colleague, you know. Very, very good advice.
Karan Arora: Yeah. Yeah, 'cause you get fatigue, right? that's known science as productive brain power is two hours for any given human being. I and some human beings have more, just like pathologists, they maybe have two and a half. So I think, I I just
Alae: Yeah.
Karan Arora: think, what you're trying to do is remove the variables, right? Like around you and and and give you that more capacity in a consistent way.
Alae: Yeah, I th I like that. More capacity in a consistent way, for sure.
David T: All right. We'll we'll do a couple questions to close us out here. So what Karan, what's what what six what's gonna signal success for you that we've completed writing chapter two? We've expl fully explored all of these new capabilities in computing.
Karan Arora: I think look the biggest change in conversation that can happen for me is people start with the patient first and come to the table, not asking like, my God, like all these barriers to adoption. Like they basically come to the table saying, How do I actually make it happen? Like it's inevitable. I need this to do better care for my patients. And I would love to engage in that question. Because you know, if somebody's motivated to solve the patient problem, we can do Anything. Anything. The rest of the stuff
David T: So
Karan Arora: becomes a discussion on how we make it happen.
David T: No, that's interesting. So what you're saying is right now you're developing these things and you're trying to sell them, but you're not people aren't coming to you like demanding, you're just develop is that am I understanding that correctly? You're pushing for it and and you feel like people should be asking people should
Karan Arora: Yeah, it it's a push, and even in the push, it's a push and yeah.
David T: okay.
Karan Arora: I I think people should be like going, my god, I I need this, I need this for my for my patients to deliver better care, to gain more capacity, to remove fatigue, to attract talent in a field that frankly not many people want to enter. it's like, so I think I we don't have the pull yet. And I think that would be the Turning point, for me in chapter two is where when you start to see the pull and people going, Yes, this is like a requirement.
David T: And after after this conversation, what would you what would you write the title for chapter three to be?
Karan Arora: Yeah. Yeah. Yeah. You know, if you put me on the spot, I think if I was to put a
David T: Yep.
Karan Arora: headline, I would say chapter three would look like you know, basically a patient saying how healthcare has changed and helped them live a better life.
David T: that's a wonderful way to to end this conversation. Karin, thank you so much for joining us for being so insightful. Everything that you've contributed to this community. We're we're really grateful to have people like you bringing bringing these really cool things that that have a lot of potential for high impact to to pathology, to oncology, and to to patients, as you said.
Karan Arora: Yeah, and and thank you for having me. And I really enjoyed the conversation with you and Ali. I think it was it was hey, I'm very enthusiastic about it. Hopefully you can tell. But also
David T: Yeah.
Karan Arora: I I think it's the right time. We should be having more of these conversations in the community and and people should be wanting to learn and wanting to change things, right? the status quo is not not okay.