The Technology that Makes Cancer Less Scary with Abhimanyu Verma, CTO at SOPHiA GENETICS

(Intro Narrator at 00:00:00) Today, we're talking to Abi Verma, CTO at SOPHiA GENETICS, about how they're making cancer less scary. You're listening to Joel Beasley, Modern CTO.

(Joel Beasley at 00:00:16) I guess the way that you described it was we're making cancer less scary. Are you guys solving cancer over there? What's SOPHiA GENETICS? What are you guys doing?

(Abi Verma at 00:00:25) So at SOPHiA GENETICS, the vision really is to make access to healthcare equitable across the world. The mission there is really to expand the access via data-driven medicine by using AI, with a particular focus on cancer and rare diseases. So what that translates to really is we have a decentralized, cloud-enabled platform.

(Abi Verma at 00:00:52) We call it SOPHiA DDM. It's used by over 780 plus institutions across the world. And what that platform does is it analyzes complex genomic and multimodal data, which is like clinical and radiomics and so on, in real time and gives decision makers, be they oncologists or pathologists, actionable insights to help with their decision making so that they can apply what we call precision medicine in the care of their patients with cancer and rare disorders. Now with this platform and obviously with this global network of users, you get data. The platform's been around for a few years, and over 1.6 plus million such analyses have already been done on the platform.

(Abi Verma at 00:01:45) So obviously, as that data gets computed and the users provide more and more input, you start to create this collective intelligence in the platform, which then has a feedback loop into the insights that the users can get from this.

(Joel Beasley at 00:02:03) So it's like an AI assistant for doctors?

(Abi Verma at 00:02:06) I think that would be a very gross simplification if you call it an AI assistant for doctors. It is really around an AI-assisted analysis to provide, to help them with their decision making, especially in the context of identifying, for helping with diagnosis so that they can really prescribe the course of treatment that would help the patients.

(Joel Beasley at 00:02:38) Why are you reluctant to say it's an AI assistant for doctors when that's exactly what you're saying?

(Abi Verma at 00:02:43) That can have many connotations. You can have an AI assistant for prescription writing, for example.

(Joel Beasley at 00:02:51) So it's specifically for diagnosis, like diagnostics? Okay.

(Abi Verma at 00:02:55) For supporting decision making around diagnosis. The platform itself will not give a diagnosis. That's really a decision that the practitioners have to take, whether they're oncologists or pathologists. And of course, they view it in the context of the overall information they have on the patient.

(Joel Beasley at 00:03:15) So it doesn't attempt to diagnose. It attempts to help the doctor...

(Abi Verma at 00:03:20) Exactly.

(Joel Beasley at 00:03:20) Make decisions...

(Abi Verma at 00:03:22) Exactly.

(Joel Beasley at 00:03:22) As they attempt to diagnose. What is that? How—so my brother and mom are physicians, so it's close to home for me. So explain to me how—explain that to me. So it's gonna help me make decisions on what the diagnosis should be, but it's not going to suggest the diagnosis.

(Abi Verma at 00:03:42) Exactly. So let's look—I'll give you a real life example as to how this plays out. Imagine this. A patient walks into a hospital or a diagnosis lab, and they have cancer or a rare disease. And what is the first thing that doctor would do?

(Abi Verma at 00:04:03) They would run a diagnosis, and that can span across genomics, clinical, or radiomics. So now we are generating a whole slew of data across all each of these modalities. Now the ability to collect all of this, connect all of this data across these different modalities, and provide insights to help with that decision making, specifically in the case of looking at genomic tests that have been run. So what are the variants, the mutations that may be causing a certain disease, which can then help inform the pathologist or the oncologist to then understand what is the course of treatment ahead? And so that is really what it's about.

(Abi Verma at 00:04:45) Now, of course, as this information comes together across a period of time, that information can then be used in the context to support drug discovery and research by biopharma. So that's the other part of what the platform can be used for.

(Joel Beasley at 00:05:03) All right. How does it help them to make decisions without suggesting a diagnosis?

(Abi Verma at 00:05:10) Again, let me give you, going back to that example I gave. Patient comes in, they run a genomic test. The genomic analysis, the data that our customer would give us, the user of the platform would upload it to the platform. The algorithms run on that data to be able to then give information back as to, okay, how do you interpret the data that is given to be able to understand, okay, this patient has these four mutations.

(Abi Verma at 00:05:42) That's it. Or if you look at this patient and if you combine the genomic data with the radiomic images, with various...

(Joel Beasley at 00:05:56) So say in pop-up notifications. Be like, hey, look, they've got this, and then this is also connected to the same thing that's connected.

(Abi Verma at 00:06:03) Yes, in a way you can understand. It's not a pop-up notification. There's a lot of very complex...

(Joel Beasley at 00:06:08) Say it's pop-up notifications like 1998 web browser, just...

(Abi Verma at 00:06:14) Yeah. So...

(Joel Beasley at 00:06:16) I get it. Okay. So that satisfies that part of my curiosity. You bring important facts forward from a larger set of data. Yes.

(Joel Beasley at 00:06:25) And you say, hey, these are areas of interest and things you might want to look at, and then you let them draw the conclusions.

(Abi Verma at 00:06:32) Right.

(Abi Verma at 00:06:33) And the beauty here is, exactly what you said. From vast amounts of data, information that they would have to otherwise either miss out or, there are multiple silos and that is not interconnected. So that's really what the platform enables.

(Joel Beasley at 00:06:51) Well, I get it. So when my brother became a physician over ten years ago, I went to visit him at his office—general practitioner, family doctor. And...

(Abi Verma at 00:07:03) Where does he practice, by the way?

(Joel Beasley at 00:07:05) Florida. Venice, Florida. Okay. All right. And then my stepmom is a bariatrics individual in Tampa, Florida.

(Joel Beasley at 00:07:15) Okay. But so the first time I went to visit him, I went to dinner with him after work or something. He had a stack of patient folders, and I said, what's—you bringing home work with you? He goes, yeah. He goes, this is my new patient load.

(Joel Beasley at 00:07:28) And I go, what do you do with that? And he goes, well, I just kind of sift through, you know, the patient to try to get an understanding of them, and they've got reports from doctor's visits in different parts of the country and different—all these, just this bunch of stuff that I have to sift through. And then I was like, what do you do? He's like, well, I just look for important things in relation to their chief complaint or whatever their primary situation is. And I said, man, I said, in the future, you're just gonna be able to scan that whole thing in and then have an AI tell you the most important things you need to know from your patient's history.

(Abi Verma at 00:08:04) Towards what you pictured. Now it's a complex field, and you have to really get precise. It's not like a shopping experience where you get, okay, you purchased 20 different things before, and then so... It has to be precise. And of course, biology is complex, not very well understood.

(Abi Verma at 00:08:29) So you have to rely on the current evidence base that is there, but then also give the opportunity for the doctors or the pathologists to share back their experiences, which then feeds back into this what we call the collective intelligence platform, which then helps a doctor maybe sitting in another part of the world, let's say Mumbai or in Nigeria. And they see a patient with similar profile or characteristics walk through the door, they learn from the experience that the doctor in Florida may have encountered, which is what we call about—the tagline that we use and the statement we have is called democratizing data-driven medicine.

(Abi Verma at 00:09:14) So just to break that down. You're making decisions based on data. So that's data-driven medicine. And you're democratizing it because you're now spreading the access and the knowledge worldwide. So to give you a few examples of some of the institutions that we work with. You're familiar with the U.S.

(Abi Verma at 00:09:40) So, for example, in the U.S., we work with organizations like Tennessee Oncology or Memorial Sloan Kettering Cancer Center.

(Joel Beasley at 00:09:56) I'm in Tennessee.

(Abi Verma at 00:09:57) Yeah. Okay. So there you go. Yeah.

(Abi Verma at 00:10:01) And or BioReference Labs, which is one of the largest lab networks in the U.S. And then there are similar organizations in Switzerland, France, in India, in Nigeria, that we just are onboarding onto our platform called SyndicateBio, or in Canada, a network of hospitals whom we are partnering with called Exactis. So you can see that as all of them start to work on the platform, they're all learning from each other and each other's experiences and able to add to that insights network for that collective intelligence, but then also provide input into it.

(Joel Beasley at 00:10:45) So I have a question for you, kind of a nerdy question.

(Abi Verma at 00:10:48) No, no.

(Joel Beasley at 00:10:48) I've been getting into the decentralized AI, and all of this is very unscientific and just my feelings. But I've been noticing that AI tends to be pretty good narrowly, similar to how humans are. Like, we're really good at one or two things.

(Joel Beasley at 00:11:07) It makes sense because we design large language models to mimic human biology, and as far as how they learn. So I said, okay, well, so now you've got, let's just say, all these little apps. All these little individual models that are really good at one thing, and we'll call that intelligence models. And then we've got another layer that's emerged as inference. So basically, think about it like a really high-powered executive assistant that knows how to get things done.

(Joel Beasley at 00:11:36) So you can tell them what to do and then they know how to achieve this outcome by leveraging multiple different intelligence people in gathering and organizing and structuring, and then come back to you with the result. So I've been—I don't know if this is how the educated people talk about it in AI, but I've been calling it inference and intelligence. And inference is essentially like a distribution, retrieval concept, delegator between the intelligence and the end user. Now, does any of that make sense? Is any of that sound true from what you're experiencing?

(Joel Beasley at 00:12:09) Because you have different modalities of data. You've got genomic test mutation intelligence. You've got radiomics intelligence. Am I on the right path here, or am I in left field?

(Abi Verma at 00:12:20) Without stepping into the actual technical approaches that are there. Because there are large language models and they're great in languages. As yet, it's very, very early days on how they apply in the context of biological data and in the field that we operate in. Still some ways to get there. And I mean, there's some initial work that has started in that context, but still some ways to get there. But if I were to take the analogy that you mentioned, so yes, definitely, the foundation has to be the data, and how good the data is and how well it is interconnected. And that's challenge number one.

(Abi Verma at 00:13:01) Because data is very, very—unlike language, English is English. Yes, it's written differently and spoken differently. But overall, the base structure is the same. But if you look at data structures across the world, especially in this context, there is a large degree of variation. So one is how do you overcome this challenge of variations in data, and even how the data is generated?

(Abi Verma at 00:13:28) So even in the—take the example of genomic testing. The output file that comes out of genomic sequences or radiomics that comes out of the scanners will vary depending on the chemistry used or the dye used or the instrument used, and then that can have a huge impact or influence on the results that you have. The differentiation and what, here at SOPHiA, what we've been able to accomplish is to sort of create that normalization across all of these different sources of input and the noise that may be there to be able to have very precise analytical output.

(Abi Verma at 00:14:11) And, you know, to put in very layman terms, think of it this way. You have handwriting recognition software, and different people have very, very different styles of writing.

(Joel Beasley at 00:14:25) Doctors are the worst, I'll say it. Yeah.

(Abi Verma at 00:14:29) And over a period of time, you've had now, you know, just with that, these models and AI and approaches have been developed to be able to overcome all of these different styles of writing and still have a good enough output and a fairly accurate output. I'll take that same example and apply it in the context of what I was talking about in the context of genomics and radiomics, and then to be able to bring that together. So that's what we do.

(Abi Verma at 00:14:55) So over here, to go back to how you were describing it, so the foundation being the data. And then you have the layer on top, which is how do you apply intelligence on top of it using AI, machine learning, stats, different techniques to be able to normalize across all of these different variations to be able to get that inference or the very precise insight, which is able to work across all of these different variations that are there.

(Joel Beasley at 00:15:32) So how are you guys making cancer not so scary?

(Abi Verma at 00:15:37) Okay. So now to take a big step back. So cancer—if you start thinking about cancer being very scary, why is it scary? Because you don't know what your outcome will be. Because, you know, in most cases, it's, the mortality rates are fairly scary in that context. However, you have treatments coming through now where one's quality of life and one's survival, the chances of surviving increase.

(Abi Verma at 00:16:15) And why is that possible? Because these therapies function best if they are applied in a very precise way in the context of that one individual, which means understanding what is in that patient's biological makeup that is the cause of the cancer. Is it a genetic mutation? Is it a combination of other factors? How's the cancer progressing? Is that therapy really working or not working? What combination works or not? And to be able to support those decisions, you need an insights platform like SOPHiA DDM to help you understand and make sense of all of this information.

(Abi Verma at 00:16:58) So that's which then, you know, as an oncologist, you're able to make those decisions to help your patient.

(Joel Beasley at 00:17:05) Things are scary that we don't understand.

(Abi Verma at 00:17:07) Exactly. And then what I described, absolutely. But now take this concept and say, okay, how do you take this across the world?

(Abi Verma at 00:17:16) If you're lucky to be in the West or in developed countries where the cancer care or patient care is very evolved, a lot of knowledge, but then, you know, there are cancer patients the world over, right? And they're very smart doctors the world over. How do you create access to the latest technologies and these latest approaches across the world?

(Abi Verma at 00:17:41) And I'll give you an example of where we are doing that. So at the Memorial Sloan Kettering Cancer Center in New York, over the last few years, they developed a test around liquid biopsy, right, which was New York state approved, and they use it regularly for all the patients or the people walking into MSKCC for care. And in our partnership with them, what we've done is we've taken that test or that diagnostic test that is there, that assay, and being able to engineer it in a way with the same precision that it runs on our platform. So now you have folks in India and Nigeria and in Europe having access to the same level of precise insights.

(Joel Beasley at 00:18:36) How do you do that while still making money?

(Abi Verma at 00:18:40) How do you do—

(Joel Beasley at 00:18:41) Like, how do you take this technology and this intelligence and make it available and still make money?

(Abi Verma at 00:18:51) That goes to the business model, right? So ours is a consumption-based platform. So for every analysis that is run on the platform, the users, obviously, you know, that's what they pay for in terms of pricing. So it's as simple as that.

(Abi Verma at 00:19:20) So the more we expand across, the more we expand usage across different institutions across the world, the more the number of analyses on the platform, and that's how one makes money. And that's the most enabled model from a business perspective.

(Joel Beasley at 00:19:37) Nice.

(Abi Verma at 00:19:37) And you—

(Joel Beasley at 00:19:37) You guys are a for-profit company. It's not like a—

(Abi Verma at 00:19:39) Like a—

(Joel Beasley at 00:19:39) Philanthropy. Okay.

(Abi Verma at 00:19:41) No, they're a company. That's right.

(Joel Beasley at 00:19:43) Oh, nice. Nice. Yeah. Again, all of my questions come from a good place of love. I'm just a curious person, which, you know, does well for when your job is to ask a bunch of questions.

(Joel Beasley at 00:19:55) Yeah. So do you have any good stories of a person who actually went to a doctor that was using your technology and had a better life because of it?

(Abi Verma at 00:20:11) I do, actually. So, you know, we very often, and especially also for our data scientists and engineers to really feel the impact of the work we are doing, right, we often—and of course, I also visit customers very often across the world—and you often hear these stories where, you know, they had a specific patient.

(Abi Verma at 00:20:33) In fact, recently we had the story where one of our users who is in Brazil, right, part of—is in Brazil—had a very difficult case. And there was a similar situation also in France, right, where they just couldn't understand, you know, what was causing the cancer, what were the mutations. And using our platform, they were able to get those precise insights that really helped them make that decision. And then they were able to put their patient onto the right course of treatment, which really helped the patient.

(Abi Verma at 00:21:16) And I mean, now there are so many stories like this.

(Joel Beasley at 00:21:21) Yeah. It's super interesting to me. For people who don't know, I used to think of cancer as like a thing, like a headache, like a very specific thing. And cancer is not a very specific thing. It's like a broad category that can express itself. And because I'm sitting over here as a normal person, not a physician, like, man, as a community and as a country, we spend so much money trying to solve this cancer problem. Why isn't it solved, you know? But then you realize that it's like a bunch of different things, and then you're like, oh, okay.

(Joel Beasley at 00:21:58) That's why it's not solved, because it's not just one thing. It's a bunch of things.

(Abi Verma at 00:22:01) And then even that, right, even, you know, per organ, there's multiple variations of that. And this is where the concept of precision medicine comes in, right? How do you create treatment regimens or courses which are tailored to an individual, right?

(Abi Verma at 00:22:18) That's the holy grail, right? And which is where a platform like SOPHiA DDM helps progress the journey towards that vision, right? And that's, again, you know, what I know we've been emphasizing a lot is on the diagnosis side, right? Or on the initial or supporting the diagnosis side, right?

(Abi Verma at 00:22:37) But then you go further upstream, right, is when cancer research is going on and therapies and medicines are being developed. There again, you have to make informed choices with regard to what are the kind of patients that should be put on a clinical trial, right? What works, what doesn't work, and, you know, who are responding better and who are not. And, you know, while there are established methods in the pharmaceutical industry in terms of running clinical trials to do this and to get to the necessary output and approvals around that, again, to inform those decisions, they need to use what is called real-world data.

(Abi Verma at 00:23:20) That is data that is coming from daily practitioner use, which helps them understand, okay, what's the best way to design a trial, so that their therapies are better targeted and are serving the right patient with the right dose. And again, you know, a platform like ours with all the data that has been collected and that intelligence is built in, is used by pharmaceutical companies to help them with things like, you know, helping them with trial design, selecting the right kind of patient profiles for the clinical trials and so on.

(Joel Beasley at 00:24:02) So my background was mostly in business logic, building software applications for real estate and finance. So I know what it's like to work with those types of people in business. I don't know what it's like to work with, you know, you're trying to share data with another cancer center or Tennessee Oncology or—so my question to you is this. What is the culture like amongst the technology professionals and business people that are trying to share data and build this better world? Like, if I want—if I wanna put together a group that's gonna share radiographs—

(Joel Beasley at 00:24:38) Is that something that's really hard to do, or is that something that people are pretty eager to figure out how to make happen?

(Abi Verma at 00:24:46) So from a motivation perspective, people are definitely eager to make that happen. And that's because, you know, they're all driven by the purpose, right, this higher purpose and this larger purpose to make cancer less scary or to democratize precision medicine. However, there are, you know, this is very sensitive data that you're dealing with, right? You just can't share it just like that. I mean—

(Joel Beasley at 00:25:12) Put it on GitHub, dude. Just write it on GitHub. No, I write it with my Social Security number.

(Abi Verma at 00:25:17) Right? Because there are concerns that one has to be mindful around patient privacy, around data security. And so a lot of these considerations have to be put into balance as you make decisions. That definitely influences the way—the operating rhythm, what the checks and balances you put in place, what you can do, what you can't do. So, yeah, for sure, right?

(Abi Verma at 00:25:45) And—

(Joel Beasley at 00:25:47) But people are pretty eager to figure it out. Like, they're working on the—I mean, at the doctors, I sign 8,000 forms to get my temperature checked where I'm giving my data to a billion people. So it seems to me that people—

(Abi Verma at 00:25:59) At the back end, I think that at the back end, that doesn't happen in that end, right? So even if the data was getting, you know, it doesn't get to a billion people, right?

(Abi Verma at 00:26:07) For sure, right? And even when it does, right, the data is so different from each place to each place that there's a whole bunch of work that is required to just stitch this all together, right, and to harmonize it and get it in the same way.

(Joel Beasley at 00:26:21) Well, it reminds me a lot of the PII or the PPI, the personal identification when you're dealing with production and development in an enterprise application, right? You have to strip away some things—

(Abi Verma at 00:26:35) Yeah.

(Joel Beasley at 00:26:36) Even when you're, you know, running your own test.

(Abi Verma at 00:26:38) Just take that—so take that concept and just amp it up by a hundred.

(Joel Beasley at 00:26:42) I know. I know, right? It'd be pretty cool, though, if it could take the—like, I wonder if there's a technology out there. Josh, maybe we could see if this exists or not. But if I give all my information to my doctor and then my doctor's involved in some data-sharing program, if it could somehow abstract my height and my weight and my medical history, but remove my name and social—

(Abi Verma at 00:27:07) There are approaches there on how to retain privacy of data and to strip out these sensitive fields. But the downside of that is the more you strip out, the less insightful the data becomes, because you're taking out a lot of the same information that helps inform better insight.

(Joel Beasley at 00:27:29) Yeah. You just take out my name in social. Like, you know, have age of human, biomass of human.

(Abi Verma at 00:27:38) No. That—height of, yeah, yeah. So for example, right, if you did that, right, you could still be identified based on your genetic information, based on—

(Abi Verma at 00:27:51) On your ZIP code and sort of bringing this together, right? So there are techniques out there which allow you to balance between utility from an analytics perspective versus staying within what is the expectation around privacy and data security.

(Joel Beasley at 00:28:10) It should be like Apple. You know, when you have the app—

(Abi Verma at 00:28:13) In itself. That's a whole field in itself, right?

(Joel Beasley at 00:28:15) Do you have any Apple devices?

(Abi Verma at 00:28:17) Yes, I do. Yes.

(Joel Beasley at 00:28:18) It should be like Apple where it's like, do you wanna share this with the app developers? So when I go get a test run, I'd be like, do I wanna share it with the medical community? And then I can choose, you know, if I wanna share it. No?

(Abi Verma at 00:28:30) No. That—no, no, no. You wouldn't do this. Put it in context like this, right? Would you as an individual, right, and you said this, right, you sign all these forms, right?

(Abi Verma at 00:28:44) And you seem pretty cool about, okay, let the data be with a million people, right? Well—

(Joel Beasley at 00:28:51) Not all the data, but like—

(Abi Verma at 00:28:53) There you go, right?

(Joel Beasley at 00:28:54) That's—okay.

(Abi Verma at 00:28:54) So what does "all" mean, right? And how do you break that down, right? In terms of what is allowed versus not what is allowed. So there are nuances out there which are required and need to be thought through. And yes, as a concept, yes, you should. And then technology exists, right, to how to be able to share the data. And there's been lots of—I mean, I personally have been involved in a lot of projects even in the past where, how do we enable this kind of data sharing?

(Abi Verma at 00:29:18) Is it in the right format? Is it encrypted? Is it in the right—are you not losing the utility, the analytical utility? And how do you approach these things? But you have to be very mindful of how you approach these in different contexts, staying within the—

(Joel Beasley at 00:29:35) Look, I'm not the guy for compliance. Like, you're talking to the wrong person here. If I'm in charge of compliance, things have gone horribly wrong. We need a better business model so we can hire compliance people, because I'm too entrepreneurial, "get it done at all cost and figure it out later" situation. So at least I know that about myself, right?

(Abi Verma at 00:29:55) So which kind of links it back to just saying that, how do you keep that entrepreneurial spirit, right, and make progress, while leveraging the laws and the regulations in a positive way, right? And you don't see them as barriers, right? You see them more as enabling because it provides structure.

(Abi Verma at 00:30:21) Right? It provides you guardrails as to what is allowed and what is not allowed, right? And in the case of SOPHiA, that's sort of built into the architecture from day one, right? So it's a SaaS platform, but, you know, we are operating across multiple countries, and we have to comply with the laws per country. And there are very clear frameworks like GDPR and so on, which we comply with, right? And our architecture enables that. Our decentralized architecture and our distributed architecture enables that.

(Abi Verma at 00:30:56) In, you know, what data stays where, in which region, how it is, what is allowed to be—and what insights can be centralized versus not. So, you know, that's a lot of work to figure that out, and it works very well. And that sort of reflects in the trust that, you know, so many of our users place in us with the growth we have. Now put it in context of what you see now is happening even in the context of AI, right? There's a big movement that are—what is called responsible AI, right? How do you make sure that, you know, that the hallucinations or—and that, you know, there's no—the outputs are not biased, and—right? And the answers are—

(Joel Beasley at 00:31:39) All outputs are biased. That's how it works.

(Abi Verma at 00:31:41) So, and to have the right kind of thought around what are the guardrails, what are the checks and balances you have to put in place, what kind of testing you need to do, what's the—is your data complete? All of this also regulations support, right, that kind of thinking. And it's really something that, well, we view it as an enabler, really.

(Joel Beasley at 00:32:09) Right? Yeah.

(Abi Verma at 00:32:10) And as a burden, right? That's something that, oh my God, I gotta do this. It's more like, okay, the framework is there right now. Let's use the framework to come up with the solutions and then also to inform decision makers as to what's the right construct here and to evolve that in terms of—

(Joel Beasley at 00:32:26) Well, please, when you're out there making decisions, keep truth at the root of it and not political stuff, because truth is what's going to give us the best outcomes, right? Because I've had a lot of conversation about AI and bias, and there's two conversations I've come across. One is bias in the sense that the AI is coming to the correct conclusion. We're just unhappy about it because we don't think that that's the right way for our future.

(Joel Beasley at 00:32:54) And so for that, I'm like, no. If it comes to the right conclusion, it comes to the right conclusion. If we're not happy about that, that's a separate conversation, but don't change our algorithms—

(Joel Beasley at 00:33:08) To not be happy about it. But, yeah, so especially in medicine.

(Abi Verma at 00:33:09) In the context of biology, right?

(Joel Beasley at 00:33:12) Right. Thank you.

(Abi Verma at 00:33:14) Yes. I mean, you have to be—going back to the examples you were giving earlier. Why can't we just have, you know, a simple app that or a pop-up reminder, right, and you sort of—uncomfortable sort of pushing back on that because it sort of links back to this, right?

(Abi Verma at 00:33:31) You cannot give, you have to be very precise about the output that you're generating or the insights that you're giving, right? And that has to be rooted in the context of not just the technical precision, but then it also has to be linked to known knowledge around biology and what reference information you're using, as well as what user input is there to handle, as well as the variations I described before in terms of how the data is generated, how you bring that all together and rigorously test it and validate that. That is extremely, extremely important, to get that right, which is what we pride ourselves on, right, as having a very intense focus on precision of our analytical outputs.

(Abi Verma at 00:34:26) Even if it comes at the cost of, okay, I mean, when you were linking it back to pricing, right, in different geographies, right?

(Abi Verma at 00:34:34) And there is a certain premium one has to pay for good quality, right? And sometimes we are unable to get users on board because, you know, they still haven't reached that maturity where the price value is understood.

(Joel Beasley at 00:34:55) Oh, I get it. I'm a business owner. I understand it 100%. Yeah. But it seems like your heart and your business model are all in the right place in the sense that you're trying to make this more accessible. You're trying to give better tools to people who would otherwise not have them, which will ultimately create better outcomes for patients. And I'm on board with that. I like that.

(Abi Verma at 00:35:13) Yeah. And then, you know, that's what gets us out of bed, gets me out of bed every morning and highly motivated to make progress.

(Joel Beasley at 00:35:23) So just as we wrap up, when will you guys determine the cure for cancer?

(Abi Verma at 00:35:32) Like I said, we're not in the business of curing cancer. We're in the—

(Joel Beasley at 00:35:35) I know. I know.

(Abi Verma at 00:35:36) I know.

(Joel Beasley at 00:35:38) I know. I latched on to that. It was in my prep about making cancer less scary, and my first thought was, are you guys curing cancer? Yeah. But about making cancer less scary—

(Abi Verma at 00:35:44) Yeah. But about making cancer less scary? Oh, absolutely, right? Yeah.

(Abi Verma at 00:35:48) And I explained to you how, right?

(Joel Beasley at 00:35:50) And I agree with it. I don't think it was a bad explanation at all. I mean, people are scared of what they don't understand, like, just by default. And the fact that you're making it more understandable means you're making it less scary.

(Abi Verma at 00:36:02) Right? And then, you know, you understand the causes of it, and then if you know what's causing it, then you know what to do about it.

(Joel Beasley at 00:36:12) Yeah. 100%.

(Abi Verma at 00:36:13) And if you're able to spread that kind of know-how and knowledge all over the world? Yeah. That's fantastic.

(Joel Beasley at 00:36:20) I'm a fan of it, by the way. I'm a huge fan of it. So I'm really glad you guys are doing what you're doing. One of the reasons why I was interested in this when my team pitched it to me is because my biological mom passed away from leukemia. And so I got to see her go through the identification and the treatment process and then, you know, the full cycle of it. And so I got to experience what that process was like. And so anybody who's out there who's making progress in our medical advancements, I'm always really keen to have them on the show and to support their work the best that we can so that better solutions can come faster for humanity at large.

(Abi Verma at 00:37:01) Yeah. Absolutely. No, thanks. And, you know, like, every one of us, most of us have this personal story, right, where we know someone who's been impacted or touched by something as scary as cancer or even on the rare diseases side, right? And it's a similar story over there. So, yes, absolutely. That's what gets us highly, highly motivated.

(Joel Beasley at 00:37:25) Well, Abi, we did it. We made a podcast. How do you feel?

(Abi Verma at 00:37:29) That's great.

(Joel Beasley at 00:37:30) Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics that you would like to hear discussed on the podcast, either add me on LinkedIn or send me an email, [email protected]. Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.