Follow The Brand Podcast with Host Grant McGaugh
Are you ready to take your personal brand and business development to the next level? Then you won't want to miss the exciting new podcast dedicated to helping you tell your story in the most compelling way possible. Join me as I guide you through the process of building a magnetic personal brand, creating valuable relationships, and mastering the art of networking. With my expert tips and practical strategies, you'll be well on your way to 5-star success in both your professional and personal life. Don't wait - start building your 5-STAR BRAND TODAY!
Follow The Brand Podcast with Host Grant McGaugh
The Power of Explainable AI in Cancer Diagnosis with Dr. Akash Parvatikar
Ever wondered how artificial intelligence could transform cancer diagnosis? Join us on the Follow the Brand Podcast, where we sit down with Dr. Akash Parvatikar, an AI scientist at Histowid. Dr. Parvatikar shares his unique journey from electrical engineering to pioneering explainable AI for early breast cancer detection. We promise you'll gain a deep understanding of how AI can classify medical images and why making these processes transparent is crucial for improving diagnostic accuracy and reducing misdiagnosis rates. This episode is a treasure trove of insights into the future of healthcare and the revolutionary role of advanced technology.
In a series of enlightening discussions, Dr. Parvatikar breaks down the integration of AI and digital pathology in personalized medicine. Discover how deep learning and graph-based approaches are identifying subtle clues in medical images, bridging the gap between misdiagnosis and correct diagnosis. We also simplify these complex AI concepts for a young audience, likening AI learning to everyday experiences like recognizing kitchens from photos. Listen in to learn how digitizing tissue biopsy slides is revolutionizing pathological diagnoses, enhancing both the quality and reliability of cancer detection. This episode is a must-listen for tech enthusiasts and healthcare professionals alike, looking to understand the transformative power of AI in medicine.
Thanks for tuning in to this episode of Follow The Brand! We hope you enjoyed learning about the latest marketing trends and strategies in Personal Branding, Business and Career Development, Financial Empowerment, Technology Innovation, and Executive Presence. To keep up with the latest insights and updates from us, be sure to follow us at 5starbdm.com. See you next time on Follow The Brand!
Welcome to another episode of Follow the Brand.
Speaker 1:I am your host, grant McGaughan, ceo of 5 Star BDM, a 5 Star personal branding and business development company.
Speaker 1:I want to take you on a journey that takes another deep dive into the world of personal branding and business development using compelling personal story, business conversations and tips. Development using compelling personal story, business conversations and tips to improve your personal brand. By listening to the Follow the Brand podcast series, you will be able to differentiate yourself from the competition and allow you to build trust with prospective clients and employers. You never get a second chance to make a first impression. Make it one that will set you apart, build trust and reflect who you are. Developing your five-star personal brand is a great way to demonstrate your skills and knowledge. If you have any questions from me or my guests, please email me. At grantmcgaw, spelled M-C-G-A-U-G-H at 5starbdm B for brand, d for development, m for masterscom. Now let's begin with our next five-star episode on Follow the Brand. Welcome to the Follow the Brand Podcast. I am your host, grant McGaugh, ceo of Five Star BDM, where we help you build a five-star brand that people will follow.
Speaker 1:Today we are about to embark on a journey that could revolutionize how we fight one of humanity's greatest enemies cancer. Imagine a world where misdiagnosis becomes a relic of the past, where artificial intelligence works hand-in-hand with our dudes to save lives. Sound like science fiction? Well, our guest today is turning this vision into reality. Meet Dr Akash Parvati, the brilliant mind at the intersection of AI and oncology, with a PhD in computational biology and a passion for pushing boundaries. Akash is just not a scientist. He is a modern-day explorer guarding the uncharted territories of healthcare technology. At Histowid, a cutting-edge biotech company, akash is pioneering the use of explainable AI and cancer diagnosis. But here's the kicker this work could slash misdiagnosis rates that currently affect nearly half of all breast cancer patients. Let that sink in for a minute. But Akash isn't just about cool bird data. This journey from electrical engineering to becoming an AI wizard in healthcare is as fascinating as the technology he developed, and, trust me, you will want to hear how a guy who started with robotics ended up revolutionizing cancer detection. Now we're about to dive deep into a world where algorithms meet biology, where computers are learning to see what even trained human eyes might miss, and where the future of healthcare is being written in the lines of code.
Speaker 1:Please join me in welcoming our Cubs for a B-Cup to the Follow Brand Podcast, where we are building a five-star brand that you can follow. Welcome everybody to the Follow Brand Podcast. This is your host, grant McGaugh, and I have been doing a lot of work with my counterparts that originate from India. I don't know what it is with India, but I have an affinity with people from that area and I love the way that they just embrace technology and they're really embracing these new emerging technologies and AI, and they have just owned this space. And they have just owned this space and I am truly enamored when I get an opportunity to talk to someone that truly understands, from a layman's standpoint like myself, all the way to the very, very deep ends really start solving some of these complex problems that we have, and that's what we're going to do today with my friend, akash. Please, please introduce yourself.
Speaker 2:Yes, first of all, thank you so much, grant, for having me here.
Speaker 2:It's amazing that, you know, you gave a platform like this for us to share our story and, like you know, not just build an awareness but also be, like you know, a voice of motivation for those people who are looking to step into, like you know, the technology world, and I would like to thank you for that. So, to the viewers, I'm Akash Parvatikar, originally from Bangalore, india, and right now I'm working as an AI scientist in a biotechnology company called Histobis, and prior to this, I got my PhD in computational biology, specializing in developing AI explainable AI, to be more precise, in the field of early breast cancer diagnosis and I'm super excited to be in this field. I'm personally motivated to drive, uh, like drive a change, uh, for a good cause and, as we talk today, uh, we'll be happy to, like you know, uh, go through like, what is it that I love in the field and what is it that I'm looking to contribute? Uh and uh, you know, yeah, happy to be on this show, trent, thank you so much.
Speaker 1:We've got to unpack a lot of that. That is amazing what you just talked about. First, you threw out a big buzzword I don't think a lot of my audience have heard before. You've heard the word AI, right, but AI is a big term. Explainable AI is something different. I mean you've got to tell us some, give us some examples different. I mean you've got to tell us some, give us some examples of how do you explain AI from a diagnostic accuracy perspective, especially something like breast cancer. I mean, what are you doing around this world of explainable AI?
Speaker 2:Right. So that's a very interesting question, grant. So when I talk about AI for breast cancer diagnosis, so it's mostly classification or making a decision, and, to keep it simple, we can put the breast cancer into two categories, say it is cancer or no cancer, right? So now you develop, like these artificial intelligence tools to basically do that. You give it an image and you ask it to tell okay, does this image show you cancer or it doesn't show you cancer? So at a very high level, there has been a lot of advancements in this area.
Speaker 2:Now, to add another layer of complexity is the explainability part of it. Now, many a times the AI that you have developed it might not resonate with what an expert, such as a pathologist, would see when they diagnose the image, whether the cancer or no cancer, because this AI would unravel in a very complex space. Look at like several different features. It might be like black box, where you don't know what's happening inside the black box, but it comes out like you know, it gives you the right answer, but you don't know as an expert how did it come to that answer. So for clinical application, to put this AI into the clinical realm, you need to build the tools which are explainable from a pathologist perspective, or to make it more trustworthy and, of course, to have regulatory approvals, such as FDA, for that, because you would want to know as to why the AI came to this decision and does that resonate with the decades of medical literature for cancer, how the doctors were diagnosing from all these years?
Speaker 2:Does AI follow that path or does it do something of its own?
Speaker 2:Because if it is something of its own, then you have to completely identify what are the failure modes, because you need to know when this AI will not work, and that's very crucial. And that can happen only if you invoke explainability into your system. And as we talk more of the project that I did and more than happy to give more depth into this, but at a very high level of when people hear about XAI, which is short for explainable AI, what that means is there is an AI tool which will not just give you an answer, but it will also tell you why. Did it come to that answer, sufficient enough to understand, okay, what's happening behind the scene, and that just helps us get trust into the system. And again, trust is something. It's a process, right, it doesn't happen instantly, so you need to give it several different images and see what it's doing, see why it failed, and that's when you build that trust. But then explainable AI lets you have a platform to build that trust into the system.
Speaker 1:That is so important. I've heard the term causal AI. Is that the same as explainable AI, or is that a little different?
Speaker 2:Well, causality is one way in which you can invoke explainability. Like you know, I came to this because of this, this and this, and then you know, all these things came together to make this decision. So, yes, like causality is like one of the ways to achieve explainability.
Speaker 1:Wow. So like it's got, it's multimodal, to what I'm hearing, and now you do a lot of work in digital pathology and that's a big word, right? So when you start adopting this type of technology for digital pathology and you work with histo-works and you're talking about workflows and efficiency, Just give us an idea of what this looks like, especially from a preclinical researcher. Lens.
Speaker 2:Yes.
Speaker 2:So just to give a context, I work for Histowiz, which is a biotech company, and we provide histology solutions. What I mean by that is, say, a customer gives us like paraffin blocks, like tissue blocks, so we, we do the grossing, embedding, cutting and all that workflow and then we digitize the tissue. More simplistically, we take a photograph of the tissue at a very high magnification and make it available to our customers, who can now switch from looking at the tissue under the microscope to completely viewing it on a big screen, on a big monitor. So basically, this is the digitization process on like big monitors. So basically, this is the digitization process, and we were established in 2013, basically as a solution for lab researchers to take away all the mundane tasks of what goes on to do histology. And from then on, we have grown as a company and we have adopted digital pathology solutions, and the core of our business is we produce high quality data set with the fastest turnaround time in the industry currently. So to achieve that, we have adopted digital pathology technologies to get there, adopted digital pathology technologies to get there, and one remarkable tool that we have is automated quality control.
Speaker 2:So just to give a context, so say you have an image and say there is an issue with the image. Right, it's out of focus, it's blurry, or there are like pen marks or you know, or there is some dirt on the image, things like that. So it's like then you feed AI, then you feed the image into the AI system. It's as simple as garbage in and garbage out. Like you know, the AI is only as good as the data you feed it and the algorithm. So what we did is a couple of years back and it's evolving we developed an AI for automated quality control. Now this takes a lot of human personal time to identify quality of images, which is quite repetitive and boring from their perspective, because they want to focus on more important tasks as to identifying answers to some questions, some research questions. In Histowiz we have an AI-assisted automated quality control that flags images which are like bad quality and it automatically like detects any issues.
Speaker 2:Uh, so, uh, we we have made like a lot of progress in uh this front, in uh trying uh to sort of accelerate pre-clinical research. So most of our customers are in pharma, academic institutions and biotech who are doing preclinical research, maybe for drug discovery or things of that sort. So our job is to make their research more streamlined and give them exactly what they need, which is a good quality data set, and provide a platform for them to do research. That's essentially what we do at Hisdoos, well that is, a good quality data set and provide a platform for them to do research. That's essentially what we do at HisDoits, well, that is a huge advancement.
Speaker 1:We're talking about biomedical imaging, intelligent diagnostics. We were talking earlier how things are moving from just information information technology to intelligent technology, information technology to intelligent technology, that there is a knowledge component that it can do things that faster number one than human thought All right, and it can present you with some different options that you didn't have before. As you look at that, you know biomedical imaging, intelligent diagnostics, and you marry that in the context of personalized medicine, what are you seeing as some future advancements that we don't have presently?
Speaker 2:Right. So that's a very good question and I keep thinking about it as I'm in this field. So what I envision is again to take a step back the field that I'm in right now, which is digital pathology, digital scanners for primary diagnosis and that is when people started digitizing the slides on a very large scale. So this was like seven years back. And then what happened? Parallelly, on parallel track, the field of AI was growing as well. It had nothing to do with digital pathology. These were like two independent entities, right. So this field emerged on a rapid scale and then AI took in as well. Now I think there is like a marriage happening between these two which needs to be evaluated, like. What I mean by that is, as the technologies are evolving within AI for example, llms and Gen AI and all that they might necessarily not be a solution to some of the digital pathology problems. So just because it's evolving both are evolving at a very rapid scale some of the AI that is coming up might not be the solution.
Speaker 2:So what I feel is, like you know, to keep it quite simple and basic and first to think like what are the problems in cancer diagnosis as of today? Like if it's breast cancer. Like you know, many women undergo misdiagnosis and this number is quite alarming. So there was a study in 2015 where it was published that, for pre-invasive breast cancer, more than 1 million women undergo biopsy every year, and I'm more than happy to attach a link to the paper in the comment section once the podcast goes live. But among that 1 million women, close to 48% of the women have misdiagnosis. So either they're over-interpreted or under-interpreted. So the problem really lies in this disagreement between pathologists to come to a certain diagnosis. So that's a huge problem, not just in breast cancer, but in lung cancer as well, in prostate as well. So that's one big problem. Second problem is the difficulty in identifying those clues within the image which can lead to a certain diagnosis for a certain stage and grade of cancer. So that needs to be thought of as well.
Speaker 2:So, to answer your question, how do I envision this in the realm of personalized medicine? I feel we still need to be on the track to address those problems. We shouldn't deviate ourselves to, like you know, maybe creating new problems by taking on new technology. We should still focus on what the problems are and try to bridge the gap between, like you know, misdiagnosis and correct diagnosis with AI and then see what more can be done with biomedical imaging and intelligent diagnostics. And I would say maybe you can think global but act local. You can think big about personalized medicine but then focus on what problems are there right now and, you know, try to solve that one by one, take it off and then move forward in the technology.
Speaker 1:I completely agree, and I've heard this a lot when working with people that are in the world of cancer, specifically even in breast cancer, in which there are women that have more dense tissue than other women, and then when they go through radiology and that type of thing, they get some of these images that it doesn't pick up as much and they're not attuned, I guess, to those anomalies. So I am moved by the fact that there's a different way of getting a better result, because what you said that number of misdiagnosis that's something that we can't live with. We're talking about precision medicine. We're talking about AI. We always feel it's more precise. We like the precision and the speed of the precision.
Speaker 1:Another aspect is deep learning. I love deep learning because when you really understand this is when I see the AI and machine learning really taking off is that it can go beyond our normal textbook understanding of certain things and come up with different results. And what you just talked about explainable AI, I said you know what, if the AI goes off the reservation list, let's say, and comes up with this whole new, different care package on how we want to diagnose and treat a patient, we've got to understand where it's coming from and is the data actually accurate? You do a lot in this deep learning world what you're doing with graphing and what you're doing with your company with Histoworld. Just talk to us a little bit about how you're engaging with that kind of technology, ready to elevate your brand with five-star impact.
Speaker 1:Welcome to the Firebrand Podcast, your gateway to exceptional personal growth and innovative business strategies. Join me as I unveil the insider strategies of industry pioneers and branding experts. Discover how to supercharge your business development. Harness the power of AI for growth and sculpt a personal brand that stands out in the crowd. Transform ambition into achievement. Explore more at FirestarBDMcom for a wealth of resources. Ignite your journey with our brave brand blueprint and begin crafting your standout Firestar feature today.
Speaker 2:Yeah, so I have had exposure with a lot of deep learning technologies, not just at Histovist, but with my background as well, doing PhD and even before that, like you know, while doing master's. But more so from the past six to seven years it has been for digital pathology as a use case and uh, again, like many of the technologies have come and some of it has fit, uh, rightly, into the field and uh, like you know, for example, there is like graph-based deep learning approach, right, uh, so there, uh, what you, what it captures, is not just the image and the features within that, but then how they are spatially arranged. So, again, like to give a very simple example, like, uh, just like, how important is a spatial relationship in identifying cancer For our viewers? To make it more simple, say, you have an AI to detect like a kitchen versus a living room, right, as simple as that. So you can say something is a living room, like you know, say, if there's a couch, there's a television, there's a coffee table and uh, things like that. But it is not just that, right, it is also like there's a couch, but then in front of the couch there's a television, and then you have a coffee table which sits between the television and the couch. Now if you identify each of these features independently like couch, television, coffee table, you may say it's a living room or you can say it's something else. But then if you combine the spatial arrangement of all these, like you know, your confidence to say something as a living room increases, like dramatically.
Speaker 2:Uh, similarly, in the field of cancer as well, it is not just these features, but it is the spatial arrangement of those features which matters a lot in uh diagnosing, like you know, the cancer types. So graph-based deep learning allows you to do that. Wherein you know uh, you can look at a spatial neighborhood and create a graph of that neighborhood and then you can have these multiple graphs and use that as a classifier. So it preserves these spatial interactions and it will give you more understanding of how the biological entities interact with each other. So there are different flavors to deep learning. This is one of that and at Histowiz, as I told you previously, we use AI for quality control and that is based on a simple deep learning method with few layers into the network and we have a training data set and we feed it images and then it says, like you know, if it has a specific issue or not. So the exposure has been great. But again, like you know, it's quite fascinating to see how different deep learning methods give you, like different kinds of information.
Speaker 1:I completely understand what you said because I got this image. I remember someone had told me this story small, short story a while ago and they said we took a man and we blindfolded the man and then we brought him into a room and we had him describe to us what you see in the room or what you're experiencing in the room. And he went in the room and he would touch something and say, oh, this is a trunk. And then he would go to another part. He said, oh, this is a tail. You go to another part, oh, this is a hook.
Speaker 1:And what he couldn't understand was is, when he took the blindfold off, that you were touching and feeling an elephant, and so you would have never known that by just getting the parts. As he was explaining it, he said, oh, this is very simple when you see the whole. So I feel that what you're giving us in your technology that you can see the whole room or the whole home and then the entities within it all, instead of just trial and error and seeing certain pieces of it, which is not giving you the entire diagnosis or the entire spectrum of what's happening I think that's very influential, absolutely Grant.
Speaker 2:And so to add to that here, what we're trying to do is we are trying to first build AI which is as good as a pathologist and then improve upon that. And now to do that, we want to mimic, close to how a pathologist would look at an image. They don't look at each feature. They have this perceptual and cognitive ability to absorb multiple features, put it together, view it as a whole, not as parts, to come to a decision. And that's quite fascinating to understand perceptually how they look at these images and come to a decision. So I leverage to take a few courses in cognitive neuroscience as well during my PhD to gain a deeper understanding as to how does decision making happen and you know, how are biases formed, how are confusions formed, like you know and use that to like sort of like you know, develop my own AI framework during my PhD.
Speaker 1:I think that's so important because we're human beings, we have to look through the neurological lens. That is, our own state of awareness is a part of that. Whether it's right or wrong, it's a part of that and we have to look through that lens and see what we can see and discern what's really there. Now you went to Carnegie Mellon and I think that was the University of Pittsburgh School of Medicine. How did that shape your approach to pathology?
Speaker 2:So first I joined the University of Pittsburgh for my master's and then I got into the PhD program, which is a joint program between Carnegie Mellon and UPMC, which is University of Pittsburgh Medical School, trained by an advisor who sort of like made, made it simple for me to understand, because I don't have the background of cancer biology. So my background is electrical engineering with a signal processing specialization and robotics. So I come with a different background. But my advisor, like you know, helped me like get into this field and make me understand as to what the problem is, and I spent a lot of hours in understanding this field and trying to not just from an algorithmic perspective.
Speaker 2:So we collaborated with Meggie Women's Hospital for a data set to work on breast cancer and my advisor was quite clear that to be able to graduate from PhD you cannot do that just by developing a deep learning approach, getting some accuracy, and that it has to be explainable, framework and trying to understand the core of where this data set comes from.
Speaker 2:And for that I had to spend hours together at the hospital along with a very well-known pathologist, one of my advisors, and spend a lot of time with him to understand, ok, from his perspective, what does he see when he looks at these breast tissue images and why is this case difficult versus that, and I had to gain all that knowledge before building AI, so I would say my PhD was at the intersection of being a computer scientist developing AI and computational pathologist as well. We're trying to understand the data set, but it was a fabulous experience and that experience has taught me that you know to be able to develop any good AI, you don't have a good understanding of the data that is coming from when you train these algorithms.
Speaker 1:So, yeah. I want you to help us understand this, because we have some younger folks that are listening here. Some of our people just aren't super technical, but they need to understand this because the framework of how we use technology is changing. A lot of people understand how to use a computer or a mobile phone, understand social media, things of that nature. Television has been around for a very, very long time, but our technology is changing at lightning speed. If you were talking to a fifth grader, how would you explain data?
Speaker 2:How would you explain data, ai and what you're doing with both of those things, right? So, to keep it very simple data, like you know, for my field it is just images that you see, right? And again I would like to explain it. Forget about cancer diagnosis, digital pathology and all these jargons. Just say you look at an image and you want to be able to tell what that image is right, and again say you have this small camera, you take a picture. You take a picture of a kitchen, you know it's a kitchen, but then say you want, like a system to tell, oh, it's a kitchen, like you know, wow, then for us, for a fifth grader, like he or she might think it's magical, right? Uh, so the data here means like a lot of pictures of your apartment, like you know it's kitchen, bedroom, playroom, like you know, uh, living room and all of that.
Speaker 2:Now, uh, ai will come in and tell you as to say, you give a new picture. Say you go to your friend's house, right, and you click picture there as well. Uh, now the ai will tell you, oh, it's, it's a kitchen. And now you you're like, oh, wow, okay, how did it do that? So then you start thinking of.
Speaker 2:Like you know, there's this technology component that comes into play between data, which are the images, and a decision, which is, like you know, if it's a kitchen or not a kitchen.
Speaker 2:So that's like you know. Then how does it do that? That means it's looking at the image, it is looking at things in the image, like you know, say, with the kitchen, it's looking at a refrigerator, a microwave, a kitchen top and things like that, and you know, uh, putting it all together and saying it's a kitchen. So when you go to your friend's house, when you take a picture again, the system tells you that you know it's looking at the same refrigerator, microwave and all that. Now your friend's kitchen might be very different, it might not have a refrigerator, it could have something else, but then the AI may make a mistake there. So then you come back and improve your data set, which is now. You go to your friend's house and take pictures there too and then add it to the data and then go to your other friend's house, and now we're in the continuous process. So I think that's how I would explain it.
Speaker 1:That's very good, and everyone has a cell phone and in that cell phone, I guarantee you have a lot of images and you know it's about understanding what those images are like. Probably let's not go to india. I'm gonna go to india. I understand there's multiple languages in india. I understand english.
Speaker 1:I'm very one-dimensional, right, but I know that if I was there for more time, I would begin to understand the languages, because language are images, because they're explaining something they're seeing and sensing and over time, I would begin to be able to make the sounds that go with that image. Yeah, and so as we begin to understand how ai works because it needs good data and then it understands just like a basic decision tree, if then right, it's the same thing. However, it's doing it a very, very fast, very, very fast level. So, getting back to our original and we're going to get to to, you know we're getting to the end of our podcast and what I heard you saying is that you're really doing some interesting work around cancer diagnosis, because a lot of people understand that, just in case you did have cancer and you're going through something like that, you go get a screening and then they're going to be looking through radiology A lot of times, you know as a user.
Speaker 1:You go in, they put some electrodes on you or they put this vest on you, and then you might hear some buzzing. You might not hear anything, and then they're getting an image and then someone has to then look at that image to see if there's some kind of anomaly that's taking place. You're doing something unique. I want the audience to truly understand what it is that you're doing and why this is going to be so impactful as we roll out more and more into the community.
Speaker 2:Yeah, so even now, the pathological diagnosis is considered as the gold standard. The pathological diagnosis is considered as the gold standard. So, like you know, you get a tissue biopsy and then it gets sent to the lab. They do all the processing. At the end of the day, the pathologist looks at it under the microscope and comes to a decision right. So I am in the field of what is called as a digital pathology so, which is like one step ahead, wherein you know, uh, now it's not like a physical glass slide under the microscope, but then you digitize it. And now, uh, you will be able to again, like you know, which I mentioned previously you'll be able to see it on big screen and, like you know, you can also, like uh, zoom in, zoom out. You'll have all the capabilities off the screen to look at those images. So now, uh, since you're digitized, it opens up lots and lots of opportunities to add in intelligent diagnostics on top of it, because it is digital uh, which was a limiting factor when there were just physical glass slides, where you couldn't do anything with that, just look at it and diagnose. So I am in the field wherein you have gone one step further and you took the images of the glass slide and then you try to answer questions as to can you detect these features, can you detect the quality of the tissues? Can you be a guiding tool for the pathologist for them not to make a mistake?
Speaker 2:Again, very simplistically, if you imagine a Tesla car, the thing is you are still the driver, but it is equipped with so many new functionalities and intelligence that will not, like you know, that will lesser the probability of you getting into an accident, like you know, with all of that. So it's like an assistive tool which has, like, several different features, like you know, and then you can either turn it on or turn it off. You have the capability to do that, uh, to avoid an accident. So, in my case, accident refers to misdiagnosis. So we are trying to build algorithms so that, uh, the misdiagnosis is reduced and it's more assistive, wherein, you know, uh, it cannot be only a pathologist, it could be a clinical researcher or, like you know, uh, in pharma or biotech companies, where they could turn it on to, sort of like, quickly see what's happening and then they can ask more important uh questions in the future. So, uh, right now, uh, we have a couple of ai tools on our digital platform called Pathology Map at Histobles, and we're actively venturing into third-party AI tools through our strategic partnership, because we have a wide variety of data set in the preclinical space across multiple different organs, not limiting to breast or lung.
Speaker 2:We have skin samples, prostate and many different organs. So we are launching a platform called Pathology Map where we would also let in other companies who are developing specialized AI, say for skin, or specialized AI for, say, lung they can come onto the platform and try to plug it in and our customers can use all these tools. And just to give a context from the past decade, we have worked with more than 400,000 researchers across 3,000 institutions, spanning across very popular pharma companies such as RevMed, calico, sanofi and all these companies, and we have scanned more than 2 million glass slides and we'll be doing it more. So to the platform we will be enabling, again keeping the message quite simple we want to build tools to help researchers, assist them in what they do, and just to the platform enables you to do that and accelerate research.
Speaker 1:You got us on the edge of our seats. I know I'm on the edge of my seat. We've got to know how to contact you. This is so important to make sure that this kind of research, these kinds of technologies. We've got to get over the fact that, you know, if everyone gets a little bit antsy about AI, I think it's going to take over the world. We watch too much sci-fi.
Speaker 1:To a certain degree, it's like no, these are assistive tools. I'm glad you stated that this is a better way of doing it. Most people are not very concerned when they go out and they get an x-ray, so you're doing something similar to that, but now you have a much more intelligent component that works around that, and I love the fact that. That explainable AI like all right, how do we get to the answer? We don't want, as we've heard, like in generative AI LLMs, you get hallucinations and there's certain biases and you've got to root those things out, because you've got to be very, very accurate, uh, and getting these diagnoses, uh correct, and we can only improve from where we are right now according to your statistics that you gave earlier. So, akash, how do we get in touch with?
Speaker 2:you. Oh well, you could reach out to me uh, either on linkedin uh, like you know, you can look up my uh profile or you could send me to me uh, either on LinkedIn uh, like you know, you can look up my uh profile or you could send me an email if you want to chat with me, uh, and probably you can put both this information in the description, uh, when the podcast goes, uh, live, uh, so I'm, like you know, just like an email or a message away, uh, message away. If you want to chat about the opportunities in digital pathology or opportunities and where the field is going for AI in healthcare, happy to have a conversation. Or if you just want to brainstorm ideas for how to adopt and invoke AI in healthcare, I'm all in for that. I'm all in for that.
Speaker 1:I'm very in for that. We got to get you into PEMS. I'm also doing another oncology summit. Her Health will be in Orlando October 26th and 27th. We'll be talking a lot around these different things to a great group of oncologists, gynecologists and others around career health. So be on tune for that. I also wanted because this is so important you've been a guest speaker at a lot of different conferences and you were just at a pathology conference. Tell us more about that.
Speaker 2:Yes, so this was in san diego uh last month.
Speaker 2:Uh, so this is uh the digital pathology and ai congress uh, that happens every year all over the world.
Speaker 2:Uh, it's it's a premier conference for uh digital pathology, which attracts like uh leading experts, uh not just scientists, but like pathologists and pathologists and the stalwarts of this field to the conference and gives me a great exposure.
Speaker 2:I have been going to this conference for the past three years as a speaker, as a speaker, and uh, this year was it was, it was beyond uh fabulous uh to seeing uh the kind of exposure and learning that I received uh during the talk or during like tea breaks and conversations and all that uh just to get like a like a brief sense as to where the field is moving, what people are talking about, like the assistive ai for pathology.
Speaker 2:That's good, and I think, like going to these conferences, like you know, it also put things into perspective as to what you are lacking and what you need to cover, to sort of like bridge that gap of understanding. And you know it's like it's one umbrella under which you will meet like all great minds together. So it's a very like energetic uh atmosphere and I'm sure it's the same thing with, with Helms and with uh, the other other conference in Orlando which I'm super excited to be like a part of and join Uh, but uh yes, so this is more tailored to only pathology and uh it and it was a fantastic experience.
Speaker 1:I can see that. I can see that this has been wonderful and I want to invite your entire audience to tune in to all the episodes of Follow Brand. They can do so at 5 Star BDM that is the number 5. That is Star BDM B for Brand, d for Development and for Masterscom Akash. This has been wonderful. I want to thank you so much for being on the show and I can't wait to work with you again, fantastic.
Speaker 2:Thank you, brad. Thank you so much, you're welcome.
Speaker 1:Thanks for joining us on the Follow Brad podcast. Big thanks to Full Effect Productions for their incredible support on each and every episode. Now the journey continues on our YouTube channel Follow Brad TV series. Dive into exclusive interviews, extended content and bonus insights that will fuel your success. Subscribe now and be a part of our growing community sharing and learning together. Explore, engage and elevate at Follow Brand TV Series on YouTube. Stay connected, stay inspired. Till next time, we will continue building a five-star brand that you can follow.