AI in Healthcare: Revolutionizing Pathology With Deep Learning

Chen Sagiv, PhD, is the Co-founder and Co-CEO of DeePathology, which provides AI solutions for digital pathology. She is also the Co-founder and Co-CEO of SagivTech Ltd., a firm dedicated to image processing, computer vision, and GPU computing. Through her work, Chen empowers pathologists with AI tools to enhance diagnostic accuracy and efficiency.

Here’s a glimpse of what you’ll learn:

• [2:15] How Chen Sagiv’s early fascination with physics and mathematics shaped her academic and career path
• [10:52] Chen talks about pursuing her PhD in applied mathematics and physics
• [21:40] Chen’s decision to become an entrepreneur and launch a company with her husband
• [29:51] How Chen shifted her company’s focus from traditional computer vision techniques to deep learning and AI applications
• [39:04] Recognizing opportunities in AI and preparing for the future of medical technology
• [47:45] AI’s role in augmenting medical professionals and improving healthcare diagnostics
• [56:12] The future of AI in preventative medicine and its potential to revolutionize healthcare

In this episode…

The rapid advancements in artificial intelligence are transforming industries, but in healthcare, the stakes are especially high. Physicians, researchers, and entrepreneurs grapple with using AI to enhance diagnostics and treatment without replacing human expertise. How can AI improve efficiency and accuracy while preserving the critical role of medical professionals?

AI-driven medical imaging expert Chen Sagiv explains that AI’s primary function in healthcare should be augmentation, automating tedious tasks, and improving diagnostic precision while ensuring human expertise remains central. Deep learning models can analyze pathology slides with remarkable accuracy, aiding in early disease detection. Beyond diagnostics, AI’s predictive capabilities can accelerate preventative medicine, helping identify at-risk patients before symptoms arise. By integrating AI into healthcare, professionals can make faster, more informed decisions to improve patient outcomes.

In this episode of Transaction Healthcare, Zak Eisenberg interviews Chen Sagiv, Co-founder and Co-CEO of DeePathology and SagivTech Ltd., about the evolving role of AI in medical diagnostics. Chen shares how pursuing a PhD in mathematics and physics led her to entrepreneurship, the ethical considerations of automation, and how AI can expand access to high-quality healthcare worldwide.

Resources mentioned in this episode:

• Zak Eisenberg on LinkedIn
• Merritt Healthcare Advisors
• Merritt Healthcare Advisors email: contactus@merrittadvisory.com
• Chen Sagiv on LinkedIn 
• DeePathology
• SagivTech Ltd.

Quotable Moments:

• “AI should not replace doctors but serve as a tool to enhance their decision-making and efficiency.”
• “Deep learning has shattered the glass ceiling in medical imaging, solving problems we once thought impossible.”
• “The future of AI in healthcare lies in predictive models that enable true preventative medicine.”
• “AI is only as powerful as the data it learns from; good data is truly the new gold.”
• “Critical thinking is essential; AI should never replace the human ability to ask the right questions.”

Action Steps:

1. Leverage AI for augmentation, not replacement: AI should enhance medical professionals’ work by automating repetitive tasks and improving accuracy. Keeping humans in the loop ensures better decision-making while allowing AI to handle high-volume data analysis efficiently.
2. Adopt deep learning for medical diagnostics: Deep learning models can detect patterns in medical images that may be missed by the human eye. This technology improves early disease detection, leading to faster treatment and better patient outcomes.
3. Utilize predictive AI models for preventative medicine: AI can analyze patient data to identify early indicators of diseases like cancer and neurological disorders. This proactive approach enables early interventions, reducing the burden on healthcare systems and improving survival rates.
4. Ensure AI tools are accessible to medical experts: Giving doctors and pathologists direct access to AI platforms allows them to develop customized solutions for their specific needs. This democratization of AI empowers professionals to integrate technology without relying solely on external developers.
5. Maintain critical thinking in AI-driven decision-making: While AI can generate insights, human oversight is crucial for interpreting results and ensuring ethical applications. Encouraging skepticism and evaluation of AI-generated conclusions helps maintain trust and accuracy in healthcare.

Sponsor for this episode…

This episode is brought to you by Merritt Healthcare Advisors.

Merritt Healthcare Advisors is an investment bank with a unique focus on healthcare providers and their businesses.

Merritt leverages the healthcare industry expertise of its owner-operators, clinicians, investors, and advisors to develop surgical facilities that perform safe, efficient, and cost-effective procedures.

To learn more, visit https://merritthealthcare.com/.

Episode Transcript

Intro  0:04

Hello and welcome to Transaction Healthcare. I’m Zak Eisenberg, Vice President at Merritt Healthcare Advisors. Merritt Healthcare Advisors is an investment bank with a unique focus on health care providers and their businesses. Transaction Healthcare is a podcast focused on addressing questions and concerns at the intersection of healthcare transactions and business.

Zak Eisenberg 0:24

I’m Zak Eisenberg, a partner at Merritt Healthcare Advisors, and your host for Transaction Healthcare This episode is brought to you by Merritt Healthcare Advisors. Merritt Healthcare Advisors is a full service investment bank with a unique focus on healthcare. Merritt Healthcare Advisors leverages healthcare industry experience of his owner, operators, clinicians, investors and advisors to advise owners, physicians and entrepreneurs in the health care sector about maximizing growth and running successful transactions. To learn more, visit www.merrittadvisory.com I’m joined today by Chen Sagiv, who began her career as an image processing and computer vision algorithm developer. After pursuing her PhD in applied math, specializing in Gabor analysis and variational methods for computer vision tasks, and a year in Germany for a post doctoral, she went back into industry as an entrepreneur. In 2009 she co founded SagivTech, a company that specializes in the high performing computing and non recurring engineering research and development projects for imaging processing since 2018 she is the co founder and a co CEO of DeePathology. Chen, good to see you. How are you today?

Chen Sagiv  1:34  

Good to see you. Zak, I’m fine. How are you good?

Zak Eisenberg  1:38  

Good so. Chen, I know, just before we started this, we were talking about your background a little bit, and I’d love to start there and just hear really about your your time as a developer, and then what made you decide to pursue a PhD in Applied Math. That’s a very like all PhDs a pretty heady pursuit, um, so I’d love to hear what, what led you to all this, when we can start as early as you’d like, you know, in your childhood or in college, wherever, wherever you want to start.

Chen Sagiv  2:15  

Okay, so, okay, so let me, let me start as early as the age of 15, because, yeah, when I was 15, I knew that I want to, I knew I wanted to do a PhD in physics. I was fascinated with the physics, with the laws of physics, and with the, you know, the very structured paradigms. I had the privilege to have excellent teachers for physics and mathematics throughout my childhood and also high school. I was really privileged, and I learned to appreciate that more and more as I grow older, because I understand how influential they were in my decisions making, because they allowed me, I love mathematics, and they did it, you know, they didn’t turn it off. They turn it on. So, so I had this dream to do that. And so for me, it was very clear that this is going to be my direction. I had no hesitations. The only dilemma I had in my when I was 20 was to either go and become a medical doctor or to go into physics and mathematics and

Zak Eisenberg  3:31  

wait. We need to pause there for a minute. You were thinking about becoming a medical doctor at one point, tell me. Tell me about how you went from physics and math to a totally different, totally different interest area.

Chen Sagiv  3:47  

Yeah, I think that doing something that will benefit people and will to find cure to cancer and to I think this always fascinated me. My mom passed away when I was 21 she passed away from cancer. So I think that I always have this urge. I had this urge to save her. I failed. And I have this urge, I think, to do something which is really beneficial. So I think that it came from a very emotional point, because I’m really, I love the structure and the logic and determinism in math and physics, whereas medicine is not so much deterministic and, you know, logic. And then I realized, I mean, thanks God. I was clever enough when I was 21 to realize that, yeah, I will probably be a very emotional doctor. So it’s better for me to stick with no, I really thought so it’s I’m too afraid. I’m too it will make me it’s too much for me emotionally. So I. Um, you know, I still, I tried to enroll into medical school because it even then, it was considered to be quite challenging to be accepted. So, you know, so I wanted to to, you know, to smooth my ego a bit. So I got accepted to med school. But still, I, I went to learn physics and mathematics. And, you know, the joke was that, what if you wanted to learn physics back then you all needed to bring your birth certificate, because physics, I mean, physics was not, you know who’s going to to study physics. I mean computer science or engineering or but, but physics was considered to be too difficult and like, what are you going to do with that after you graduated? And, you know, I really remember that we started the first year, 120 students. Second year, only 60 survived. Because it really, it was really, really tough studies. I mean, it was super tough to study, both physics and mathematics. And since I knew that,

Zak Eisenberg  6:03  

so was this your undergraduate physics

Chen Sagiv  6:06  

undergraduate then, then I almost immediately continued to do my master’s degree in Medical Physics. So then it was the first time that I combined my passion to I would say, to to math and physics and to medical sciences, I had, like the best advisor, Professor Ray Solange Axelrod. She is really, she, she passed away about a decade ago, but she really was a role model for me, because she she combined things and she, she was, she was a fantastic person, a great mentor and leader, and very, very professional and, and

Zak Eisenberg  6:51  

by combined things, you mean she was multi disciplinary in her approach is that she didn’t come in, it just from And, and this is indicative of that exactly program you’re in medical physics is not something I’ve heard of many times. I know what you’re talking about, but, yeah, it takes uniquely, I think, creative people to take thoughts and concepts from different different fields and different disciplines and combine them together. But it makes sense knowing you now that she’s a role model for you, because it seems like you do this as well.

Chen Sagiv  7:31  

Yeah, I really, I think that she all, she also she was, you know, when, when you are making a journey in life. I think that, I think that you really, when you have like people that you can really look up to, that guide you through your journey, I think this is so not trivial, and it’s so I was really, in a way, blessed, because, you know, I, I want to great give credit to my parents. I had fantastic parents. So i My father passed away two years ago. It was my best friend, really. And my mom, she was a super, super, fantastic mom. So I think that I really appreciate that, that when you have people that simply want the best for you. It affects you. And it’s it affects you in the sense that you you are empowered by their support of you, by their belief in you. And I think that this is something that I as a parent myself. I think that I try to convey to my daughters that no matter what they do or they choose, I support them. And I think that this is just, this is a lesson for life. But I really believe in that. So, so, so I did my master’s degree, which was the evaluation of the volume of amniotic fuel fluid based on ultrasound images. It was really fascinating. And then, and then I went back to the to the industry and and I became a mother, and I became a mother. And I feel after becoming a mother, I said to myself, Okay, so the dream of becoming a parent checked. So now I want to pursue my dream from the age of 15, and I want to do my PhD. So so it was really no it was really fulfilling a very important dream for me and and and I was really privileged to have two fantastic mentors during my PhD, professor shukaze Ave and Professor Neil. So sometimes I did feel like there is a comedy of Moliere, the slave of two masters. So sometimes I did feel like I’m the slave of two masters. But but on the other hand, it was. Really, again, the multi disciplinary approach of two people who are, you know, just are looking at the problems from different perspectives. And I think it’s really made me push my limits intellectually. And it was a fantastic experience doing my PhD. I worked hard. I did my PhD when I was a mother to two twins. Were babies constantly sick, and I think also it was a challenge for my, let’s say not only enthusiasm, but also to my determinism to finish that, because it was absolutely far from being a simple and then and

Zak Eisenberg  10:46  

a challenging time in your in your life, to start something new like that.

Chen Sagiv  10:52  

Oh yeah, absolutely, absolutely. But let me be honest with you, I think that in some aspects, you know, I started my PhD after spending a few years in the high tech industry, and then I came back to the high tech industry, but when you do your PhD, it gives you some freedom, because you you are managing yourself, and you are responsible to to working hard whenever you have to. So it could be the middle of the night, it could be whenever you have time. So I think that to have this self responsibility and to be my own boss, this is something that I on one hand, is very, very difficult because you have to be very disciplined with yourself. But on the other hand, I think that this, this, this were the seeds of becoming, later an entrepreneur, because you have to take responsibility of what you’re doing, and you really you have to manage your project, being that your pH research, or then later on, being project, or, you know, your product or your company. In some lessons, it’s the free to be independent, to be your own boss. It’s, it’s, it’s a great deal of responsibility, and you have to accept the fact that you will have sleepless nights because of this responsibility. But on the other hand, you have freedom. And I think that I really loved this way to to to handle the tricky situation of having young twins were preemies and all the time sick on one hand, on the other hand, doing my PhD. So I it was a great journey that really culminated in having the opportunity to do a post doc, which I never thought I will do. I simply got an opportunity to so my my doctor, my doctorate was two faced. One was to do what my mentors wanted me to do. And then after they were happy, they told me, okay, now you can do whatever you like. So I spent like

Zak Eisenberg  13:03  

three months, and this was in Israel, right? Your pH? This was in Israel, in

Chen Sagiv  13:07  

Tel Aviv University, yeah. So I spent like three months reading all kind of things, and I I suddenly found a fantastic paper on the uncertain principle of information, and this is related to the Gabor analysis. So you probably are familiar with the uncertainty principle in quantum mechanics, where velocity and position, you cannot define them simultaneously. And in information theory, you cannot define the location of a signal or the value of a signal in its frequency simultaneously. And it’s very, very deep. It’s amazing. And this was the direction I was really fascinated by. And so my other half of the PhD was really around that, and I got the opportunity to do my post doc with one of the leaders in this field. So Professor Peter mass from Bremen University in Germany. It was really a fantastic opportunity that I could not resist, and for one year, you know, post doc is amazing, because when you do your bachelor degree and your master’s degree and your PhD, you have a goal. You must fulfill this and this, you must be successful. You must, you know, do these courses, and you, you have to do something which will satisfy other people, otherwise they don’t give you the degree in your post doc, you can fail. You can fail. You can fail. And this ability to fail actually allowed me to write papers that, to be honest with you, now, if I try to read them, I think I’ll lead a month or two to recall all the mathematics. It was really, it was really the very. Very deep information theory and uncertainty principle and group theoretic approach. And I really, you know, I was, it was really amazing. But after one year of doing pure research and and really a lifetime experience. I understood, it’s not for me. I need to do something which will be more time to Earth, more practical, that will have an impact on, on the on, you know, on real life, and not being some kind of, you know, a a some kind of a bubble where you are defining sophisticated mathematical rules. It was a lot of fun, but I think that it even drove me more into the direction of becoming, of going back to the industry, and this is one and the second thing to become an entrepreneur, because I really wanted to to do something that I believe in, that I think that brings value. And I loved the, you know, the high pace of the industry, where, you have to do things,

Zak Eisenberg  16:23  

and it seems like you also it’s so interesting, just going back to your time in your PhD program, lots of academics that I’ve been exposed to have almost the opposite view. If they leave academia, it’s not structured enough for them, and they have a hard time adjusting. Sometimes, there are some that fit into your category, where you’re in charge of your own schedule and your own your your own boss, as you were saying. But I think you’re probably more the exception than the rule. Most academics, if they if they decide to leave academia, I think they end up in like, a very structured corporate environment, because it provide, yeah, yeah. This is my experience. And again, this is just an anecdotal I know if there are any studies on this, but it, it’s, it’s, it’s so interesting that that’s when this idea, for you, looking back, start to blossom for yourself of how do you want to live your life, and how do you want to pursue things? And then on top of that, entrepreneurship, I think, in industry, is a perfect analogy or symmetry to a postdoctoral program of being able to fail and pursue interesting things. And yes, of course, it’s with a capitalist band, because that’s, that’s our global structure of how you enact change in this system, right? The the companies, the method of change in this system. But that’s, that’s such an interesting head space that you you come from, and kind of the evolution of of your thinking about this is really, it’s quite fascinating, because I, yeah, again, lots of friends and and family members and just people I’ve met who that have a PhD Background and and even pursued into post doctoral, and they had a lot of trouble adjusting to industry. I think also part of what probably helped you is that in this as well, is that you had some time in industry even before you pursued your PhD. So you were you had more life experience and and I’m also biased here, because I live here in the States, and in the States, most people who pursue a PhD never leave academia. They go from their bachelor to masters to PhD from the time they’re 18. So they go right from high school into academics, and then they finish their PhD, and they have no idea what they want to do with their life, and they actually don’t enjoy what they’ve been pursuing.

Chen Sagiv  19:09  

And I think during my master’s degree, I actually worked full time, so because the the scholarship for Master’s degree was too small to, you know, to actually, you know, just isolate myself and do my master’s degree. So I worked full time. And actually, when I took time off from work, it was really to to study. And this was, like, one of the most intensive times in my life, because I was either working or doing my master’s so even, you know, on a Saturday when I wanted to go and have some fun, I was feeling guilty because I had to do either this and that or that and then. So I think that I really spend a lot of time in the industry. I loved working in the industry, and I think that the. I do understand what you said. I think that when you are having, you’re spending, you know, like, several years in the academia, and I also have friends who did that you are, you are getting out into the real world in a way when you are already, you know, in your late 20s or in your 30s, and it’s, it’s different, because you are very used to the way things are going on in the university, and they are, this is quite different from what is happening in industry. So I had a good, I think, you know, a good experience, to evaluate these two, these two things. Yeah, and

Zak Eisenberg  20:42  

that’s not to say that I think training for especially PhD programs, is quite good for entrepreneurship as well, because lots of people who pursue PhDs think from a principles First approach, they’re really very skeptical of new ideas, but oftentimes they’re not scared of trying new things. All this lends itself very well to entrepreneurship. So, you know, I think what, where I jumped in earlier you were about to transition into, you know, part two of Chen’s life. So you become a mother, you’re pursued academics, and you leave your post a doctoral, and you decide now’s the time that you want to become an entrepreneur. So tell me about that thought process. And was it an immediate decision, you left and you knew that’s what you wanted to do, or did you take, you know, six months to decide and and figure out what you were going to do next? Yeah. So,

Chen Sagiv  21:41  

so actually, when I finished, so when I finished my first post doc here, I got a proposal for doing a second year from from boyman University, and I it was really attractive because it was a lot of fun, and the people there were great. And we got to spend, like, summer holidays in a fantastic, you know, mention of the university on the river. So it was not too bad. It was actually pretty good, but, but I said, No, it’s, I have i i need to do something which is more intense and more energetic. So I knew I wanted to go back to the industry. So I spent, like, I think, almost two years in another company, which was, it was called the diplar, and it was really fascinating, because we did, we did some kind of development there of a lens that could focus at any length. And I think that at some point I looked at the entrepreneurs of the company, and I said, My God, I I envy them. I mean, it’s, I want to be the, to be the entrepreneur in the company, not one of the, not being in the group of the algorithms developers a lot, although I really loved the research, etc. So I think it was really a process of that, you know, the pieces of the puzzle started falling through, through life. I mean, it was like, think that you are, it’s not like some kind of an I stood there and some, you know, some understanding just jumped on me. I think it was really a process, whereas I think that, and in my case, it was a process where at some point, both myself and nit my husband, we became entrepreneurs together. And this is even a more difficult thing to do because of many aspects. So I think that we both matured into that the time was spending a lot of years in the high tech industry, and I think that we really had the passion to do something that we believe in, something that we are. We bring our values, our passion, because both of us were, really, were colleagues, and we put a lot of effort and thinking and, you know, emotions into our the places that we worked in and and we felt, well, let’s, let’s take all of these and just put it in our own operations. So it was a journey, and I started being like an independent consultant for a while, and when things started to be a bit more solid, and things started to move on. Then, man, it San joined me, and I think it was a very looking back. I think it was one of the most courageous decisions that we made as a couple. You know, I like to joke and say that after you managed to raise twins without. But without killing each other. What’s What’s so difficult about starting a company? And maybe there’s some sense of truth in that, but I think that at some point we said, Yeah, we are complementary in many aspects, in the sense that we are not. I mean, I’m more in the tech and technical operations, you know, and things which are going to more technological side, and it sounds more in the business and the legal and the commercial side. So, so we are really complimentary and and we, we decided, let’s, let’s give it a try. And it was scary, because you can imagine that in the first in the first three years of establishing we did not draw any salaries. And, you know, it’s, and this is something that we had twins at home, and we had a trace at home. We and it was pretty scary and but I think that both me telling myself that the I think that we are not afraid of taking risks on one hand, on the other hand, we always make sure that we have some kind of net, of safety net, that even if we fall so, you know, falling down is is never fun, but even if we feel we have some safety net. And we were lucky enough after spending like, you know, not not some good years in the high tech industry, we had our safety nets, and we decided to take this risk. And you know, in 16 years later, I think that this was really one of the best decisions that we have made, because I love working with nitsan. I have fun. We managed to create a work environment which is very supporting, very friendly, very professional. And I think that this is something that I’m very proud of, that we managed to do that and very profitable. So this is also something that we are happy because and, and it was really, I think it was, and it is still very hard work. But I guess that in some way, we love it. We are a addicted to that, a little bit to the challenges we love to to be busy, to have a lot of things, to make decisions, to work together.

Zak Eisenberg  27:55  

When running a company is like a constant puzzle that you have to solve. It’s always changing and evolving. And for anyone who’s creative and enjoys solving puzzles, business is a is a very interesting puzzle generally, and it involves lots of different disciplines. So to your point earlier, it’s very inter disciplinary. So it fits you. It fits you quite, quite well. It sounds like I’m curious over that time and again. This is not in the healthcare space. We’re getting there with your your latest venture, and anyway, we’ll come back to some of your early experiences in relation to that. I’m curious how, over 16 years with SagivTech, which is your first entrepreneurial baby, let’s say, how have you seen that business evolve and your role within the business evolve over that time, obviously, it’s matured as a company, and what you were doing in those first three years, you’re probably not doing many of those same things today, or maybe you are. And outside of that, I’m curious how you’ve thought for that company which is more mature versus how we come to know each other through DeePathology. How you’re thinking around the goal of that company has changed, and from a macro perspective, how does it fit into the industry that you operate in and and also, how are you continuing to create value for the company and build it into something bigger and better and more impactful.

Chen Sagiv  29:47  

So, yeah. So yeah,

Zak Eisenberg  29:49  

a few questions in there. Yeah,

Chen Sagiv  29:51  

a few questions. So let me, let me try to answer, and if I miss anything, just, you know, I think that the when you. Are. When you are an entrepreneur, there is a question which is very sharp, although probably everybody has to ask this question. But when you’re entrepreneur, it’s really, really crystal clear that you have to bring value. You have to create value. So someone will be willing to pay you for this value. And bringing value a was something that we understood that this is something that we have to focus on. So as SagivTech we were, we knew that we wanted to work in the image processing and computer vision space, and back then, GPU computing was really super challenging, definitely much more challenging than it is today. CUDA was really, you know, very, very, you know, just a baby, baby CUDA and and we understood that there is an opportunity here. Because if you want to accelerate image processing tasks and you want to do it on a GPU, you need to do optimizations. And this is how it evolved that we had like, after a few years, we had like, I would say, one of the strongest teams in GPU computing, not only in Israel, but I would say, worldwide. And I think this proof to that is that we worked boo with, you know, we worked for a few years, and one of the leading projects of Google, the now, it’s not night that I forgot a name, so, so you have to edit that. But the what’s the name? What’s the name? A, okay, so we worked for a few years on one of the No, but I, how did you call the 3d CA, the phone of Google?

Zak Eisenberg  32:01  

Phone? Yeah, Google Glass. Did you mean glass? No, Google

Chen Sagiv  32:07  

Tango. Tango. There was a project of Google Tango. And Tango was they developed their own phone where you had the ability to have that a, let’s say, sensing. Oh,

Zak Eisenberg  32:22  

I remember this, yeah, yeah.

Chen Sagiv  32:24  

So it was a of Google eight app, and we were, like, leading the GPU optimization of that project for four years. It was absolutely a amazing and fascinating and, and, you know, and I think that really we had a chance of very good years in doing optimizations, especially for GPU and and we brought value. And I think it turned out to be a very successful operation. But you know when, when you have a business that is running for some time. The problem with reality is that reality changes and and you need to adapt. You need to change, and sometimes you need to change even before the changes actually happen. Otherwise you become irrelevant. And I think this is one of the most important lessons that I’ve learned through entrepreneurship, that what worked yesterday might not work tomorrow. So this is really exhausting, because you need to keep track on things. You have to constantly learn. You have to constantly change. And you know, it’s very tiring. It’s very tiring, especially when I think about the how the work space was my for my parents, for my grandparents, so and now I think that the work environment is not only is constantly changing, not only for entrepreneurs, but only also for people who are just, you know, they are working in a company and One of the

Zak Eisenberg  34:01  

and the rate of change is faster now as well. Absolutely and 50s, yeah,

Chen Sagiv  34:07  

yeah, so, so back in 2014 I got tips for two very good friends of mine who are really super clever people, both of them entrepreneurs. And they told me, you know. And if people in the computer vision industry, if they don’t go into deep learning and AI, in a few years, they will be completely obsolete. And back then, in 2014 it sounded super strange, because, you know, okay, AI was really not, you know, it was really getting started so, so it was not part of the it was not what it is today. But I listened to them, and I said, Okay, we must be prepared to this. Ai, a thing. And,

Zak Eisenberg  34:53  

yeah, deep mind was just in its adolescence. Then, speaking of Google, they were only founded in the. 2010 or 2011 Yeah. I remember reading some of those early papers, and they probably were keeping track of some of these early research, early research institutions, yeah,

Chen Sagiv  35:15  

but I think that at that time, I mean, we couldn’t expect, I mean, we couldn’t, I think that we couldn’t really expect what would happen in a decade, but still that I feel that in 2014 there was, like we started going into deep learning, started to do projects in deep learning. And it was really mind blowing, because when you do computer vision in the old fashioned way, like in classical computer vision algorithms, there was always seem to be a glass ceiling of what you are able to do. And in some projects that we did, we there were some, you know, like some, some problems that we we couldn’t really solve, and and then we revisited these problems using deep learning tools. And we were fascinated, really, our jaw dropped because we saw that, my God, we these problems can be solved all of a sudden. And I think that then there were some processes within the SagivTech that really changed the way that we perceive it, because we we the GPU computing optimization per se was starting to decline because there were more libraries and GPUs were becoming stronger and stronger. So really tweaking every bit and byte was not so needed. This is we saw one. The second thing was the, you know, the incline of AI and deep learning. And also, I think that we SagivTech is, was the was, and still is very dear to us. But the role of SagivTech in our lives changed in the sense that we started viewing it as some kind of an incubator to ideas that can lead to our own products, because after spending, like, I don’t know, six or seven, five to seven years as delivering high end technological projects to companies and then,

Zak Eisenberg  37:16  

but truly services, But truly services, other companies creating

Chen Sagiv  37:20  

something fantastic, and then, you know, like growing a small baby, and then someone else is having it. So I think that, in that sense, for myself and its on, we really started to say, Okay, we have our own company that does services, and we’re doing a good job, but we want to take to look at SagivTech also as a platform that will allow us to grow into making our own products. So and so this was, again, it was not something that just jumped us, you know, but it was really a process, and we took the time to to really evaluate ideas and and to to stay tuned for opportunities. Because, you know, opportunities are things that you meet from time to time in life, and I think that the and I mean, usually you have some control on opportunities, but not really. I mean your opportunities, you can look for opportunities, but but the actual opportunities you would meet. It’s not entirely up to you, but what you can do, I think you can be prepared to the point in time when opportunity might meet you. And we had, really,

Zak Eisenberg  38:43  

there’s a perfect saying in English for this, which is that fortune favors the prepared mind exactly, because you’re so right. We’re all presented with opportunities in life, and some with more than others, and this is comes down to luck. The question is, how do you take advantage of those opportunities when they come

Chen Sagiv  39:03  

along? And since we were already in the state of mind that we were ready to to do something which will be different from SagivTech, we also understood that going to the phase of having our own product might really make the role of SagivTech smaller in our life professionally, because maybe our focus will be on something else. And to be honest, it was emotionally a bit difficult, because, you know, SagivTech was, was it still is very dear to us. But then we the opportunity came along. We in 2017 we started doing a project in Roche. It was really, I think, the first time that we did a deep learning project in the field of pathology. And the aim was to assist, to help the in. The part of the pharma which does the early research and diagnosis in the, you know, in the long process of a drug development, to really try to build tools that will help pathologists to assess slides quantitatively and and it turned out to be pretty successful. It was a good project, you know, and just just another project. But then the opportunity came, where we got an invitation to join the Roche digital health accelerator, which they decided to do, because they said, Roche is a giant, you know, it’s a huge pharmaceutical company. 125 years of experience, a huge boat, and they were searching for small speed boats that are, you know, just are swimming, are navigating the water more rapidly. So, so they invited us to join their first, let’s say, trial of having startups say, going through their healthcare accelerator. Now it is called the Korea sphere, and I think they already are in batch 15 or 20. Back then it was batch zero. So we were like, we it was also allowed to fail. And this was an opportunity to to understand how we can take our knowledge, the technology, the the insights, and to have the opportunity to talk to leaders in Roche that will tell us what is needed, and it’s it was a fascinating time of six months that we’ve spent in Munich. So we were traveling back and forth. It was before COVID, so it was very trivial to then travel back and forth. And at the end of this adventure, which was really a huge, huge opportunity, we realized that we want to do something which will bring value to drug development, and we can bring this value with creating a platform that will bridge the gap in a way between AI tools and AI solutions and medical practitioners and pathology specifically, and and then we were joined, myself so

Zak Eisenberg  42:39  

well suited for your your background, personally, and that’s what’s so amazing about this, because you did your master’s degree in Medical Physics, you thought that you wanted to be a doctor at some point. And now fast forward, however many years later, you’re presented with this opportunity, which overlaps perfectly, with your interest areas throughout your whole life. It’s destiny for you that you should be working

Chen Sagiv  43:08  

in just a couple of days ago, I had a very interesting discussion with a potential, potential collaborator, slash customer. And they they told me about the problem that they have that has to do with some kind of making decisions on on, on where to cut a tissue during surgery, and being able to determine that with making a a pathology analysis during surgery, and so, so we are discussing that, and we had that, we are setting a meeting, etc. Okay, that’s great, and, and, and it fulfilled me with appreciation to the fact that what we are doing is we are trying to bring technology where it matters the most to we and to really bring value where it’s super important. And I told nitsan just after the call, you know, of course, you know, when you cancer is not fun, it’s not it’s not something that makes you happy. But if you are creating a tool that really helps in diagnosing, curing, creating a developing medicine, medicines to difficult, you know, to severe diseases, to severe conditions. This really is, was and still is a source of motivation for us, because we really feel that what we do brings value, not only from the technological point of view, but with the fact that technology really brings. Value to to life, to medic, to the to medical research.

Zak Eisenberg  45:06  

It’s such an interesting point because, you know, of course, lots of people listening to this will know some of what’s happening in the AI space overall, and probably most of what they’ve heard is some of the more trivial uses for, say, you know, there are AI tools now for writing jokes that have been LLM models that are specifically tuned to this. And this is lots of fun, right? This is great and,

Chen Sagiv  45:36  

and we’ll make a lot of money from that.

Zak Eisenberg  45:39  

Yeah, I don’t know about the money, but it brings lots of joy to people, and that’s important too. But there’s so much that is happening, I think, in these less known fields, particularly in health care, that have real potential of absolutely changing the world. And you know, of course, this was already years ago, this was already, as you as your friends said, this was already somewhat envisioned, you know, with alphafold and some of these other programs that came out years ago, there was always the hope that something like DeePathology would come along. I just like to spend a little bit of time now talking about DeePathology, but also hearing your perspective and as a leading entrepreneur in this space, really the tip of the sphere for this type of technology, new company, it must feel somewhat profound. I know some people who were around during the comm era, starting companies, and there was this feeling of just total optimism and ecstasy, almost because everything was moving so quickly. And it seems this way now in AI broadly, but also in your niche slice of AI applications as well. So maybe just talk a little bit about DeePathology and give you a plug for what you’re doing now. But also, I think then zooming out a little bit to the macro of how you think technology like DeePathology. And you know, again, I hope that DeePathology is a winner in this space. But how this technology you think will be used in the future, what the what the eventual goals are for, really, for your company, but I think for lots of entrepreneurs in the space, what is the utopia of usage for this type of technology?

Chen Sagiv  47:46  

Okay, so I love your question, because I spend a lot of time thinking about it in, yeah, in I know you have absolutely so I think so. Let me start from the broader perspective, and then I go pathology. I think that we are fortunate to live in an era where technology can transform health care for the best. I think that we are only scratching the surface. And I think that our probably the generation of my daughters and my future grandchildren, etc, I think that they will benefit from it immensely. Because I think that right now, there are, like a few phases of our AI, and this technology can really benefit health care. So I think that the first, very first layer, is to automate some of the tedious, time consuming, you know, where you don’t have enough medical people to do things, so, for example, to serve as decision support systems, to serve as a priority, prioritizing, let’s say cases like in when you are suspected for for stroke. So you have to go and you have to do the city. And let’s say that you have, like now, 20 cities. So who shall you start working on? And then a decision support system can help you say, go first with this guy, because the AI system thinks that is most probable to have the stroke. So it’s really to take things which people, the medical professionals can do, or to, you know, to to analyze pathology slides and look for, are these cancer cells? Are they sensitive to some kind of drug, etc? So all of these are tasks that humans can do. I mean, professional humans can do. And. But some of these tax can be automated, can be accelerated, and can they serve as the assistant for the medical professional? So as a patient, you benefit, because you actually, you can get results faster, you can get the priority you need, and you can really be served better. So I think this is the first layer, and I think that this is mostly what we see nowadays in radiology, pathology and general data.

Zak Eisenberg  50:34  

Maybe a good word for this is augmentation of current current practitioners, or current professionals in the space you’re augmenting their productivity? Yeah,

Chen Sagiv  50:47  

absolutely. I think that the next the next day layer would be related to prediction models. And prediction models are really something that you can think about in many, many aspects. But I think it will be easier for me to go to the pathology point of view, where, for example, when nowadays it’s very wide spreading to for pharma companies to look back at clinical trials retrospectively, because in a clinical trial, let’s say that you have the result of a medical trial was that if you use this drug, there are 60% of success for the patient now, but when you look at the specific patient, you have no idea where this patient is, what are the chances of this specific patient? So if you look at all the data that was gathered in this clinical trials, and you and for each person, you have the success of the drug during this clinical trial, then a task which is now being done a lot of times, and we are also participating in that is to create drug outcome prediction models. So when you are you want to to know what are the chances of this drug to be efficient for this specific patient, you can use a model that will guide you and will give you personalized, I would say, prediction, and this goes hand in hand in the very important concept of personalized health care. So I think that after augmentation, the next phase is really a AI as an enabling technology for personalized health care and to facility to facilitate that, you need to resolve a few things. You need a lot of data, and you want this data to be legitimate and regulated and anonymized, and, you know, and it is true that data is the new oil, and good a good data is, really is gold, okay, so, so it’s not enough to have just data, but you need to have good, created and legitimate data. So, and I think that this is something that we see, and we hear of, and we discuss that the data should be entity that should be handled very, very seriously and carefully and based on the data, you can really develop tools to match the right diagnosis to match the right medication with the individual. And so I think personalized health care is something that we definitely see in pathology, but also in other domains where it could be also related to other kinds of data in health care, and we are now, I think, I think that the market is now really the front of this market is in the personalized health care. How do you develop new biomarkers? How do you create prediction models? How do you develop drugs more efficiently? And no, it’s super interesting to see that new AI companies are actually pharmaceutical companies. I know. Okay,

Zak Eisenberg  54:32  

it is fascinating. It is fascinating. Some, some of some that I’ve heard of that maybe you have as well, is is taking markers in speech that are in detectable, of course, to humans, but to a AI model can detect pre, pre markers of say, dementia or. Various other, yeah, various other neurological conditions well before there’s actual symptoms of this. And you can start treating it early. And there’s lots of examples of this from, you know, heat biomarkers throughout the body. And it’s, it’s a fascinating time because it’s, it’s so hard to predict what these models will be able to predict, because we we have no idea how to really connect this data ourselves, but we know that these models can can potentially and draw conclusions in different ways. So, so So we first, have first layer, and I think this is true outside of healthcare as well. I think what you’re going through is sort of true for any application, really, of of AI, but it is specific for health care too. So the first is augmentation, the second layer, and then I know DeePathology is working on this as well, or prediction models. And what’s the third, and maybe, you know, the the the potentially, the final layer of of advancement, yeah, potentially. This is why I say potentially, because who knows, maybe there’s more. But I

Chen Sagiv  56:12  

think that, you know, AI, I think that it was one of the things that it has taught, probably not only me is that sometimes you are surprised. You I think AI was not expected to be so dramatic back in 2010 because we had a lot of discussions back then with, with, really, with, with leaders in the industry and and engaging in very interesting discussions of where is the what is the next thing, and, and I think that it took some time to understand that AI is going to be the next thing. So I don’t know if it’s going to be the final layer, but definitely, in my view, the next challenge, or the next layer, would be preventive medicine. It means that while, at the moment, and I think it’s really corrupt, also a corresponds to what you were saying, but where augmentation, and then to be able to really be very specific and personalized when someone is sick, so to to match the right drug for them, and then to really be able to evaluate and to predict sickness and to predict a, let’s say, life threatening conditions, being that stroke or heart attack. And I don’t know one, one of the fantasies that I am I know, if you know it’s, it’s to have, like a wearable that will alert some caretaker when you are and it will tell you, Hey, you’re a little father or whatever is about to get a stroke in like, the next day, 24 hours. Be aware of that so. So this is maybe this might sound a bit of science fiction at the moment, but I think it’s going to become science in the next next decade or so, and I think that

Zak Eisenberg  58:16  

the sound unrealistic to me. I mean, some of the companies I just mentioned, which I know this research is happening are showing results in this case. And you can imagine lots of different types of biomarkers that we have no idea about, which AI is going to inform us our potential biomarkers for some of these diseases. Yeah, it’s a fascinating space. I would just add also to preventative medicine. In this space is also the advent of AI agents, potentially, so going back to kind of first layer augmentation, potential, just replacement of pathologists or radiologists and and, look, this is, this is a somewhat, this is a somewhat controversial viewpoint. I do think there’s always going to be a use case for human radiologists, but at least in radiology, it’s very clear that AI agents doing this will be far more, far more efficacious than human radiologists. And so I think at some point the health care industry globally will need to face this reality. And and it may, it may not mean that 100% of radiologists are replaced, for example, it but the likelihood is that 90% 95% will will be replaced because I will want to trust it from the patient standpoint and from the society standpoint, it will be cheaper and it’ll be more efficacious. So why wouldn’t we replace absolutely ideologists?

Chen Sagiv  59:54  

I want to remark on that because, well, maybe I’m old school, but I. I honestly believe that there will always be a human in the loop. I think that the radiologist. I don’t think the role of radiologists and pathologists and surgeons. I don’t think that it will vanish. I think it will probably dramatically change in the sense that some of the tasks which are more quantitative, or, you know that where you have to look for mathematical predictions, which humans do not do, I think that these tools will eventually serve doctors to make better decisions. Because, you know, I think that where, for example, if I may use the example of ChatGPT, yeah, that I strongly believe in the human in the loop, always. And I wanted to to give it as an example, ChatGPT, where you know somehow, if you think that ChatGPT can know everything for you and can write essays for you and can program for you, it, I don’t think so. I think it definitely can assist you in doing these tasks, but you, as the human, you have two roles, which I think will not change, or it will, they will still remain the human tasks. The first thing is asking the questions. So I think that as humans, we have the task to really create the right questions. Maybe these tools can give us the right answers, and maybe not but, but we need to be to have critical thinking, to be able to ask the right questions. And you know, in like in in one month from now, I’m going to speak to give a talk to a group, a group of Gifted Youth and and the topic of my talk is going to be how not to lose your critical thinking when you’re using AI, because you know, for example, when you are using it, when we are using a navigation system, at some point, you may lose your understanding of your surrounding, of where does where’s the north, where’s the you know this city and that city, because all you know is go right, go left, go right or left and and I’m not joking. I see that, especially with with younger people who have no idea, they have no context of the environment. So so really, maintaining critical thinking and asking the right questions is essential. I think that this will stay in the human side. And the other thing, and it’s still about critical thinking, is assessing and being able to understand whether the results the recommendations of the AI system. Are they making sense to us? We will need, as humans to maintain our ability to question these, let’s say, to question these platforms and to be able to have metrics to see how good they perform. And so I view AI tools in general and in medical applications in particular, as tools to assist humans to make better job. And so really, I view this not as a replacement, but as enhancement. And I already think that the roles of doctors is changing. They really need to use more sophisticated tools, and definitely in pathology, we see, you know, we so now I go to DeePathology. In DeePathology, we our solution is Do It Yourself platform that allows pathologists and medical experts to create their own AI solution and to the problems that are of interest for them. So rather than the usual flow where you have a problem, you have data, you give the problem and the data to the AI experts, and they develop the solution. They consult with you, but they are the owners of creating the solution. We did things a little bit differently. So we say, Okay, we have a platform, we have AI tools, and we give you the pathologist, the AI tool. So now you have a playground that you have to understand the problem, you have to understand what you’re looking for, and you have to have. Have good data, but once you’re giving

Zak Eisenberg  1:05:01  

it to the researcher instead, exactly right?

Chen Sagiv  1:05:05  

And I think that once the people on the other side understand that they now have the ability to develop AI solution, they love it, because they don’t have to beg some computer scientists to develop the solution for them, but they can really train a new AI model for a specific type of tasks. Or we don’t solve all the pathology problems in the world. In our platform, we aim to expand that naturally but But definitely, there will always be room for customized solutions, but we have like several problems that pathologists can build AI solutions on top of our platform. And we feel that in some way, we democratize the AI solutions creation because we bring the AI capabilities to the hands of the pathologies, and this really is also in sync with our philosophy that AI should really be a tool, and not only for computer scientists, not only for AI researchers, but a tool for People who are experts in other domains. So then one of the one of the challenges of AI would be to bring AI to people who are experts in different domains, and to bridge the gap between real life applications and the complexity of AI. So how can you bring AI to more and more a, let’s say, a professions, so not in order to push people away from the profession, but on the other hand, to make them more sophisticated, more professional, and to keep to keep them up with the advances of this technology. So

Zak Eisenberg  1:07:06  

I tend to agree with this philosophy. I just wonder, and I think it’s still too early in this space, or on lots of spaces, to really know how this will shake out. I do think, of course, there are lots of companies that have a totally opposite philosophy, which is that we should replace all people with AI agents and and never have another. Never have people doing things again. But I, first of all, I just don’t know, in that scenario, I don’t know how society really functions. Have that type of outcome, but

Chen Sagiv  1:07:44  

I think for the sake of society, people should be kept busy.

Zak Eisenberg  1:07:49  

Yes, so this is ultimately my point is, this might be possible, but is it practical for humanity? And the answer is probably no. But the second piece here is, and I think you brought it up really well, is that the it’s not totally clear, I think, to anyone I speak in this space, and it sounds like you have had similar experience, is that it’s not totally clear that AI is really going to be able to fulfill the why aspect. They can help you with the how and the what, but it can’t really fulfill the why. The as you the term you were using is critical thinking, but really asking the hard questions, not the easy questions of, where is this, or what is this definitional things, or even potentially stepwise Solutions, and I’m sure you’re familiar with with what’s happening in the in the AI software development, developer agent space, which is advancing extremely rapidly, Absolutely, but even in that space, the why of how you need to or what you’re trying to accomplish, and why you’re trying to accomplish it for what need in the real world, those systems are not able to answer those types of deeper questions.

Chen Sagiv  1:09:17  

So I agree, yeah,

Zak Eisenberg  1:09:19  

it’s, a I could talk about this all day, and I often do with my brother, who’s in the space too. But hen, this was great. I think this is a great place for us to end. So really appreciate you coming on transaction health care. I think this was a fascinating conversation. I’m sure the audience will agree.

Chen Sagiv  1:09:38  

Thank you, Zak. I appreciate the opportunity, and I have to say that I will be happy to join the discussions of your brother in yourself, because they sound fascinating. I really enjoyed the talking to you. And, yeah, I think that we are living in a in super interesting times. And I really believe that AI. I will enhance good in our lives, and will bring more good to more people. And I think that the the ultimate challenge for us as society would be, in my opinion, when a I would bring value to to people who live in poorer countries, and to bring high end medical care to places that are not experiencing it right now. And I think that this, this would be probably a very, very huge challenge of us as a society. Let’s see if we will live up to it. Hopefully we will. Thanks Absolutely. Thank you.

Zak Eisenberg  1:10:45  

And that wraps up another episode of Transaction Healthcare. Hit the subscribe button to get notified when we release new conversations. And if you are someone interested in learning more about these topics, visit us at merrittadvisory.com, or send us an email at contactus@merrittadvisory.com.