Read the First Edition of The Lens, a new magazine by LSI arrow-icon

Dusty Majumdar, PredxBio - Spatial Analytics & AI for Precision Pathology | LSI USA '24

PredxBio is converging spatial analytics and artificial intelligence to transform precision pathology.
Speakers
Dusty Majumdar
Dusty Majumdar
PredxBio

Dusty Majumdar  0:06  
I'm Dusty Majumdar, CEO of PredxBio. At PredxBio, we are transforming drug development and discovery using spatial biology and spatial analytics. So very happy to be here. So there are two major problems in pharma 90% of drugs don't make it and the drugs that make it only work for small populations of patients. These are the two problems that we are attacking right now. At PredxBio. For the last five years, we have been accelerating discovery and optimizing clinical trials using Explainable AI and unbiased spatial analytics. For a long time, it has been known that the tumor microenvironment is extremely heterogeneous, and it will start to be very chaotic. One of the things that we discovered around five years ago was that there is some order to this chaos. And there is some organizing principle that determines the order that exists in this chaos. That order really is in the form of the micro domains that we discovered a few years back, which serve as guides to explaining the heterogeneity. And once we discover the micro domains, we are able to get into the network biology of cancer, which leads to novel targets and biomarkers. So who are we? We are a company out of Pittsburgh, Pennsylvania, have been around for five years. Today, we have several customers as you can see at the bottom of the page, who we serve, with a proven ability to predict response to therapy with over 90% confidence. We're here to raise a Series A round the series a two we already raised series, a one which is $2 million, and we oversubscribed heavily already on that. Today we are here to hear raise series A two it's a participation preferred stock with a 22 million pre money valuation. Huge momentum for us the last six months or so where we have got a number of pharma customers to come in Bristol Myers Squibb Mark genmab, AstraZeneca, and 20 plus peer reviewed publications in journals like Nature, communication, cell reports, etc. 10 Plus patents in the last few years granted, a number of these are US patterns and some international and Stan marks is the chairman of our board. St. Mark's is the chairman also of the Hillman Cancer Center in in Pittsburgh, Pennsylvania. So what's the big problem that we are tackling here? For the last 10 years we have known that immunotherapy clinical trials have been growing tremendously to around 50% increase over the last few years. The thing that has remained constant is the response to therapy. 17% 17% of patients respond to immunotherapy at advanced age, the rest do not really respond. As a result of this what happens in pharma, of course the patients are in agony by the farmer. After spending $2,000,000,000.10 years, only 10% of the drugs are successful in immunotherapy at this point of time. This is largely because as I said about the heterogeneity in the tumor microenvironment and biology and cancer is immensely complex. This is what is extremely difficult to get into the underpinnings of. And that's what we focused on. So five years ago, we asked the question, what if you could identify subpopulations of patients who would respond to immunotherapy with over 90% confidence and radically improve clinical trials and the discovery process. This is what we have been able to deliver. And we're very proud to say that over the last few years, we have been consistently delivering this kind of accuracy in terms of predictive response to immunotherapy versus the PDL one based tests which are the standard of care in the market. We have been doing this consistently from biopsy tissues, which are labeled with X number of biomarkers, it could be anywhere from two to three to five 200 biomarkers. And we are able to leverage spatial biology to look at the entire tumor microenvironment in the tissue with the complexity of the tumor that you have with stromal cells, tumor cells, immune cells, T cells, B cells, all tweeting at each other communicating with each other, we capture those in the base of the proteomic biomarkers. And we also capture these micro domains, which are constellations of cells coming together in an identical spatial orientation. And this is going to be no secret sauce to connect that to the outcomes. Once we get the micro domains we are able to leverage what we call Explainable AI, which is the next level of AI over deep learning, machine learning etc. which is there in 99% of the different AI algorithms that you see from different companies. With Explainable AI, we are able to discover and reason about cause and effect. And that's going to be no secret sauce to leverage that and expose the network biology. This is our platform. We can ingest data from proteomics transcriptomics, simple brightfield pathology, and within a fraction of a second we are able to phenotype every single cell All That Is there on the tissue using this recursive tree and leaf algorithm on the right hand side that you can see. Once the cells are phenotype, we know who's who, what tumor cells, stromal cells, macrophages, etc. And once we know who's who and who's located where we are able to generate these micro domains, which are constellations of cells coming together in an identical spatial orientation, has said, these micro domains are critical, in terms of some of these could be tumor promoting. Some of these could be tumor restraining. And once we discovered these micro domains, we can correlate that to the network biology of cancer. So we have published all of this over the last few years in a number of different journals, from nature communication, to sell reports to discovery, etc. Marquee journals across the board. And we have also patented this technology 10 Plus patents, as I said, so let me show you some use cases in the last few minutes that we have, from discovery to clinical trials to translation. One of the things that we do is we not only predict all of this with a very high degree of accuracy in terms of response to therapy, but we also expose why why is this happening? Why is the subgroup of patient responding versus Why is the subgroup not responding because we get to the pathways of cancer. And that has kind of been our secret sauce over the last few years. This is our customers some of the projects that we're working on Gen Mab optimizing immunotherapy clinical trials, Bristol Myers Squibb predicting response, a combination of immunotherapies, etc. Let's get into some of the nitty gritty this is for a company called Gen Mab that we have worked on for the last couple of years. Disease diffuse large B cell lymphoma. The challenge was to identify responders versus non responders, in immunotherapy clinical trials, were able to do that with 93.3% accuracy, compared to where genmab was, basically point zero au Roc dramatic improvement over the last six months that we've been working on with them. And we have shown this consistently across different trials from biopsy tissues labeled with eight biomarkers. And as you can see, now, Gen math has put us across their entire range of clinical trials that are going on lymphoma, as well as some of the other solid cancers. And we are working our total pipeline from phenotyping, to microdomain. Discovery to network biology with them. This is one we're very happy customers, Chun Li talks about how we help them in phase one and two. And now they're going into Phase Three in terms of predicting the response to therapy, this is for colorectal cancer. The challenge was to predict the progression of colorectal cancer with a certain degree of accuracy, we're able to do that with 93% accuracy in terms of a 400 plus patient cohort, predicting their five year risk of recurrence compared to some of the other players in the market, accoya, exact sciences, etc, who are in the 60 to 75% range in terms of accuracy. This is for BMS, revealing the mechanistic differences in response, the combination of two immunotherapies were able to really explain why the combination works way better than a single checkpoint inhibitor. Again, Sara Hirsi one of our BMS VP is very happy with what we have done in terms of exposing the mechanistic underpinnings of to checkpoint inhibitors versus one. This is our pipeline at the moment. And, you know, these are customers on the right hand side, the market is easily over $100 billion for what we're doing right now across various cancers that across different therapies coming up including Carty and cancer vaccines, this is our revenue profile, we're going to make $4 million this year and rapidly increase some of the offerings that we have right now, you know, going from opportunistic services all the way to spatial concert services, as well as platinum service with micro domains and network biology. Come to our booth, take a look at this amazing view of the tumor microenvironment. With our spatial computing, with the Apple vision Pro, we don't really have at this point a direct competitor and we can talk more about it when we meet. This is our exit strategy. We are going to exit to the end of 2016 Seven very good valuation at this point to come in and get a huge return. So our target acquisition partners once you could expect from pharma, as well as on the data companies. This is the team I come with around 25 years of experience from GE HealthCare, IBM exact sciences etc. Chakra was sitting here comes from Jeff Hinton's lab, the guru of AI. And I hope that you come in and have more conversations with us about investment, that de risk platform great market demand, good customer traction, a world class team that we've assembled of oncologists to a physicist, of computer scientists and AI folks together, and we hope to talk to you about investment in series A two Thank you very much

 

LSI USA ‘25 is filling fast. Secure your spot today to join Medtech and Healthtech leaders.

March 17-21, 2025 Waldorf Astoria, Monarch Beach | Dana Point, CA Register arrow