0:00:10 - David Williams
The microbiome holds fantastic promise for unraveling the mysteries of human health, from autoimmune conditions to cancer to longevity. Insights at the individual patient level should prove invaluable. But today it's awfully hard to interpret microbiome tests due to their complexity and the difficulty of keeping up with the fast-moving scientific literature. Today's guest, Leo Grady, is tackling this challenge head-on as founder and CEO of Jona, which developed a proprietary large-language model to leverage artificial intelligence to generate microbiome insights tailored to the individual. Hi everyone, I'm David Williams, president of Strategy Consulting firm Health Business Group and host of the Health Biz podcast, a weekly show where I interview top healthcare leaders about their lives and careers. Leo, welcome to the Health Biz podcast. Thank you for having me, david. So it was fun to meet you earlier this week in Washington DC. I got a good sense of everything that's going on, so very excited to do this podcast today. I wanted to start off by asking you about your background and your upbringing and, in particular, any childhood influences that have stuck with you throughout your career. 

0:01:19 - Leo Grady
Well, I grew up around microbiology. My father was a virologist, and so I loved going into the lab with him as a kid. Ultimately, my own path took me into computers and then ultimately into artificial intelligence, but I always had that grounding in microbiology from a very early age. 

0:01:39 - David Williams
Nice. So what did you study in school? 

0:01:42 - Leo Grady
As an undergraduate I did electrical engineering and anthropology and then in graduate school. My PhD is in cognitive and neural systems, which was a very early neural networks-focused graduate department at Boston University. 

0:01:57 - David Williams
Great. I've seen in your career progression things that make a lot of sense for where you are right now. A lot of our listeners and viewers are people that are earlier mid-stage in their career journey, may aspire to be a CEO or founder at some point, but I think it's very useful to be able to just sort of trace through with the steps of somebody like yourself. I'd love to hear about your first jobs and then moving on to Siemens and HeartFlow and Page and so on. It's like a good progression. Certainly in looking in retrospect it looks good. I don't know if it was so planned along the way. 

0:02:33 - Leo Grady
Well, I started really as a kid working in data entry jobs. I grew up in the 80s and 90s when computers were still new. This was pre-internet. In a lot of cases I was really interested in computers. I worked with computers a lot. I was doing data entry jobs. As I went into college in electrical engineering I started working. I did a summer internship at IBM and worked with an economic policy group building models for them computer models. Those were definitely interesting. They were just scratching the surface. They were using computers but not really taking advantage of the power of them. It was in graduate school where I got excited by signal processing and computational modeling and computational biology. That led me into graduate school then, ultimately through the career trajectory that you're talking about. 

0:03:38 - David Williams
Yeah, it's interesting we're roughly the same age. I might be slightly older. I worked with computers too as a kid. When doctors first got ahold of computers I remember working with some people that were trying to do some things like clinical decision support, worked on some of those first models and decision trees, really using mumps. The physicians were mainly interested in computers for tracking their investments and things like that. That was a way that they first got interested. Then a lot did start to tinker with them as well. Some of those early programs done by doctors or technically-oriented ones are also very interesting. Say a little bit about. I think HeartFlow and Page in particular were interesting to me. What were those companies about? What did you learn there? 

0:04:28 - Leo Grady
As you pointed out, I started my career at Siemens, the big German multinational, where they primarily build imaging equipment MRIs, cat scanners, ultrasound, pet machines. It was great to work there because I had exposure to so many different areas of medicine the cardiology, the oncology, radiation oncology, OB/Gyn, musculoskeletal. I got this really great picture of the whole all of medicine and got to work with many different types of doctors, learned a lot about anatomy and physiology. At that time I had this feeling that as AI was getting more and more powerful, that it could be more and more transformative. At some point it occurred to me that Siemens is ultimately tied to the hardware, that this was a hardware company that built really top tier, excellent hardware. If the value was going to shift to the software, I didn't think it was going to come from a hardware company. I had the opportunity to join HeartFlow, where we built this cardiovascular diagnostic test for coronary disease. That was a combination of AI and computational fluid dynamics. We took that test. It was a purely software-based test. It took a cardiac CT as input and then we ran the test on top of it. We could work with hardware from Siemens. We could work with hardware from GE or Philips or Toshiba and now Canon. That test was ultimately separate from the hardware, from the data acquisition. It was the first time that a test like that, a pure software, ai-based test, was paid for by Medicare and then ultimately all of the private pairs. We brought that to market in North America, Europe and Japan. 

I was recruited to PAGE in New York, which was a spin-out of Memorial Sloan Kettering building AI and pathology. The reason it was a good fit for me to go to PAGE is that PAGE wanted to do something very similar. They found the value to be in the AI and the software that we were building. It wasn't as important for them and their technology whether the pathology tissue was scanned from a Philips scanner or a Leica scanner or Hamamatsu or some other scanner. What mattered was that if you got a good quality image, what could you do with it? What kind of value could you extract out of it? Because of my experience with HeartFlow, we were able to leverage a lot of the lessons from HeartFlow to do the trials, to build the products and ultimately receive the first FDA approval for an AI system in pathology. 

0:07:17 - David Williams
Great. Then I saw Briar Capital. I don't know if that was just like a way station as you were getting Jona established, or what the progression was from PAGE to there. 

0:07:27 - Leo Grady
Jim Briar is a well-known investor. He was known for his investment in Facebook early on when he was a sell. He's been involved with many, many great companies, big and small. At some point he went on his own and formed Briar Capital. He'd never done anything in healthcare, but PAGE was his first healthcare investment. He was really involved in bringing PAGE out of Memorial Sloan Kettering. 

As he became more and more excited about healthcare, particularly the opportunities for AI and healthcare, he and I started working together more and more closely. When I left PAGE I joined forces with Jim to really help him invest in AI companies and healthcare and then support the portfolio of AI companies that Jim was invested in. When I left PAGE, I knew I wanted to start a microbiome company, but I wasn't exactly sure what the entry point was or what the strategy was going to be or what the initial product was going to be. I knew the problems I wanted to solve. I knew where I wanted to go, but being able to work on the VC side for a while was really helpful to the survey the landscape, learn a lot and then have an opportunity to really work on the business plan for Jona while I was working with Jim. 

0:08:49 - David Williams
Great, let's talk about Jona. My first question has to be what is the microbiome? I have a general sense of it, but how do you characterize it? What does that actually mean? The microbiome? 

0:09:03 - Leo Grady
Microbiome is all of the organisms that live in your body and on your body. It's all of the bacteria, the fungus, the viruses, protists, archaea all of these organisms that live inside you and on you. We think about historically. We think about germs as bad and antibiotics as good. That's certainly true. I mean, everything from the black death to polio has a microbial origin, and we have to seriously deal with those pathogens. 

At a certain point, though, it became obvious that all microbes aren't bad and that many of them could actually be beneficial and, in fact, not essential, but critical to someone's health. Over the last few decades, we've seen the rise of this probiotics industry, with this idea that they're not only bad guys, but they're also good guys. You can take probiotics in pill form, you can take it as kombucha or yogurt or something like that, but there is this concept of good guys and bad guys. The reality is that you have a pretty complex ecosystem that lives inside your body, and it functions almost like another organ and can have really dramatic impact on your health. I think one of the lessons that we've learned, particularly over the last five or 10 years, is how tightly linked the microbiome is to your immune system. 

0:10:42 - David Williams
I don't know how literally to take your point about the microbiome being like an organ. It would suggest that, is that emerges, that you the same way have a cardiologist or hepatologist or nephrologist, you might have a microbiomist or something like that is. Is that where we're heading is medicine? I know that's not what Joan is doing, but is that, is it that profound? 

0:11:02 - Leo Grady
I think so. 

I've heard a lot of doctors say exactly that, particularly in the GI space. Many doctors believe that this microbiomologists that you're talking about will ultimately fork out of GI and become its own discipline. You know, I think thinking about the microbiome like an organ is a really interesting way to think about it, because if it's an organ, it's not like a heart or a kidney or the kind of organs that we're used to that you can do surgery on or that you can take out, that you can touch, you can feel, that you can take an MRI of. If the microbiome is an organ, it's really an organ of data, and therefore I think that's the only way that we're going to build an MRI of the microbiome, if you will is through really advanced computational tools, and that's spans the gamut between collecting a lot of data but also building the tools to be able to interrogate that data, the AI, to pull together the patterns in that data and then to be able to make that accessible to this microbiome doctor of the future, but also to the patient themselves. 

0:12:15 - David Williams
So I hear a lot about the microbiome or just sometimes just saying people say the gut and its relation to you know illnesses or ailments as you wouldn't necessarily expect. I had a pitch today from someone who's got a company focusing on the connection between the gut and Parkinson's and so like all these things coming across that are hard to get a handle on and they all are sort of seem like one-offs. And you mentioned probiotics and so on. So what is? I guess? I sort of get the idea of the promise of the microbiome, but what? What are the major challenges in realizing that that promise? You think you're getting at it partly when you're talking about the data, the massive amount of data and analysis needed. 

0:12:55 - Leo Grady
Well, you mentioned Parkinson's. So the microbiome has been related to Parkinson's, to basically every autoimmune condition, to GI disorders, to food allergies, but also to diabetes, to obesity, liver disease, to depression, to anxiety, to Alzheimer's, to autism, even to human longevity, and so how all of that happens is something that we're still learning more about, and there are certain questions of causality. Is the microbiome just reflecting all of these different disorders or is it causing it? It's probably a combination of both, depending on the different conditions. But a lot of the challenges have to do with the complexity. So the microbiome itself is large. You can have 2000 to 20,000 different organisms, different species, that live in your body. If you take all of the genetics of all of those species together, the number of genes is about a thousand times the number of genes in your human genome, and it's spatially distributed, so you don't have the same organisms living in the same places in your body, and it changes over time, and so that complexity is very difficult. 

And then there's also the complexity of the literature to understand what's happening to and what we're learning. So every single month there are more than 2000 papers being published in PubMed on the microbiome. It's impossible for any human being to keep up with all of those papers. So complexity is one of the big issues, yeah, causality is another one. Accessibility is another one. So we we can sample the gut microbiome through stool sampling, but it doesn't give us a great reflection of what's happening in the small intestine or in the stomach or in the trachea. Trying to sample the microbiome in the lungs, for example, is very difficult, and so there's sampling issues involved with the microbiome. And then then, lastly, I think trying to work out the different mechanisms by which these organisms are interacting with your body and all the different pathways is is also very complex so great fodder for, as you're saying, thousands of publications a month, because there's a lot of promise and a lot of challenges in doing that. 

0:15:25 - David Williams
How are some new technologies or you know other approaches helping to unlock at least some of the promise? I think I'm getting the sense from you. 

0:15:33 - Leo Grady
It's not all going to come at once well, what we've done at Jona to address this issue is we built a large language model to be able to read all of those papers and to synthesize them together so we can take a. 

We can offer and take a microbiome test result, a list of all the different organisms that live in your body, and the AI can summarize all of that literature that's being published and say for you, david, here is a summary of everything that your microbiome has been associated with all the different diseases, the conditions, the food allergies, the symptoms and, if you want to steer your microbiome away from those associations, here's a summary of all the different actions that you can take, everything that's known about ways in which you could change your microbiome through lifestyle, through diet, how medications interact with your microbiome, and really make all of that information accessible to somebody so that they can not only look for answers in their microbiome about different experiences they're having, but then also get some guidance on what they might be able to do about it to change their microbiome in a beneficial direction. 

0:16:53 - David Williams
So you know, the use of AI there sounds very promising because you've got all this information that people can't keep up with. You need to synthesize it, maybe ways that are unique for whatever that individuals test results are. How do you consider you know some of the challenges that we hear about AI itself, whether it's the bias or what model is. You know what it's trained on. And then just even the issue I don't know whether hallucination plays into the type of large language model that you have, but how do we use the AI in a way that's going to be most effective? 

0:17:29 - Leo Grady
So, if you look under the hood with our technology, there are really three pieces to it. So one is the AI system that is reading all of these papers and saying this is a paper that studied Crohn's disease. It looked at 2,000 different individuals. This was their age range, this was their gender, this was their whatever and compared to healthy people, they found that people with Crohn's disease are higher in E coli and lower in acrimancy or whatever it is. And so the one AI system is really looking through all these papers and pulling out these details of how big was the study, what was found, what did they study, who did they study and just pull all that information out. Then the second part of our technology actually puts all those together and says, okay, well, if we looked at 100 papers on Crohn's disease and they found these differences in the microbiome relative to healthy people, you know they find different. They find each one finds different things. They all find exactly the same thing, and so you need to be able to merge all of that data together to all those findings, to synthesize a way of saying, to operationalize it, to say you know your microbiome fits the profile and the literature well or not well for someone with Crohn's disease. And then there's the third part, which is taking all of those matches and then being able to present something to the user that they can read and understand and make sense of, whether that user is a doctor or as a patient. And so those three pieces have different parts to it. Right. 

So to your question on bias. All we can do is reflect the current literature. And if the current literature never studied certain groups of people, was over-weighted to other groups of people, the results that we get are going to reflect that literature. You know, the moment that you become a diagnostic test, the moment that you have to go prospectively validate a technology, then you're obligated to go sample the population appropriately. But if you're just reading the literature, you have to work with what you have today. And then, as far as hallucinations go, because everything that we are presenting to the reader, to the user, is based on the science and it's amalgamating together all of these different papers, it's not creating new things in the same way that a chat sheet would yeah, where it's, you know, making up case law or anything like that we really restrict that portion of the AI system to only work off of what's in the knowledge base and what's in the knowledge base, or all these? 

0:20:28 - David Williams
different papers Got it. There may be some interpolation, so some result that wasn't literally published in any particular paper might show up based on, or is that not the case? 

0:20:40 - Leo Grady
Well, it's a synthesis of all the different papers, so it wouldn't be. It wouldn't be creating results that aren't in any paper, but everything that that gets reflected back to the user has all the citations, so it's very transparent where all this information is coming from. Somebody can go read it, and we're also working on publishing a white paper that details how we put all these different papers together so that we're completely above board and completely transparent in the results that we're giving back. 

0:21:13 - David Williams
Now I saw that I think you were launching from on a beta basis at least, with the direct to consumer model. Can you say a little bit about you? Know what's your initial customer base, and is that the long term plan? 

0:21:25 - Leo Grady
So we are are offering it through two channels. One is through concierge medicine or functional medicine, integrated medicine, longevity clinics, aspects of medicine that do testing that goes beyond what is normally done in a regular primary care setting and often doing tests that are not reimbursed, and so that's one group. And then the other is we are offering a direct to consumer and to those people we are offering them access to a practitioner who can help answer their questions or talk through different results or help them operationalize what they're finding. I think a lot of the people who are coming to us are people who are either in one of three groups. They're either people that have had symptoms and are having some sort of health struggle. Maybe they've gone to a doctor, maybe they've had tests, but the tests come up negative and they're still looking for answers, and so they want to find out if those answers live in their microbiome, and so by scouring the literature and telling them what's known about what's been associated with their microbiome, maybe we can point them in the direction to ultimately get the right diagnosis. 

The second group of people are people who have a diagnosis but they're still wondering what to eat, how to change their lifestyle. They know that what they eat affects them. They don't have great guidance on a personalized basis of what can help, and so, again, they're looking to their microbiome to help them figure out what's going to work for them and what's not. And the third group are people that are not really suffering from any health issues but they're really interested in longevity, their athletes, they're interested in performance and they want to take their health to the next level. And we know that the microbiome is related to exercise. We know it's related to longevity, it's one of the hallmarks of aging, and so there's a lot that we can tell somebody about the literature associated with those areas as well. 

0:23:42 - David Williams
Now, when you think about you know what is somebody going to do differently. So in some cases it might just be helpful to know hey, this is a root cause, explanation of you know why I have this, that or the other. It helps them to confirm a suspicion. The types of interventions or behavioral changes that you're discussing mostly have had to do with diet, if I understand right. But what are the? What's the nature of the kind of recommendations that might be made? Are there things related to not just behaviors but environment or anything else like that that comes out of it? 

0:24:09 - Leo Grady
Well, we're reflecting the literature, so it's everything that's been studied, and I would say most of the studies are around diet and some around lifestyle. So there are some studies that look at exercise. Some say, is it like it's sleep and the impact on the microbiome. There are other kinds of studies. I mean, if you get a dog, it changes your microbiome in certain ways, yeah, and you know that may be beneficial or not beneficial to some people. So we're really reflecting everything that's known. I would say most of it is dietary oriented, because that's most what's been studied, but not everything. There are a lot of other kinds of studies and papers out there on things that can affect your microbiome from an environmental standpoint. 

0:25:04 - David Williams
Leo, how would you contrast what you're doing with something that people might be a little more familiar with, like 23andMe? Is it the same sort of thing? Is it complementary? Does it all kind of go together at one point, recognizing they don't have the same AI based interpretation of the literature? But should I think about it the same way? 

0:25:23 - Leo Grady
It's got some similarities. So, as you know, 23andme looks at your genetics and they also will tell you. For example, when I did 23andMe, they told me that my genetics were related to atrial fibrillation, that I was a high risk for AFib, they told me that I was a high risk for prostate cancer and they pointed me to the literature that had studied my genes and linked them with these different conditions. So in that sense it's similar. However, I took that test 10 years ago. 

I haven't taken it again because my genes haven't changed and maybe in some future I can crisper up my genes, but not yet, and so for me there's not a lot of value in going back and there's not a lot of action that I can take to lower my risk for atrial fibrillation or for prostate cancer beyond generic recommendations. There the microbiome is different in the sense that it does change and that it is changeable through different actions. So it is similar to 23andMe in the same sense that we connect your body to the literature and tell you what has been linked and what's been associated with these different aspects of your biology, but we're able to go a second step and tell you what actions you can take to actually move your, or what's known about how you can move your biology in a direction that's going to be more favorable to you, and then you can monitor progress of that over time. 

0:27:07 - David Williams
You've been involved in the past in some companies that have required a fair amount of capital to get going and I wonder is this a capital intensive business? What do you do in this funding environment? How do you think about how that will roll out, given your experience and what you see for this company specifically? 

0:27:25 - Leo Grady
We raised initial capital from prior capital and we had some other investors come in as well and, however, not at the same level that I did at previous companies. So at Page, for example, I raised almost $200 million over the course of my time there as CEO, and a lot of that money was put into data collection and creation. So because of our deal with Memorial Sloan Kettering, we had access to pathology slides, but they were still in storage. So we had the agreement with Memorial Sloan Kettering to be able to scan them, but we still had to go do it, and that was a very expensive and time consuming process to pull slides out, clean them, scan the barcode, scan the slide, put it back and then to do that for all of the patients that have come through Memorial Sloan Kettering over the last decade. 

So with Jona, I've taken a different strategy and a different approach and started with this product. That is something that people are already buying microbiome tests. There's already an interest among consumers and among practitioners in studying the microbiome, but often these tests are really just geared towards selling probiotics and not really providing you any information, and so I felt that there was a real opportunity, especially now with the maturity of large language models that we've seen with ChatGPT and Lama and over the last year, to be able to access that information and to make it more accessible. And this was something that we were able to build fairly quickly and get on the market very quickly. And because people are already paying cash for these tests, we felt that there was an opportunity to give them a significant upgrade from what's out there today. But then, as we move forward over time and we start wanting to go beyond the current literature and build clinical grade products, new diagnostics, new therapeutics, we'll have the opportunity from a data and a revenue standpoint to be able to go through that. 

0:29:44 - David Williams
I'm just going from a pattern of two, but I have one other example of an entrepreneur that did something similar in concept what you're doing there. First company, which was VirtualScopics, required people to go get an MRI that might not otherwise be doing it Pharma companies usually paying for it. They'd analyze that data. And then the next company, iCardiac, took ECGs which were already being collected and already out there and were cheaper to do and then worked from there. So they were working on the analysis. It was a lot less capital intensive. It was easier to scale the company as well. So maybe there's something similar here and maybe that's a lesson for aspiring entrepreneurs as well. 

0:30:22 - Leo Grady
I think it's a really great analogy. Selling into an existing market with a brand new technology and a new value proposition is something that can be easier than creating a new market, and I very much believe that, as this microbiome doctor of the future comes into reality, we are going to be the company that's powering that doctor and giving them the technology and the tools where they can analyze the microbiome. But that's not something that you do overnight and that's not something that you can do without a lot of technology investment and data investment, and so, rather than trying to go straight there, we're starting in an existing market and then moving up into really, really, really redefining this new area of medicine. 

0:31:19 - David Williams
Leo, my last question for you is going to be turning away from business a bit, and I want to ask you if there's any good books that you've read lately, anything that you would recommend to our audience. 

0:31:34 - Leo Grady
That's an interesting question. You know, I recently did read a book that struck me it's a little unusual, though it was a book written by a magician on how to basically story tell in the course of, you know, presenting to an audience, and they they really talked a lot about different mechanisms and techniques that you can do to tell a story effectively, and that was something that you don't really hear very many people discuss, and so you know a lot of if you were to go read magic books. It's mostly about how to do card tricks and things like that. 

Yeah, this was really more about how to present to an audience in a way that gets them excited and gets them engaged and involved with the presentation. So that was something I read that was that was unique. Recently I'm also reading Powerful, the story on building culture written by Patty McCord from Netflix, in conjunction with Reed Hastings, and so I think this is an area that I'm thinking about at this stage of Jona how we build a really high performing culture and how we are able to, you know, build a company that lasts. 

0:33:07 - David Williams
Yeah, I like the idea of the magician. I heard a Moth podcast recently and I was talking about a magician and some of his. You know some of his early challenges and things that he had and how and his parents weren't supportive eventually became very supportive, I think. You know, people are always learning these tricks and things and somebody really masters the microbiome. There'll be a lot of magic that can be done from that as well. So that'll be a topic for our next podcast. 

In any case, Leo Grady, founder and CEO of Jona, thank you for joining me today as a guest on the Health Biz podcast. Thank you so much, david. It's wonderful to be here. You've been listening to the Health Biz podcast with me, david Williams, president of Health Business Group. I conduct in-depth interviews with leaders in healthcare, business and policy. If you like what you hear, go ahead and subscribe on your favorite service. While you're at it, go ahead and subscribe on your second and third favorite services as well. There's more good stuff to come and you won't want to miss an episode. If your organization is seeking strategy consulting services in healthcare, check out our website, healthbusinessgroupcom.