Latest / HumanifAI HealthTech / How AI is Unlocking the Human Microbiome w/ Dr. Jenny Yang, Co-Founder & CEO, Outpost Bio
Transcript
- 0:00Brian: We share 99.9% of our DNA with every other human, but the trillions of microbes living in our gut tell a much more personal story.
- 0:12Brian: In fact, no two microbiomes are alike, and those differences shape how we respond to food, drugs, and disease.
- 0:21Brian: Today's guest wants to turn that hidden biology into personalized medicine built around the patient, not the population.
- 0:36Brian: Welcome back to Humanify Health Tech, where we explore how AI is rewriting the rules of healthcare and putting people back at the center of where they belong.
- 0:46Brian: I'm Brian Chi, Senior Director of Content and Communications at Net Health, or your host for short.
- 0:53Brian: Today I'm joined by Dr. Jenny Yang, co-founder and CEO of Outpost Bio, a company combining lab-grown human microbiomes with AI to make medicine as unique as the people it's meant to treat.
- 1:06Brian: Before she took on the microbiome, Jenny was using AI to hunt for cancer drugs.
- 1:12Brian: She earned her doctorate at Oxford as a Marie Curie fellow, one of Europe's most competitive research fellowships, and built her career at the rare intersection of real bench science and machine learning before co-founding I- Outpost Bio in 2025.
- 1:29Brian: She's that rare expert who's equally at home growing microbes as she is building AI models.
- 1:35Brian: And today, she's here to help us make sense of both.
- 1:39Brian: Good luck with that, Jenny.
- 1:41Brian: But thanks for joining us.
- 1:43Dr. Jenny Yang: Thanks so much for having me.
- 1:46Brian: Awesome.
- 1:47Brian: Okay.
- 1:47Brian: So first question I wanna jump into is, um, it's gonna be a little basic, right?
- 1:53Brian: But I think it's good to sort of set there, s- begin the conversation there.
- 1:58Brian: We're about to spend this entire conversation on the microbiome.
- 2:02Brian: So before we do that, let's just say that most people have never had to think about it as a health factor at all.
- 2:13Brian: So first question, what is a microbiome?
- 2:17Brian: And in your own estimate, how much of a person's health actually traces back to their microbiome versus their genetics?
- 2:27Dr. Jenny Yang: That's a great question.
- 2:28Dr. Jenny Yang: So I guess first with the definition of a microbiome, it really is the community of, like you mentioned, trillions of microorganisms that inhabit a specific environment.
- 2:38Dr. Jenny Yang: And I guess to contextualize this conversation, we can focus on the gut.
- 2:42Dr. Jenny Yang: Um, so this would include, um, all the bacteria, viruses, or fungi that live in your gut.
- 2:49Dr. Jenny Yang: But there are other microbiomes, so there's a microbiome on your skin, in your mouth, et cetera.
- 2:54Dr. Jenny Yang: Um, but it really is that community of organisms that inhabit a certain environment.
- 2:59Dr. Jenny Yang: Um, and I would say a lot of people actually have thought about the microbiome, but it might have been phrased differently.
- 3:05Dr. Jenny Yang: Um, you probably hear the term gut health a lot, um, and that's probably much more tractable, but I'm sure a lot of people think about, you know, the foods they eat and how that affects their health, and that is directly related to the microbiome.
- 3:17Brian: Awesome.
- 3:18Brian: So one of the things I noticed in doing some research for, for this podcast, I noticed that you have a lot of passion about this topic, a lot of enthusiasm for your mission at Outpost.
- 3:28Brian: And one of the things that stood out to me was your work with Professor Stephen Jones at the Genome Sciences Centre.
- 3:34Brian: That, that seemed to be a key point in the journey that you're on with Outpost.
- 3:39Brian: What was it about that experience that led you to where, where you are today?
- 3:44Dr. Jenny Yang: Yeah.
- 3:45Dr. Jenny Yang: I, I think back very fondly and, and owe a lot to, um, my, my career journey because of Steve.
- 3:52Dr. Jenny Yang: So I started my undergrad as an engineering student, um, but wanted to have some experience in, in research, specifically within healthcare, and Steve gave me an opportunity to do an internship, um, in his lab that, uh, one, one summer.
- 4:06Dr. Jenny Yang: Um, and he really introduced me to the whole concept of both AI within the healthcare space and personalized medicine.
- 4:14Dr. Jenny Yang: So I got to gain experience working alongside both machine learning practitioners, bioinformaticians, and clinicians on building tools that would help, um, push forward the field of personalized medicine.
- 4:27Dr. Jenny Yang: Specifically in his lab, we were focused on oncology, and we really wanted to have a better understanding of how we could both classify cancer subtypes, but also do better patient drug matching so patients are matched with the drug that would work best for them.
- 4:42Dr. Jenny Yang: And specifically, we were focused on human genomics.
- 4:45Dr. Jenny Yang: Um, and I also joined the lab during a really exciting time.
- 4:48Dr. Jenny Yang: So this was probably over 10 years ago now, when we were really seeing the onset of a lot of, uh, big data, um, specifically in the genomics and transcriptomics space.
- 5:00Dr. Jenny Yang: Um, so it was very exciting to see the growth of all of that.
- 5:02Dr. Jenny Yang: And I would say that really got me fully passionate about pushing forward personalized medicine.
- 5:10Dr. Jenny Yang: I feel like I've been trying to pursue that at every step of my career since then.
- 5:16Brian: Yeah, it's, it's, it's exciting.
- 5:18Brian: To me as a, as just a, as a, as a, as a mere mortal, as a human being, that what you're doing is incredibly exciting.
- 5:25Brian: Because we all know that feeling of not being, not being, not feeling like a person when you go to the doctor or, um, when, when, when there's struggle in finding what, um, finding answers to your symptoms and, and, and having that experience.
- 5:40Brian: So, and that's part of it, right?
- 5:42Brian: That's part of what's so exciting here, is that patients experience illness as individuals, yet medicine has always treated patients as a population, you know?
- 5:52Brian: And, and, and the greater pool, rather than take that more personalized approach that you're, um, that you're, um, that you're putting together.
- 6:01Brian: Why has there been such a disconnect between how we treat patients and what medicine predicts?
- 6:08Brian: And, and follow-up question or two-part question to that, how can microbiomes close that gap, or more broadly, what can it actually solve?
- 6:19Dr. Jenny Yang: Yeah, that's such a good question, and I think it's such a big question as well.
- 6:22Dr. Jenny Yang: Um, I think to start, human biology is just so complex, and when you take into consideration the individual rather than the population, you're adding a lot more complexity.
- 6:32Dr. Jenny Yang: Uh, and to be fair, I think there are a lot of incredible patterns to be learned.
- 6:35Dr. Jenny Yang: Like, I think you see one case, and you can take that anecdote and all the lessons from that and apply it to a new case, and I think that's why we've really taken this pattern recognition approach to look at health on a, a population scale.
- 6:49Dr. Jenny Yang: Um, and I think there is a lot of, a lot of truth in that, and we've seen it especially with, um, even, like some of the, the generics, uh, like generic therapeutics that have come out.
- 7:00Dr. Jenny Yang: Like it, it does work on this population scale, and that's because we can find certain pathways on this causal level that relate to all of us, um, as humans.
- 7:09Dr. Jenny Yang: Um, and then technology is also growing at the same time as our knowledge of just biology.
- 7:15Dr. Jenny Yang: So when I got experience in the genomics, the, specifically the human genomic space, I was able to see how we can identify exact genetic mutations that lead to particular subtypes of cancer, and then people are now developing therapies that target that particular mutation, and that's super effective for better understanding how we can start refining the resolution that we, we try to, um, diagnose people with or, or cure people with.
- 7:43Dr. Jenny Yang: Um, but I would say it is relatively new.
- 7:47Dr. Jenny Yang: Like, I think it's been a journey that people have been on in the last 20, 30, even, even 40 years.
- 7:52Dr. Jenny Yang: But as our knowledge of the, this complex space is growing, technology is also growing, so I think we are limited by that pace of development as well.
- 8:02Dr. Jenny Yang: Um, but it's a hugely complex task.
- 8:05Dr. Jenny Yang: Um, people always talk about how the human is a system that is impacted by many different things.
- 8:11Dr. Jenny Yang: So we're impacted by our human genomics, we're impacted by our environment, our eating habits.
- 8:16Dr. Jenny Yang: Um, so there's just so many things to consider that it makes it really complex to look at everybody on a personalized level.
- 8:24Dr. Jenny Yang: Um, I do feel like the microbiome space is also one of those growing spaces where now we might have the right tooling to be able to add it in as an analysis that brings us more to that personalized level.
- 8:38Dr. Jenny Yang: So I-- Yeah, because I've spent a lot of time and seen the growth in the genomic space and seen the impact that's had, I think it's time to add all the non-human DNA.
- 8:48Dr. Jenny Yang: 'Cause like you mentioned in your intro, when you consider, you know, the entire audience and, and the world, we're 99.9% identical in our human DNA.
- 8:58Dr. Jenny Yang: Um, but when we look at our gut microbiomes, you know, we can be up to 90% different.
- 9:03Dr. Jenny Yang: There's- A lot more genetic diversity in there that when we want to consider what makes us really individual, we are gonna have to add in this piece of the puzzle.
- 9:13Dr. Jenny Yang: Uh, but it is just a piece of the whole human biology and whole human system.
- 9:19Dr. Jenny Yang: Um, so it is a largely complex task, but I think it will add a lot of extra useful information to bring us one step closer to personalized medicine.
- 9:27Brian: So it sounds like to me, and, and please e- educate me here, right?
- 9:31Brian: It sound-- But it sounds like to me like it's the, it's, it's a journey on understanding more personalized patterns and continually working to m- create more personalization within the patterns that we have already, we've already been basing our medicine on a- all along.
- 9:51Brian: It's just that from 50 years ago, the patterns we were looking for were broader, and now, thanks to your work and the work of a lot of people that are, uh, that are, that are dedicated to this, it's about finding more personalized, more unique patterns to be able to, um, find, find cures, find, find solutions.
- 10:12Brian: Is that-- Do I have that right?
- 10:13Dr. Jenny Yang: Yeah, absolutely.
- 10:14Dr. Jenny Yang: And like adding to the patterns that we already have and like bringing them to a more, um, higher resolution so we really get to a more personalized space.
- 10:24Brian: Wow.
- 10:25Brian: Okay.
- 10:25Brian: So using human knowledge and technology to put that all together and to, to a- and actually our knowledge in the lab is, is greater than the ability of technology to detect, find, and build those models.
- 10:44Brian: And, and if anything is slowing it down, it's the technology.
- 10:47Brian: And, and what we see today with AI is a, is a quantum leap or a huge step toward being able to do that.
- 10:54Brian: When, when-- I'm leading up to a question, believe it or not, I, I really am.
- 10:58Brian: When did you put that together as far as the ability and the, and the inspiration behind that, that concept of lab and, um, and machine learning?
- 11:10Dr. Jenny Yang: Yeah.
- 11:10Dr. Jenny Yang: I think there's two parts to this question.
- 11:12Dr. Jenny Yang: So the first part is really the, the scale of data that was being generated in genomics.
- 11:17Dr. Jenny Yang: Um, it was during a time of incredible growth where we started with like the 100,000 Genome Project, and now it's probably up to millions of, of genomes that have been sequenced, sequenced, definitely more than that.
- 11:28Dr. Jenny Yang: Um, but by seeing that and seeing algorithms that could be built on it to basically crunch through a lot of data and find patterns that we wouldn't have been able to do just, um, in our heads, that was, uh, just really incredible for me to experience.
- 11:44Dr. Jenny Yang: Um, and I was also embedded in a team of clinicians.
- 11:48Dr. Jenny Yang: So this is, this is the second part.
- 11:49Dr. Jenny Yang: So at the Genome Sciences Center, we worked directly with clinicians.
- 11:53Dr. Jenny Yang: As soon as someone created a new algorithm or a new tool, it was in the hands of clinicians.
- 11:58Dr. Jenny Yang: They were giving us feedback, so we really were building AI that was made to be translated.
- 12:03Dr. Jenny Yang: So seeing that effect, um, I think really h- it had a lasting impact on me.
- 12:10Dr. Jenny Yang: So even when I went to do my, my PhD later, I knew I wanted to focus on translational AI and work with the clinicians that would be directly working with the patients, um, to implement th- these AI technologies,'cause it really is, like you said, that, um, interaction of knowledge and technology, um, and then working with the system in, in the way that can actually bring it in most smoothly.
- 12:35Brian: Uh, let's dig a little bit into the technical side.
- 12:37Brian: So if we agree the medicine needs to become more personal, that that's the thesis, that's the, that's the, that's the statement, right?
- 12:45Brian: There's technical challenge, as you've identified, um, to do that, right?
- 12:50Brian: The microbiome isn't one thing, it's a tr- it's trillions of things interacting in ways that are constantly changing, evolving, uh, all sorts.
- 12:59Brian: I can imagine that's far too complex for myself, for mere mortals to-- or even traditional statistics to try to untangle that.
- 13:08Brian: So, so what can AI do today that wasn't simply possible five or 10 years ago?
- 13:17Dr. Jenny Yang: I think we have much more sophisticated algorithms.
- 13:19Dr. Jenny Yang: I think we've seen incredible work being done in the actual, like, theoretical, um, machine learning space and, and mathematical space that helps us find patterns in much more effective ways.
- 13:29Dr. Jenny Yang: Like before, you know, we started with linear models, and now we have a lot of nonlinear models and a lot of other types of models that make computing these datasets much more efficient.
- 13:38Dr. Jenny Yang: So that's one thing.
- 13:39Dr. Jenny Yang: The other thing is cloud compute infrastructure is, is much stronger now.
- 13:43Dr. Jenny Yang: So again, we can take the, the load of data that's actually coming in to crunch through these patterns.
- 13:50Dr. Jenny Yang: So I think there's been both a increase in capabilities on that test technical aspect, and another one is I think we're recognizing that, um, we really need high-quality data.
- 14:02Dr. Jenny Yang: So the techniques to generate data that's high quality and machine learning usable, these techniques are also improving.
- 14:10Dr. Jenny Yang: So if we can have high-quality data to train these models that are now more effective and more powerful, uh, I think we're going in the, the right direction to take on much more challenging problems.
- 14:22Brian: Okay.
- 14:22Brian: So le- l- let's talk about your personal inspiration a little bit more because I, I'm, I, I think, like, I, I listen to you talk and I hear dataset, I hear, um, algorithms, I hear, I hear things that are not necessarily personalized.
- 14:38Brian: You're applying this non-personal, non-in- individualistic approach to create medicine that is inherently personal.
- 14:49Brian: Is there-- When, when you began doing this, was there a specific person or a patient, somebody that sort of, sort of really drove you here?
- 14:58Brian: Someone that you, that you, like, still picture when you're deep in your technical work?
- 15:03Brian: Not to be sort of Pollyanna there, that's kinda corny, but, but you know what I mean, somebody that really drove your meaning here.
- 15:10Dr. Jenny Yang: So our family has been impacted by cancer, as has many families.
- 15:15Dr. Jenny Yang: Um, but that actually happened later on in my career, and I w- I would say There wasn't one person.
- 15:25Dr. Jenny Yang: I think the problem itself is, is so important, and I think my inherent passion comes from both the, the concept of having personalized medicine and just the, the innovation and kind of technical, uh, more scientific and technical aspects of it.
- 15:43Dr. Jenny Yang: I think I'm, I'm fortunate enough to say that, you know, at that time when I was trying to build my career, I… Our family was lucky to have not been impacted by that, and I can't… Other people wouldn't be able to say the same, but I would say the mission is so important, and you can see the impact it has on, on healthcare and families that I was immediately drawn to it.
- 16:07Dr. Jenny Yang: Um, so yeah, I, I think on a bigger scale, the-- when you see the numbers of how many people are affected by this and when you hear the statistics, like one famous one that everyone hears is 90% of clinical trials fail.
- 16:21Dr. Jenny Yang: Um, but 90% of clinical trials fail today as they did 30 years ago, so why are we not moving the needle on that number?
- 16:28Dr. Jenny Yang: I think that just drives me to want to find new ways and new perspectives to bring into the field to hopefully move that number.
- 16:36Dr. Jenny Yang: Because that's a lot of resources going into such an important mission that however we can contribute to making that stat better, I, I want to be part of that journey
- 16:50Brian: Okay, let's shift a little bit.
- 16:51Brian: I, I wanna talk about, um, open source a little bit.
- 16:54Brian: So there's a growing divide, right?
- 16:56Brian: Um, leaders like Dario Amadei argue that increasingly capable open models create unacceptable risks, um, and that, you know, he's saying basically the future may need to be closed.
- 17:09Brian: Um, at the same time, science kinda has always advanced through openness, reproducibility, independent validation.
- 17:17Brian: Um, where are you at with that?
- 17:19Brian: You're building open models, right?
- 17:20Brian: Have-- What, what's the f- what's your thought process on this?
- 17:23Brian: Why do you think that the-- that's the right choice for biology, and do you worry about that at all?
- 17:29Dr. Jenny Yang: Yeah.
- 17:30Dr. Jenny Yang: So I, I'm very much for open source, especially in, um, when it comes to under- better understanding biology.
- 17:39Dr. Jenny Yang: Uh, one reason I'm pro open source, especially in the microbiome field, is because it's a growing field of knowledge, um, and there's a lot of discovery still to be made.
- 17:50Dr. Jenny Yang: And I think it would be really inappropriate of us to say what we do in the lab hidden behind closed doors and then tell you as fact is just true.
- 18:01Dr. Jenny Yang: Especially in a, a growing field, we do need that independent validation and external validation and peer review.
- 18:08Dr. Jenny Yang: Um, especially for such an important field that has implications on health and medicine, um, we need to make sure what we're doing is sound, um, and proven before we, you know, put it out in the public and have other people actually being impacted by these findings.
- 18:25Dr. Jenny Yang: Um, the other thing that's really, really good about open source specifically in the machine learning world is I do believe in pattern recognition and the transfer of knowledge.
- 18:35Dr. Jenny Yang: So if we build really effective models, let's say like right now we're focused on the gut microbiome, but we have a study coming out soon that's gonna demonstrate that findings from the gut microbiome can be transferred to other types of microbiomes.
- 18:48Dr. Jenny Yang: So we don't want people in every single microbiome field to have to start from scratch, especially if we've moved the needle forward in a generalizable way, um, because that- Um, sharing that knowledge with others isn't going to affect our business model or business line.
- 19:05Dr. Jenny Yang: Like, people can work on their own tasks in their own spaces, um, but learn from some of the patterns that we're seeing in our particular niche.
- 19:13Dr. Jenny Yang: Um, so I, I, that… I think people also move faster and learn faster when they're able to collaborate with others, um, share learnings, share findings.
- 19:22Dr. Jenny Yang: Um, another reason I am very pro open source is I've been in the machine learning field for so long that I know I've always used open source models.
- 19:31Dr. Jenny Yang: I've used datasets that other people have posted publicly, and I've read other people's papers to improve the projects that I worked on.
- 19:38Dr. Jenny Yang: Um, and I, I really don't wanna pull that up from behind me.
- 19:42Dr. Jenny Yang: I, I know I got here not alone.
- 19:45Dr. Jenny Yang: Um, so the least I can do is, is give back in that way.
- 19:49Brian: We're never alone.
- 19:49Brian: Yeah.
- 19:49Brian: We always get to where we're at, right, as a, as a, as a, as a collective group, as, as a, as a, as a core team of people helping one another.
- 19:58Brian: But do you worry?
- 19:59Brian: I mean, you're, that's Anthropic talking about that, right?
- 20:02Brian: Do you worry that AI is gonna become too closed and that all the progress you're making is gonna slow down?
- 20:08Dr. Jenny Yang: I think there's certain types of AI where, and in certain fields where I understand why people need to close it.
- 20:15Dr. Jenny Yang: Um, I think I, I've just been very lucky to be in such a, a positive field when it comes to healthcare and pushing forwards, like, uh, biology and im- improving health that I feel like I've always been around people with very positive views on AI.
- 20:31Dr. Jenny Yang: Um, I think there's a lot of good things that are open source, and I think maybe it will be application dependent.
- 20:39Dr. Jenny Yang: Like, for example, like models that are being built in defense, I don't know if I want everyone to have access to those and, but that seems like such a specific case that, um, w- I can, I can understand that.
- 20:51Dr. Jenny Yang: I actually feel like when it comes to more like theoretical AI and the pr- development of new models, I do feel like that is actually going to remain quite open on a, on a general scale, and it's once you really dig down to the specific applications that we'll start to see a divide in what's opened and, and what's closed.
- 21:12Dr. Jenny Yang: And I, I don't know if I'm the right person to, to give a- an opinion on every field there, but as long as I feel like we're not providing We can only control so much what other people do with these tools.
- 21:25Dr. Jenny Yang: Um, and I, I guess I've always tried to, um, err on the side of being more, more positive.
- 21:33Dr. Jenny Yang: Um, but it is something we actively think about, so we will make sure that, like, uh, if we're using certain types of data, we get the correct ethics approval, we anonymize it in the correct way.
- 21:42Dr. Jenny Yang: So we'll take steps to make sure, um, we can do things open in as, in as safe way as, as we have control over.
- 21:50Brian: Yeah, because I mean, such a powerful, such a powerful technology, such powerful ability via the technology creates, um, creates, um, a requirement to have your own red line determined and defined based on what you're doing and the application in which it, in which it's being used.
- 22:10Brian: Is that, is that right?
- 22:12Dr. Jenny Yang: Yeah, absolutely.
- 22:13Dr. Jenny Yang: Like, um, we, we definitely think about this a lot, especially when we start taking in data that's linked to patients, and, um, that's when we really need to decide, like, like that, that stuff we'll keep in-house, um, 'cause that's also just a safe data practice.
- 22:31Brian: Do you ever worry that, that, you know, this, this conversation, this, th- this sort of theoretical sort of slightly real conversation is, uh, you know, that, that it's going to impact your own model in terms of flattening the individual signal patterns that you're trying to develop?
- 22:48Brian: You know, I mean, it's a cleaner, easier answer, right?
- 22:50Brian: Um, probably.
- 22:53Brian: What's the warning sign that you watch for there?
- 22:54Brian: What's-- When does it impact the work that you're doing, um, in terms of finding those individual patterns and that personalized patterns?
- 23:03Dr. Jenny Yang: Um, I would say the models we build are meant to learn from individual bits of data, but then they're transformed in ways that normalize them in a more generalizable way.
- 23:15Dr. Jenny Yang: So, um, we are taking steps to make sure that you can't backtrack the weights to identify individual components.
- 23:22Dr. Jenny Yang: Um, as the model is used, I think from the user perspective, you can put in an individual data point to get, uh, an outcome that is learning from patterns that the model has seen.
- 23:35Dr. Jenny Yang: So I think in that sense, we can keep it individual because we're making judgments on individual data points and learning from individual data points, but the model itself inherently is the learned patterns of a collective.
- 23:48Dr. Jenny Yang: Um, and then there's certain machine learning steps and actual computational, uh, kind of safeguards we can put in place to make sure it's not identifiable in a, in a backwards fashion.
- 24:00Brian: Okay.
- 24:01Brian: All right.
- 24:01Brian: Let's, let's talk about perception.
- 24:04Brian: Um, microbiome, trendy, hot, right?
- 24:09Brian: Lots of hype around microbiome.
- 24:11Brian: Not compared to AI though, because AI is all that and probably more.
- 24:17Brian: And, and you're living in a world of the most oversold words in health tech at right now, right?
- 24:23Brian: I mean, you got both of them.
- 24:25Brian: So all you need to do is add like transformation in there, and you're gonna be the trendiest person in the entire world.
- 24:32Brian: Um, how do you keep AI from becoming the thing that dresses up weak microbiome science?
- 24:37Brian: Because really it's the science, right?
- 24:39Brian: All, all of the stuff that you've been saying, all of this really insightful thing, information is all about the science and AI leveraging AI, leveraging machine learning to be able to create even more so and even better.
- 24:52Brian: How do you separate that?
- 24:53Brian: How do you keep that from becoming the thing that hides weak science?
- 24:58Dr. Jenny Yang: Yeah.
- 24:59Dr. Jenny Yang: I, I think this is a fantastic question 'cause I do feel like AI is the hottest topic right now.
- 25:06Dr. Jenny Yang: And, um, us as a company, like in, in every meeting we, we're, we're in, we're making sure we don't overclaim, and we're talking about validation and reproducibility and how we can validate our findings.
- 25:20Dr. Jenny Yang: Um, on the not overclaiming about AI side, we're very much of the opinion that if the best model is linear regression or a decision tree, we're not going to overcomplicate it with a fancy model just to say we've used a fancy model.
- 25:38Dr. Jenny Yang: Like I-- we do not believe in reinventing the wheel.
- 25:40Dr. Jenny Yang: That's not productive.
- 25:42Dr. Jenny Yang: And even with our first, uh, launch of a open source foundation model, uh, Waypoint, we were very clear that this is a starting point.
- 25:49Dr. Jenny Yang: We didn't say it was solving the microbiome.
- 25:51Dr. Jenny Yang: We didn't say it's gonna solve every single problem.
- 25:54Dr. Jenny Yang: And we told people how to use good practice to leverage these models.
- 25:58Dr. Jenny Yang: We said, "Fine-tune it on your own data," because especially with healthcare fields, you, you really need to have a fit for purpose model.
- 26:05Dr. Jenny Yang: So we're trying to be very, very, um Just open and fair about how to use these models and not overclaim what they can do.
- 26:17Dr. Jenny Yang: And when it comes to the actual science part, um, we always said Outpost Bio functions as a lab in the loop.
- 26:24Dr. Jenny Yang: We have the wet lab and we have the dry lab.
- 26:26Dr. Jenny Yang: The wet lab can help generate high quality data, but it can also validate findings that we're getting, uh, from our in silico approaches, and that is something we really care about.
- 26:35Dr. Jenny Yang: We wanna make sure that if we say an AI model has a certain output, we can prove it again in the lab.
- 26:41Dr. Jenny Yang: Um, we also think about reproducibility and external validation, um, outside of our, our own lab.
- 26:48Dr. Jenny Yang: So we have an academic institute as well as an external CRO that are generating buckets of data for us so we can validate the findings in our own wet lab, as well as demonstrate machine learning generalizability externally.
- 27:02Dr. Jenny Yang: Um, so we are trying to take steps, um, there just to make sure we, you know, we can do what we say we are doing.
- 27:12Brian: All right, let's wind back, um, and, and talk a little bit about, um, the patient that's just different.
- 27:20Brian: You know, we kind of started the conversation talking about that experience, talking about what you're doing to help solve that or to open up new ga- new d- new doorways, right?
- 27:30Brian: Um, so let's, let's, let's dive back into that for a couple questions and then wrap up.
- 27:35Brian: Here's a scenario.
- 27:37Brian: A clinician suspects that their patient is just different.
- 27:40Brian: It-- There's no way to prove it.
- 27:42Brian: The clinician can't put their finger on w- exactly, exactly what it is.
- 27:46Brian: The data is just not available, and it speaks in averages.
- 27:49Brian: So they're, they're trying to find a solution that, th- that they don't have the data for, right?
- 27:54Brian: Um, personally, and as the CEO and co-founder of Outpost, what does that feel like to hand a clinician data that actually backs up what their instinct was already telling them?
- 28:08Dr. Jenny Yang: Yeah.
- 28:08Dr. Jenny Yang: It, it, it feels amazing, and, and that's where, like, the, the translation part, uh, comes in.
- 28:15Dr. Jenny Yang: Um- You can do so much cool science and, and build really fancy models that are showing very great accuracy or, or patterns, but the real impact is when you actually see it having a positive effect in the real world.
- 28:31Dr. Jenny Yang: Um, so it really is the, the best feeling, and especially in the microbiome field, it's so complex, but we've seen so many wins that have been published and talked about, whether through trials or, um, in the literature where people can actually identify specific biomarkers in the gut that were responsible for certain side effects or were responsible for, um, responding better or worse to certain therapeutics.
- 28:57Dr. Jenny Yang: Um, and that we really wanna get there because that is the feeling that we're, we're going towards.
- 29:02Dr. Jenny Yang: Like, if a clinician could just, you know, take a stool sample and then find a marker, that would inform a lot.
- 29:08Dr. Jenny Yang: That's in- that's incredible.
- 29:09Dr. Jenny Yang: That's a huge achievement.
- 29:12Brian: I mean, that reinvigorates their, their reason for being, right?
- 29:15Brian: That r- the, for a clinici- for a clinician, let alone the patient, the clinician, that reinvigorates, that, that humanizing quality reinvigorates, goes beyond the top of their license, goes beyond what they're doing in terms of getting that enjoyment and the satisfaction of making a difference in someone's life.
- 29:33Brian: That's huge.
- 29:35Brian: What about the person, though?
- 29:36Brian: What about the patient?
- 29:37Brian: I mean, that, the, that person who's never been able to get a, a, a specific, a specific, um, answer to all of their allergic reactions that, that, and they've struggled for years and years and just, they've just been put on standard treatments.
- 29:53Brian: You know, 10 years from now, what's actually different for that p- that patient, right?
- 29:58Brian: The work that you're doing now, you know, if you fast-forward 10 years, what does that look like for that one sp- that one specific patient who gets an actual answer that makes a difference instead of that population guess?
- 30:11Dr. Jenny Yang: Yeah.
- 30:12Dr. Jenny Yang: It, it's huge, and I think everyone wants to feel like if they go to see a, a clinician, that clinician can say, "Based on your biology, this, uh, medication or this treatment is the best fit for you." That, I think n- that gives you a sense that there's a really diligent and responsible kind of process that someone has used to give you the best recommendation.
- 30:35Dr. Jenny Yang: Um, especially when you're w- if, if you're suffering from something, um, you don't wanna feel like you have to try one drug, realize it doesn't work, and then change to another, and it's this trial and error process.
- 30:48Dr. Jenny Yang: That's such a long process.
- 30:49Dr. Jenny Yang: It's wasting time that really matters, um, and mentally, it's, it's so draining.
- 30:55Dr. Jenny Yang: So I think having the feeling that something is right for you for specific reasons and that can be explained to you, it makes the patient process so much better.
- 31:06Brian: Okay, last question.
- 31:08Brian: Um, I'm gonna make you choose M- microbiomes, wet lab, right?
- 31:16Brian: The work you do in the lab, or your experience and expertise in machine learning.
- 31:21Brian: Which one drives the most passion for you?
- 31:23Brian: Which one gets you up in the morning and, and, and, and drives your greatest interest?
- 31:30Dr. Jenny Yang: Oh, that is such a hard question, and it is so hard to choose.
- 31:35Dr. Jenny Yang: I will say my, my dominating expertise is the AI and the translation of the AI.
- 31:43Dr. Jenny Yang: I'm very much passionate about both, but I don't claim to be an expert, um, in, in both.
- 31:49Dr. Jenny Yang: So, you know, even the way we've built up the team, we've made sure to, to bring in the experts for each, um, for each domain.
- 31:58Dr. Jenny Yang: But my background has been in AI, and I, I genuinely am so passionate about seeing such an incredible new technology actually being translated in a way that's extremely effective.
- 32:09Dr. Jenny Yang: So I think I build towards a future where we can start doing things at such high throughput because we're doing it, um, in silico.
- 32:18Dr. Jenny Yang: But I do think lab in the loop is really what matters for, for Outpost, especially at this stage, because not… It's not the wet lab or the dry lab that's more important.
- 32:30Dr. Jenny Yang: It's really working together in this flywheel that I think will really drive discovery and change and improvements.
- 32:38Brian: Awesome.
- 32:39Brian: Jenny, this has genuinely been one of the more mind-expanding and informative conversations we've had on this show.
- 32:46Brian: It's easy to forget how much health lives outside the parts of biology we've gotten used to talking about, so thank you for reminding us of that, and thank you for making it all make sense.
- 32:58Brian: We're gonna let you go now, but before we jump, before we jump into our recap.
- 33:02Brian: But thanks again, Jenny.
- 33:03Brian: Appreciate it.
- 33:04Dr. Jenny Yang: Thank you so much, Brian, and thank you for the, the thoughtful questions.
- 33:07Dr. Jenny Yang: I had a great time.
- 33:08Brian: Yeah.
- 33:09Brian: Likewise.
- 33:10Brian: Thank you.
- 33:15Brian: All right, recap time.
- 33:16Brian: Here are the top three things I took away from my conversation with Jenny.
- 33:21Brian: Patients have been telling doctors some version of the same thing for years.
- 33:27Brian: "This doesn't work for me the way it's supposed to work for me." Doctors were listening.
- 33:32Brian: They just didn't have the science yet to explain why.
- 33:36Brian: The microbiome is a real piece of that missing explanation.
- 33:40Brian: Second, getting from people like you to an actual answer for you took a computational leap and a person with the passion to bring it together.
- 33:51Brian: Modeling trillions of interacting microbes at the level of one person's biology just wasn't tractable until AI caught up to the problem.
- 34:01Brian: And finally, the harder challenge going forward is scale.
- 34:05Brian: Proving this works for one person is one thing.
- 34:08Brian: Keeping that same resolution intact across millions of people, across partners and data sets without drifting back to averages, that's the real test.
- 34:19Brian: And that should do it for today's episode of Humanify Health Tech.
- 34:22Brian: As always, thanks for tuning in.
- 34:25Brian: If you're joining us on Spotify or YouTube, we'd love to hear your thoughts in the comments, and don't forget to leave us a rating while you're there.
- 34:32Brian: Once again, I'm Brian Chi.
- 34:34Brian: I'll see you next time