52. Why Your Medicine Doesn't Work — The Microbiome Science Nobody Told You About
Available now
-
Available now -
The Missing Variable in Personalized Medicine? - It Lives in Your Gut
You take a pill. You wait. Nothing happens. Or something worse happens.
You blame the medicine. Maybe you blame yourself. But what if the real reason was something living inside you right now? Trillions and trillions of microscopic creatures — bacteria, fungi, viruses — quietly doing their own thing in your gut, breaking down everything that enters your body, including the medicine your doctor prescribed.
This is the world of the gut microbiome. And according to Jenny Yang, PhD scientist and co-founder of Outpost Bio, it may be the most important piece of the personalized medicine puzzle we have been ignoring for decades.
Mizter Rad sat down with Jenny for a conversation that starts with a simple question — why does the same pill work differently in different people? — and ends somewhere in the year 2078, where medicine, nutrition, and consumer health have been completely transformed.
The 0.1% That Changes Everything
Here is something that will mess with your head. Between you and the person sitting next to you right now, your human DNA is 99.9% identical. You are basically the same organism. And yet, your gut microbiomes can be up to 90% different.
That gap — that 90% — is where medicine starts to break.
"If we look at our gut microbiomes, we can be up to 90% different," Jenny explains. "So really, if we want to be able to finally unlock personalized medicine and understand why you respond one way and I respond a different way, we do need to consider the gut microbiome."
Think about it this way. Imagine two people. Same diagnosis. Same genetic mutation. Same drug. Same dose. One person gets better. The other gets worse or feels nothing at all. For decades, doctors have scratched their heads at this. The answer, Jenny says, has been hiding in the gut the whole time.
When you take an oral drug — or even an intravenous one that travels through the bloodstream — it passes through the gut microbiome. And the bacteria living there will do what bacteria do: they will transform that compound. Your bacteria might break it down into something inactive, so the drug never really works. Someone else's bacteria might turn it into something toxic, causing side effects that seem to come out of nowhere.
"My gut microbiome might break down a drug into a deactivated compound," Jenny says. "Someone else's gut microbiome might break down the compound into a toxic metabolite... that would lead to some sort of adverse side effect."
Medicine hasn't been wrong. It just hasn't been looking at the full picture.
The Field That Didn't Have the Right Tools — Until Now
So why has this taken so long? If the microbiome is so important, why aren't doctors already accounting for it?
Jenny has a clear answer: it's not that the science wasn't there. It's that the tools weren't.
"When it comes to the microbiome, I think this field just hasn't had the right tools or data sets to finally unlock it in a translational or very tractable way."
The human genome has 20,000-plus genes. Your gut microbiome contains 150 times more. And unlike the human genome — where researchers are largely looking at one genome, one protein, one interaction at a time — the microbiome is a full ecosystem. Thousands of bacterial species, all competing, cooperating, and chemically transforming everything around them.
"We're looking at thousands of genomic sequences essentially at the same time," Jenny explains. "And this has historically been really challenging to measure."
This is where Outpost Bio comes in. And this is where AI enters the conversation.
Flying Poop Around the World (Seriously)
Before any model can be built, data needs to exist. Real, reliable, human-derived data. Not data from mice. Not sparse academic datasets full of gaps and biases. Actual human gut communities, sourced from diverse populations around the planet.
So yes — Jenny and her team are flying stool samples in from around the world.
"We are flying poop in from around the world and we are making sure to have diverse samples from different populations because microbiomes that exist in Asia won't necessarily exist in South America or North America or the EU."
The process is as controlled as it sounds complicated. The stool communities are stabilized using a proprietary protocol — because microbes are extremely sensitive to temperature and environmental changes. Then, at high throughput, the team tests different drug compounds and food molecules against those communities. They measure what happens before and after: how the bacteria change, how the compounds break down.
This is the raw material of the model. And it has to be built from scratch.
"When we started building our model, we really started from scratch," Jenny says. "We knew that we had to have control over the data generation process to guarantee that we would have high quality data that we could reliably use to train a machine learning model on."
What Is Metabolomics (And Why It Matters)
One of the most exciting scientific tools Jenny's team uses is something most people have never heard of: metabolomics.
You've probably heard of genomics — the study of DNA. Maybe proteomics — the study of proteins. Metabolomics is the next one to know.
"Metabolomics is an omic that studies the biochemical reactions that occur," Jenny explains. "Now we can say if we have this community of bacteria plus this chemical structure, it will break down into these metabolites."
In other words, metabolomics lets Outpost Bio see exactly what the bacteria are doing to a compound — not just which bacteria are present. Are they producing a toxic version? An inactive version? Or are they actually activating the drug, making it work better?
This goes far beyond correlation. It gives a mechanistic, cause-and-effect understanding of what's happening when your gut meets a drug.
The Microbiome as a Weather Forecast
One of the most compelling ideas in the conversation is the analogy Jenny uses to describe where the field is going:
Think of your microbiome like the weather.
Right now, we can do a one-time snapshot — give a stool sample, get a result. But the real goal is a continuous forecast. A real-time read of what's happening inside you, how it's changing, and what that means for the compounds you're putting in your body.
"One of the goals of the future is to be able to collect enough data from the same person over time that we can then track changes in the microbiome more fluidly. Like, for example, kind of like the weather forecast."
And Mizter Rad pushed even further: what about a biological sensor inside your body, constantly monitoring your microbiome and adjusting in real time — like a glucose pump, but for your gut?
Jenny didn't dismiss it.
"I could see that for the microbiome — a device that will monitor your microbiome and then help modulate any sort of changes that it might see necessary. And it might not be as far into the future as you think."
There are already companies building smart pills that sample your GI tract. The next step is a version that stays, monitors, and reports — your gut as a living dashboard.
Your Toothpaste Is About to Get Personal
If the microbiome affects how your body processes everything that enters it, then the implications go way beyond medicine.
Mizter Rad asked Jenny about the future of consumer products — toothpaste, shampoo, skincare, even detergents — personalized to your specific microbial makeup. And the answer was: yes, that is exactly where this is going.
"Ideally we would be able to have this on a gradient where we could really have a formula that's geared towards our own microbial makeup," Jenny says.
Some makeup stores already have facial scanning tools that recommend colors based on your skin tone. The next version could recommend formulas based on your skin microbiome. Walk into a pharmacy, swab your mouth, wait a few minutes, and walk out with a toothpaste designed specifically for your oral ecosystem.
"Maybe you just take a swab from your mouth and quickly put it into a machine and it will give you the best recommendation."
This is not a futuristic fantasy. The tools are being built now.
A Note on Data Privacy
When the conversation turned to data ownership, Jenny was direct. Biological data — data that comes from your body — sits in a heavily regulated category.
"There are really strict regulations around that. We have to apply for ethical approval for both the use of this data for research purposes or commercial purposes. And we have to outline exactly what we're going to do with the data and we make sure that all the data is going to be anonymized."
This is worth knowing, especially as national microbiome databases start to emerge. Jenny confirmed that biobanks of gut microbiome samples already exist globally — the Human Microbiome Project dates back to either 2007 or 2017. Countries are already mapping this data. The question is how it gets used, and who owns the insights that come from it.
Cross-Reference: The Body as a System
This episode connects directly to some of the biggest themes on the Mizter Rad Show. In the conversation with Andrew Hessel on synthetic biology and Chromosome 24, the discussion touched on N-of-one medicine — the idea that future treatments will be designed for a single individual. Jenny's work on gut microbiome profiling is exactly the infrastructure that makes that possible. You can't have truly personalized medicine without understanding how that individual's gut ecosystem processes drugs.
And in the episode with Gizem Gumuskaya on anthrobots, the idea of engineering biology at a cellular level raised questions about the boundary between the body and technology. Jenny's vision of a sensor that lives inside your gut and monitors your microbiome in real time sits right at that intersection — where biology meets bioengineering.
What Does 2078 Look Like?
When Mizter Rad asked Jenny to paint a picture of the future, she had a clear and specific answer:
"Within the next 50 years, microbiology will be fully computable."
What does that mean? It means that if you give a model a microbial community — the bacteria in your gut — plus a chemical structure like a drug or a nutrient, it will be able to simulate exactly what will happen. Which metabolites will be produced. Whether the drug will be activated, deactivated, or turned into something harmful. For you specifically. Not for an average patient. For you.
"Really I think within the next 50 years we'll be able to input a microbial community and a molecule into a model and then simulate the outcome. Like running a weather forecast but for the living ecosystem that is your body."
And the implications stretch beyond medicine. Microbiomes exist in the soil. In the ocean. In every environment on Earth. The same computational tools that help predict how your gut processes a drug could one day help predict how the environment processes the chemical products we dump into it.
Medicine. Nutrition. Agriculture. Climate. All connected through the invisible ecosystems living in and around us.
The Bottom Line
The future of medicine is not just about smarter drugs. It is about understanding the trillions of organisms that decide what those drugs actually do inside you.
Jenny Yang is building the tools to decode that relationship — one stool sample at a time, from one corner of the world to another.
The microbiome has always been there. We just haven't been able to read it.
Now, we are starting to.
____
This article is based on the Mizter Rad Show episode #52 featuring Jenny Yang and was polished by AI.
Listen to the full conversation with Jenny on the Mizter Rad Show, where we explore how the trillions of creatures living inside you may be the ones deciding if your medicine actually works.
Stay curious. Question everything. And maybe, just maybe — start treating your gut like the sophisticated ecosystem it is.
Mizter Rad
-
Mizter Rad (02:11.055)
We all take medicine, whether it's an aspirin for a headache or a prescription your doctor gives you, we assume it's going to work. But for a huge number of people, it doesn't work. Not because the medicine is bad, but because of something living inside you right now. Trillions and trillions of microscopic creatures, your microbiome. And here's where it gets interesting. Imagine this headline from the year 2078.
That's about 50 years from now.
News from the Future:
You're listening to Global Healthwire.
The date is March 14th, 2078. The World Health Assembly today officially retired the term standard dosage from all pharmaceutical guidelines, closing the book on over a century of one-size-fits-all medicine. Under the new microbiome adaptive treatment protocol, every prescription must now be cross-referenced against the patient's live microbiome index.
A real-time biological signature tracked by biosensors most adults have been carrying in their digestive tract since the 2050s. The numbers speak for themselves. Adverse drug reactions are down 71%. Hospital readmissions for chronic conditions? Down 58%. And tonight, the Global Microbiome Foundation honors the 50th anniversary of OutpostBio's foundational open dataset.
The paper many credit with starting at all. Back in 2025, the team had 5 employees and were considered a very small bet. Funny how that works. Back to you, Mr. Rad.
Mizter Rad (04:12.123)
Thank you, Mark. sorry. Now that future for some may sound like science fiction, but many or maybe that future is starting right now, right here in this conversation. My guest today is Jenny Yang, PhD scientist entrepreneur and one of the few people in this planet who might be responsible for the biggest shift in how we think about personalized medicine in the next 50 years.
She co-founded Outpost Bio, a biotech startup that is decoding, and I love that word, decoding, the relationship between your microbiome and the compounds that enter your body. Think of drugs, nutrients, all of it. Jenny, welcome to the Mr. Rad Show.
Jenny (05:00.856)
Thank you, Mr. I'm super happy to be here.
Mizter Rad (05:04.685)
Amazing. Let's start simple, Jenny. What is the microbiome? And please imagine I'm 15 years old and I need to understand. I'm very curious about what the microbiome is. How would you explain it to me in simple words?
Jenny (05:19.598)
So really, the microbiome is a collective community of microorganisms. This includes the bacteria, fungi, and viruses, as well as their genes. And they inhabit specific well-defined environments. And when you think about the human microbiome, there's the skin microbiome and the gut microbiome, the oral microbiome, et cetera. So there's specific microbiomes that inhabit all those different parts of your body.
Mizter Rad (05:46.897)
And so what are the problem or what's the actual problem that you're trying to solve? Like, and if possible, maybe you can give me a day to day example, like something that everyone in the audience can relate to. like, let's say, imagine like we were saying in the intro, imagine I take ibuprofen, walk me through what's happening and why it might work differently for me than for you. Like if we take the same pill.
same dose with two different humans, why does it completely or sometimes completely interact differently in our bodies?
Jenny (06:27.35)
Yeah, so that's a great question. And I think if we bring it to the gut microbiome and we think about personalized medicine as a concept.
We all know that we're different in terms of our genetic makeup and a lot of personalized medicine has been focused on human genomics. So the last 30, 40 years, we've been looking at human DNA and saying that one person might have a specific gene mutation, which leads to a particular subtype of disease. And we can even choose a medication now that's targeted towards that subtype of disease, but a problem that we haven't been able to solve.
is why two people with the same genetic mutation, with the same subtype of disease, taking the same medication, are still responding very differently. So exactly what you said. And one of the key reasons that's been missing from the personalized medicine picture is looking at our microbiomes. Specifically, I'm talking about the gut microbiome here. So when we take an oral drug or an intravenous drug through the bloodstream,
medicine is going to pass through your gut microbiome either way. And we both have very different bacteria in our gut that will transform that drug in different ways. my gut microbiome might break down a drug into a deactivated compound. So I wouldn't get to experience the therapeutic benefits. Someone else's gut microbiome might break down the compound into a toxic metabolite or chemical structure that would lead to some sort of
adverse side effect. So because we have just very diverse makeups in our gut microbiome, that will lead to us potentially responding very differently. And to put this into perspective, beside genomics,
Jenny (08:21.582)
between you and I and all the listeners on your podcast, we share 99.9 % the exact same human genomics. But if we look at our gut microbiomes, we can be up to 90 % different. So really, if we want to be able to finally unlock personalized medicine and understand why you respond one way and I respond a different way, we do need to consider the gut microbiome.
Mizter Rad (08:50.519)
I see, I see what you mean, but let me just challenge you a little bit here, because basically I'm trying to understand here have have doctors been prescribing medicine for the past decades in the wrong way, our pharma company spending billions and billions testing these drugs that we've been designing traditionally, let's say without taking into account our micro biomes. Are you telling me all this money is being sort of
thrown to the trash and we've been missing the point this whole time of how is this possible?
Jenny (09:26.562)
So I wouldn't say, so I do think there is a lot of trial and error when it comes to prescriptions. For example, there's a lot of different types of medications that claim to do the same thing, but they're coming from different companies, and you will ultimately have.
or you may ultimately respond differently depending on which one you're on and sometimes it's doctors won't necessarily know which one to give you. When it comes to both prescribing and drug development,
Mizter Rad (09:49.169)
Hmm.
Jenny (09:59.852)
I think everyone is doing the best they can with the tools they have. One example is people are looking at genomics for both diagnostics and drug development. That's been proven to be an extremely useful variable to understand. But really, when it comes to specifically drug development and prescriptions, diagnostics, it's just a hugely complex challenge and the combinatorial space
of diversity across different populations and different individuals is so huge that considering every variable all at once is very, very challenging. When it comes to the microbiome, I think this field just hasn't had the right tools or data sets to finally unlock it in a translational or very tractable way. To put it into perspective,
When we look at genomics and the tools that have come out for genomics, I'm specifically thinking AI because there's been a lot of large datasets based on human genomics and a lot of tools created, computational tools created for these datasets. You're typically just considering one genome.
for one type of protein and trying to figure out how it interacts with another. So it's a bounded space, it's isolated. So it's a much more confined problem. But when we think about the microbiome, we're not trying to just look at one specific microbe interacting with a chemical structure, because really you have a community in there and the community of bacteria compete, cooperate, and transform chemically the environment around it.
that we're looking at thousands of genomic sequences essentially at the same time rather than just one. And this has historically been really challenging to measure. So there haven't been definitive data sets that have come out to study these interactions. And there hasn't been the right tools to be able to use it generalizably across both drug discovery and clinical practice. So I think right now,
Jenny (12:19.982)
the medical field acknowledges the importance of bringing in microbiome analytics. We just don't have the right tools yet to be able to do it widespread.
Mizter Rad (12:32.113)
And why, mean, okay, I understand that one of the reasons why it's been challenging is because of the amount of essentially microbes that we don't even understand yet. Some, some we, I understand that some we've been classifying and understanding, but many others and maybe the majority are still a question mark. And maybe that's the main sort of base challenge of why they has been so.
such a black box in a way for developing drugs based on that or advancing personalized medicine. So how does AI play a role here? Because, know, yeah, I'm sorry for that. Jenny, I use AI and I know everyone else around me uses AI these days. So I want to get this right because I understand you use AI in your company for your work on a daily basis.
But what do you mean exactly when you say that AI is an important tool in your processes? Like, how do you exactly and how do you actually use it? Because there's a big difference between, of course, using Chai GPT or Claude and building something foundational like what you guys are doing.
Jenny (13:49.57)
Yeah, so for us, AI really is the core of what we do. Our goal is to both generate data of microbiome and compound interactions and then build AI tools so people can then apply their own questions to our data set. But essentially,
When you think about human genomics, there's only about 20,000 plus genes in the human genome. But if we look at our gut microbiomes...
there's 150 times more genes there. So it's already a very, very large space. And when I mentioned that, between all of us in the world, we're almost 100 % identical in terms of our human genome. There's a very, very small portion that differs. But when we look at our gut microbiomes, we can be up to 90 % different. So that already increases the complexity and the combinatorial space that we have to look at. So we really do need
strong computing infrastructure to be able to even detangle all the patterns that might exist, as well as sophisticated machine learning algorithms to be able to take both genomic sequences or metagenomic sequences, what we're using, as well as other types of omics technology. So we also use metabolomics. So the omic that...
decodes biochemical reactions. So we can actually tell you how a microbiome is breaking down a chemical compound. So to look at this much data at a multi-omic level.
Mizter Rad (15:26.752)
Jenny, one second. me jump right in there. Can you repeat that part starting from metabolomics? And can you tell me what metabolomics you said? can you start from there? Because I think it got cut off. And explain me what that means as well, please.
Jenny (15:38.274)
Yes.
Jenny (15:46.03)
Yes, so in addition to just the
large combinatorial space we're looking at, we also want to get down to a mechanistic understanding of what the microbiome is doing. So if we only looked at the DNA sequences of the bacteria in your gut, all you're able to say really is that there's correlation between this community of bacteria existing and this health outcome. We want to bring in metabolomics as well,
is one of the newer omics that are that's coming up in the field. you've heard of genomics which looks at DNA, transcriptomics which looks at RNA, proteomics that's been one of the biggest kind of
areas of attention in the field of AI for Bio recently. And now there's metabolomics. So metabolomics is a omic that studies the biochemical reactions that occur. So now we can say if we have this community of bacteria plus this chemical structure, it will break down into these metabolites. So they're the breakdown products that we can identify using metabolomics.
And that tells us how a community of bacteria has broken down, let's say, a drug structure. And we can then use metabolomics to identify if those breakdown compounds are toxic or deactivated versions of the drug molecule or activated versions of the drug molecule.
Mizter Rad (17:28.08)
I see. I see. But something is still not clear to me. Because you must have started with a base, a foundational, like I said, like a foundational model. It's like a base model, a kind of starting point that then you used to learn from in the beginning. How did you build that in the first place? And I guess my real question is, how do you know when that
base model or your model in general is good enough to trust.
Jenny (18:04.15)
Yeah, that's a fantastic question. one reason we wanted to make sure that we also generate a data set for this field is because there really hasn't been a lot of large data sets appropriate for machine learning that are publicly available online on microbiome and drug interactions. There's a lot of data sets that are coming out of academic labs that have proven the importance of the microbiome in understanding drug metabolism.
are very small.
sparse data sets. There's a lot of bias that exists depending on the source that we're getting this data from. And many models are actually based on a mouse model. So that's where the state of the art of the field is. And these mouse models have just been known to translate very poorly to human outcomes. the data that is available makes the importance of the problem undeniable, but they're just not the right kind of data to use for
Mizter Rad (18:49.616)
Mmm.
Jenny (19:06.478)
machine learning purposes. So when we started building our model, we really started from scratch. We knew that we had to have control over the data generation process to guarantee that we would have high quality data that we could reliably use to train a machine learning model on.
Something very specific about the data that we're generating is that microbial communities, especially when you take them from the gut, are extremely finicky. They're constantly in competition with one another or working together and changing in abundancies. They're changing based on temperature, other environmental conditions. So we need that environment to be extremely controlled. So we've developed a proprietary protocol that stabilizes and controls these environments.
when we interact with communities with a chemical compound, we can be extremely confident in what we're measuring and what we're seeing. So we're actually generating this definitive data set at high throughput, and then we're training machine learning models. In terms of understanding when a model is good enough, that's also hugely complex because part of what we are gonna have to do is...
validate our findings, there are a lot of known pathways in the literature that we can validate against. So there's a lot of academic labs that have identified microbiome enzyme X plus drug molecule Y leading to toxic compounds, for example. So using our method and our models, we can demonstrate that we can predict what is already known. And then we can also expand.
Mizter Rad (20:47.536)
Sorry to jump in there, but that you said that is already known is proven theoretically, but it hasn't been practically confirmed. Is that correct?
Jenny (21:00.398)
They've been proven both within mouse models and in some human studies. It does depend on which pathway you're looking at, but there's a mix of both, but there still is so much more that we need to discover and that's what we also hope to do.
Mizter Rad (21:17.2)
Okay, I understand. So take me through the process of creating that baseline. If you say you're not using rats anymore, that makes sense to me. Are you using organoids? Are you using real urine? No, actually not urine. It doesn't make sense. Stools? How are you doing it? Take me through the process. And are the stools coming from different places in the world? If you're trying to like...
create like a diverse comprehensive map of the microbiome.
Jenny (21:53.346)
Yeah, absolutely. So we are making sure that we have human-derived stool communities. So we are actually getting stool samples and we are...
Mizter Rad (22:04.144)
stool clubs.
Jenny (22:06.218)
Yes, exactly. So we're flying poop in from around the world and we are making sure to have diverse samples from different populations because microbiome that exist in Asia won't necessarily exist in, let's say South America or North America or the EU. So we are making sure to source from diverse locations.
Mizter Rad (22:15.801)
Amazing.
Mizter Rad (22:23.855)
Right.
Mizter Rad (22:31.384)
Okay, so now that you're able to source those tools from diverse locations, what happens next? Guess to your lab and then what happens?
Jenny (22:40.8)
Yes, we use a proprietary protocol to be able to create stool-derived communities. So using our method, we make sure we stabilize these stool communities so the bacteria that exists in the stool at the abundancies that they exist at will be stable. We, at high throughput, test.
different drug compounds, food molecules, et cetera, at high throughput for all the samples. And then we take multi-omic data. So we take metagenomics to understand the community of bacteria that exists before and after interaction with the compound, and then metabolomics to measure the breakdown of that compound. So we understand what those communities do to the molecule.
Mizter Rad (23:29.488)
OK, on a practical level, I'm not a scientist. I'm never in a lab. I just want to understand how this looks like, just for me to imagine. You have this tool on the one hand. You put it on a dish, petri dish, whatever. I don't know. Maybe you have better tools than that. And then you put a compound to see what the product on the other side is. Correct? You inject that? Or how does that work?
Jenny (23:54.56)
Yeah. So essentially just think, take stool sample, mix it with drug in a very controlled environment and take measurements throughout to see how that drug is breaking down and how that community of bacteria also changes when it meets the drug.
Mizter Rad (24:00.497)
Right.
Mizter Rad (24:11.094)
I see. And then you're testing the whole population of bacteria of that tool or a specific part of it.
Jenny (24:22.52)
So we are trying to understand it on a community level. So we are taking, yeah, the community of microbes that exist in that stool.
Mizter Rad (24:26.574)
I mean it level.
Mizter Rad (24:31.6)
Okay, I understand. When you collect those tools, let's say some come from the EU, some come from the US, I guess in the US the different microbiomes depending on the region, on the weather, the exposure to chemicals, etc. Do you get to understand the profile of the specific person that this tool belongs to?
to get a better understanding of their, maybe their lifestyles or a bit of their background as well.
Jenny (25:09.356)
Yes, we do. So we do take general demographic information. We ask a little bit about kind of their lifestyle choices. But really, we... So I do want to be clear. We're not saying that the microbiome is the only thing that's important when it comes to understanding a drug response or realizing personalized medicine. It's not the only thing. Obviously, you know, our lifestyle choices, other exposures...
and
our genetics, et cetera, all matter. But what we do want to do is be able to isolate one specific component, which is the gut microbiome, and provide the right tools to be able to study this. Ideally, we would want to be able to understand the system levels, understanding of everything that affects human health. But we're really just focused right now on the microbiome. But we do collect additional clinical metadata because
there will be some associations that we can still consider alongside the microbiome.
Mizter Rad (26:15.92)
I understand. OK, so one of the things that bug me sometimes when I think of the microbiome, I've taken tests myself to understand what kind of bacteria I have myself because I'm interested in knowing. And when I did that several times, I tried to understand how this works and what the microbiome is and how it changes, cetera. And I understood that it could be that my microbiome today is very different to what it was six months ago. And you mentioned that.
before. And it changes and it's based on what we eat, but also how often we travel and where we travel. And maybe if I take antibiotics or not. And it's almost like trying to read a book that keeps rewriting itself all the time. So how do you actually and you say that you have a proprietary method or methodology process, but how do you build the reliable model?
in the years to come based on what we know that is an ecosystem that is constantly changing.
Jenny (27:27.52)
Yeah, that's a fantastic question. it's true, day-to-day fluctuations in the microbiome do occur based on what we eat, but really the main dominating bacteria that we see in our microbiomes does stay relatively stable from the time we're really young to when we're much, much older, and really dramatic changes only happen when you do take antibiotics. So there is a relative stability and we're able to find
essentially the main group of dominating bacteria that is stable. But when we think about modeling right now, we really are looking at cross-sections of someone's microbiome in one specific time. And the good news about taking stool samples is that
someone could provide a stool sample at one point and we could immediately give a outcome and then if they wanted to check again at a different point, it's relatively easy to give a stool sample rather than going in for a different type of biopsy for other types of medical diagnostics. But I would say...
One of the goals of the future is to be able to collect enough data from the same person over time that we can then track changes in the microbiome more fluidly. Like, for example, kind of like the weather forecast, it'd be great to be able to have that for your microbiome.
Mizter Rad (28:54.692)
Mmm.
Mizter Rad (29:00.812)
That would be amazing. So basically what I'm thinking here is that if it changes so regularly, I guess it is a good idea to test it regularly. And almost like, you know, what people usually do with blood every year, every six months. Do you think this is going to become a routine? Like a routine thing? Should it become a routine if you're concerned about your health?
Jenny (29:29.174)
I definitely think so. think we will be doing regular stool tests, like microbiome testing, to understand how our gut microbiomes are changing over time. Yeah, exactly, and affecting our health.
Mizter Rad (29:46.786)
Reacting. Yeah.
Yeah. And do you think that at some point in the future, maybe this is a bit out there, but we'll have some sort of biological sensor living inside of us, like a device that is monitoring the microbiome, the gut and shifting in real time, like you said, without really providing this tool, but actually just, you know, have being able to control it maybe with your watch. I know it sounds a bit sci-fi, but
Do you think that's sort of like the way it will develop in the decades to come?
Jenny (30:25.96)
absolutely. I think that's a very, very cool idea. when you think about blood glucose monitoring or hormone monitoring, glucose pumps, they look at your blood glucose level and then the pump will give you some insulin. And I could see that for the microbiome, like a device that will monitor your microbiome and then help modulate any
Mizter Rad (30:34.06)
Exactly. Yeah.
Jenny (30:53.45)
any sort of changes that it might see necessary. And it might not be as far into the future as you think because right now there's already companies that are creating these little pills that you take that sample your gut microbiome. You can poop it out after, but it really just samples the bacteria higher up in your GI tract. But you can imagine.
Mizter Rad (31:09.552)
really?
Mizter Rad (31:16.238)
wow, but so it has like an arm or something that collects the, or how does that work, do know?
Jenny (31:24.302)
I can't quite remember, but I think it's like a pill that will open up a little bit, collect the bacteria, and then you poop it out and you can have access to the bacteria. But you can imagine a version of that pill that stays in your gut a bit longer and monitors it.
Mizter Rad (31:33.381)
Wow.
Mizter Rad (31:36.687)
Right, Yeah, wow, amazing, amazing. Okay, let's shift gears a bit. I want to talk a bit about the business and also how you ended up here, because I think it's super interesting. Your story, how did you go from Canada to Boston, I guess to London also, and now be where you are?
Jenny (32:03.508)
Yeah, absolutely. So I grew up in Canada. I did engineering physics at the University of British Columbia in Vancouver. So I spent a lot of time there during my time in undergrad.
I had a lot of interest already in applying machine learning to some sort of medical or biological field. This led me to essentially creating a small little genome data set scraping tool just in my own time. And I cold emailed probably 25 professors asking them to just look at it. And then Professor Stephen Jones at the Genome Sciences Center.
out.
actually offered me a job right then and there after he saw that and that's what really led me down the personalized medicine path. So I spent five years in different positions working with Steve on different projects where I would apply machine learning to DNA sequences and really fell in love with research within that field at that point. Went down to Stanford for a little bit to a CRISPR-Cas9 lab. So when gene editing technologies was making a lot of breakthroughs, I really wanted to
again, bring machine learning approaches to DNA sequences. I then decided to fully dedicate myself to research, so I went over to the University of Oxford where I joined a clinical machine learning lab. There I...
Jenny (33:33.932)
We got to work very closely with clinicians to actually build AI based screening tools specifically for COVID-19 triaging. So when people came into emergency departments, we could triage them into correct waiting rooms based on whether or not we thought they had COVID-19. Went over to Vietnam, got to work with clinicians across two hospitals in Vietnam to build the same types of screening tools. So really got to experience translating machine learning algorithms.
to the real world for medicine and also spent one year out of my PhD working at Accentia, the AI for drug discovery company, again in the heart of the AI team working on clinical data sets. So really my passion over the last 10 plus years has been around machine learning, being able to elevate the medical fields and
that's definitely where my passion lies. would say I really am driven by the idea of personalized medicine and I have spent a lot of that time working on genomics datasets. So it really made sense for me to now move into the microbiome space because...
With the new tools that have come out, I think we can finally unlock this space. And knowing how diverse everyone in the world is based on our microbiomes, it's a really, really important field to push forward so we can hopefully finally realize personalized medicine. It really was the startup journey that's now brought me to Boston.
Mizter Rad (35:08.558)
Mmm.
Mizter Rad (35:14.894)
Right. Why Boston, by the way?
Jenny (35:17.71)
For multiple reasons, obviously it is one of the biotech hubs, which is fantastic. I also really like the city itself, so I could imagine myself living there long term. And it's also almost exactly halfway between Vancouver, where my family is, and London, where a lot of my colleagues and friends still are.
Mizter Rad (35:40.177)
I see. you talk about breakthroughs and I want to so a lot of listeners have, know, our entrepreneurs, investors, a lot of them are within the e-commerce, you know, industries. And when you think about e-commerce, you have a breakthrough in e-commerce on your e-commerce platform when you break even, right? When your sales are bigger than your
costs and in a monthly basis or in a yearly basis that is constant and so you break even. You're safe. So to say or you're profitable at what point does your model specifically talking about business breaks even or at what time at what moment does that model become actually valuable enough so that someone pays for it or buys it out.
Jenny (36:37.858)
Yeah, that's a great question. And it's definitely much more challenging to be able to pick just one break-even point because there's a lot of important tasks that I think can stem from the data sets and the models we're building. But I think even if we...
Mizter Rad (36:46.574)
Right.
Jenny (36:57.176)
focused on one specific vertical. So one example would be, let's focus on solely looking at amount of metabolism that happens to a drug based on a certain microbiome community. If we can say that, even if we turn into a binary problem, if we can say that these people are going to experience more than 50 % metabolism due to the microbiome versus another group that
experiences less than 50%, we can say this group that experiences less than 50 % microbiome metabolism will be most likely to respond well to a therapy because that therapy is more likely to reach their bloodstream. I think that's already a huge win because that tells both people designing clinical trials which populations they should better target and then clinicians who are
making prescriptions, it gives them a reason to prescribe one drug over another to one of their patients. So I think being able to even say that for one specific type of drug or one class of drugs.
is going to be a win and then hopefully we can expand this beyond just that. But of course, I think there's a lot of different wins we could have that will actually make a tangible impact in terms of improving our ability to have more drugs that work across more people and choose drugs that will most likely work on different people.
Mizter Rad (38:38.074)
Right. So again, talking about business at this point in time, do you have a clear customer or a clear target that you are aiming for? And if, well, first you answer yes or no. And if, if no, if it's not so clear yet, what do you think that ideal customer could be in the next decade?
Jenny (39:07.06)
Yeah, so right now there's probably...
two clear subsets of customers for what we're doing. we are at early stages of company building. So when it comes to understanding a community of bacteria and a compound interaction, we have a strong focus on the gut microbiome specifically and any compound that enters the gut. So that would be both drugs and food molecules. So the two buckets of customers we have would be both pharma and then CPG companies and consumer
in the nutrition space. both of those sets of people, whether it's in their R &D team or computational team, they're interested in how their specific molecules are going to be impacted by the gut microbiome and how their molecules will impact the gut microbiome.
Mizter Rad (40:02.576)
But the way it could work is that they send you, let's say, a compound, and you test it with your proprietary method system. And then you send them back the results. Is that more or less a simplistic way of seeing it? Is that how it works?
Jenny (40:22.444)
So that's what we can do at the moment because we're still building up a large enough data set. So at the moment, we can run these screenings in-house. And then ideally, as we build up a large enough data set for machine learning applications, we can then do it in Silico.
Mizter Rad (40:40.024)
Hmm. In silica, does that mean?
Jenny (40:43.214)
computationally essentially. So we can either provide them a machine learning model or the data for them to apply the tool in-house or we can just screen their compound using our model and then we wouldn't have to rely on a wet lab anymore.
Mizter Rad (41:02.736)
I see. see. Okay, so talking about dry labs, data, all this produces a lot of data. Where does this data actually live? Who owns the data? And, you know, maybe here's where I'm going to push you a bit. If I give you my microbiome data, what guarantees do I have that it's not going to be used in ways that I didn't agree to?
Jenny (41:30.464)
Yeah, so luckily there are a lot of regulation surrounding data privacy, data safety, and data usage across both, like all countries around the world really.
So even when we are collecting data, we have to apply for ethical approval for both the use of this data for research purposes or commercial purposes. And we have to outline exactly what we're going to do with the data and we make sure that all the data is going to be anonymized. So there are really strict regulations around that, we follow, other companies follow or should follow. But there are a lot of ways.
Mizter Rad (41:57.72)
Mmm.
Mizter Rad (42:12.822)
Okay, I didn't know this. I didn't know this, but let me just write jump right in there. Is this is this specifically to data that belongs to your sort of body and metabolism? Because I feel like the data that you produce with your phone and so on is it lands on a different bucket kind of is that correct?
Jenny (42:34.314)
That is correct. And there will be different rules around different sets of data. So there is specific rules around data that comes from a human.
Mizter Rad (42:35.929)
Okay.
Mizter Rad (42:44.536)
I see. Interesting. Interesting. Yeah, because I understand from conversations I've had before that on a country level, there are some countries already collecting and mapping their citizens' genomes, for example. So a lot of data being collected there on a country level. I think the UAE here, Dubai, Abu Dhabi, they're already doing that as well. Estonia is doing that.
Do you think we should be doing the same thing with microbiomes and on a like a national microbiome database? Do you think that's a good idea or rather terrifying idea?
Jenny (43:24.654)
think it's a fantastic idea and it is already being done. I actually think that the human microbiome project might have been started back in, it's either 2007 or 2017, but people are doing this around the world. There are bio banks of these gut microbiome samples. The only challenge about these data sets with respect to what we're doing is that,
A lot of these data sets won't have the bacteria still alive. And to run the experiments we do, we need the bacteria to be alive so they can actually interact and chemically transform the chemicals that we interact them with. So a lot of these biobanks will be able to represent the diversity of different microbes that exist. We just wouldn't necessarily be able to use them for the purposes we need to right now.
Mizter Rad (44:01.041)
Okay.
Mizter Rad (44:19.12)
Okay, I see what you mean. Okay, this one is going to sound a bit crazy, but follow me here on this. If the microbe or microbiome affects how we absorb everything that enters the body, you know, then, you know, food, but also maybe the air we breathe or maybe the perfume we use. Maybe it's not just about medicine. And you were talking about before.
fast consumer goods, foods as well. Do you think we could end up with maybe personalized toothpaste and personalized shampoos or personalized detergents or even like the fabric of my clothes could be personalized? Because ultimately maybe this also affects maybe not just the gut microbiome, but the other microbiomes. Or do you think I'm going too far here?
Jenny (45:20.971)
I think that's...
Absolutely the vision that we would want for the future that completely makes sense if you think about it the microbiome on my skin is different from the microbiome on your skin and even when you look at all the skincare and makeup products that are coming out They're targeting different types of skin profiles They're saying if you have dry skin use this if you have combination skin use this if you have oily skin use this and Ideally we would be able to have this on a gradient where we could really have a formula that's geared towards our own
microbial makeup. And like you said, we have an oral microbiome as well. So having the right toothpaste, you know, some people are more prone to cavities than others. I would be great to see why and fully understand that. For sure, these CPG companies are looking into that. Other research labs are looking into that. But ideally, you know, we would be able to have personalization across all our microbiomes on our body.
Mizter Rad (46:20.656)
So if we want to paint a picture of the future, if I go to a pharmacy or CVS or, you know, in Germany, we have something we call DM or a supermarket just to buy a toothpaste. How would that look like? Like physically, that is not going to be like the, you know, the traditional Colgate or whatever since just standing there one fits all.
But how do you see that? Like machines that where you like sort of scan your face and it produces real time your precise toothpaste or have you thought about this?
Jenny (47:02.73)
Yeah, absolutely. So I think this is going to depend on which microbiome you look at. But for example, right now when I go into a makeup store, there's some makeup stores that have a tool that will scan my face and then it will recommend, let's say, a color of makeup that would best suit me. So I could imagine if you're trying to access your skin microbiome makeup, you could probably do that in store with a tool because it's just touching your skin.
If you wanted to look for a toothpaste, maybe you just take a swab from your mouth and quickly put it into a machine and it will give you the best recommendation. When it comes to something like your gut microbiome and a stool sample, maybe that will take more time. Maybe provide a stool sample. And kind like when you go to the pharmacy right now, you maybe wait an hour for them to run the diagnostics and then they give you recommendations after. Exactly.
Mizter Rad (47:41.776)
Mmm.
Mizter Rad (47:55.854)
Right. The test.
Mizter Rad (48:00.977)
Yeah, that makes sense. That makes sense. OK, last question, Jenny. And this is something I like to ask all of my guests because the Mr. Rat Show is about the next five decades. And I would like to know how you see the world in those next five decades. Like, what does the decade from 2027 to 2037 and
after look like through your eyes from your perspective.
Jenny (48:37.194)
Yeah, so really, think within the next 50 years, microbiology will be fully computable. So really, like we mentioned today, if you give me a microbial community, for example, the bacteria in someone's gut, as well as a chemical structure, like a drug or some sort of nutrient, we really can't reliably predict what will happen. We might be able to know what a single bacterial strain does in isolation, but real
biology is not happening at that isolate level. It really is happening at that community level where there are thousands of species interacting and transforming their environment. So because this is such a high dimensional combinatorial space, I think within the next few decades, we're going to be unlocking more and more of that space, specifically at a level where we can now computationally calculate what's going to happen when a community of bacteria interacts with a
structure. So really I think within the next 50 years we'll be able to input a microbial community and a molecule into a model and then simulate the outcome. Like I mentioned earlier, maybe we can liken this to running a weather forecast but for instead the living ecosystem that is your body. Once we do that I feel like medicine, nutrition, consumer health...
will really shift from just trial and error to a more predictive design. I think if we're really talking on that 50 year milestone, we can go even bigger in our vision because essentially we're looking at a community of bacteria and how it interacts with the chemical structure. At Outpost Bio, we're interested on the human aspect of that, but you could see how this also applies to
different climate applications. There's microbiomes in the soil, and we're clearly trying to produce a lot of products for the environment, which are chemical products. So we would want microbiology to be able to be at a point where we would understand those interactions as well. So I think in 50 years, microbiology will be computable.
Mizter Rad (50:56.762)
Jenny, this has been incredible. Thank you for helping us see what's actually invisible.
Jenny (51:04.034)
Thank you so much, mister. I've really enjoyed the chat today.
Mizter Rad (51:08.699)
Thank you. Thank you so much. Beautiful humans until next time. Stay curious. Question everything. And maybe, just maybe, start treating your gut like the sophisticated ecosystem it is. Because the organism living inside you might just be the ones deciding how long you stick around. Hasta la vista. I'm going to stop recording now.