How TECHNOLOGY Is Preventing & Detecting CHRONIC DISEASE! | Noosheen Hashemi & Mark Hyman

Generic advice doesn't work. Personalized advice works. People are infinitely interested in what works for them, not what works for everyone else.

To me, the holy grail is applying the technology of AI and big data analytics to all the new framework we have for medicine, which is, let's call it functional medicine or systems thinking or whatever you want to call it, network medicine. which understands the body as this network as opposed to diseases. And that combined with the omics revolution, all the data from omics and all the data from Quantified Self provides an enormous potential to revolutionize how we diagnose and treat disease and a potential for providers as well as patients or people to have support for decisions of what they do for their health. And we're not there yet, but to me, there is no way, given the complexity of human biology, that we can continue to actually sort of do the medicine the way we've been doing it. And, and I think the, the work that you're doing, Ai and, and the technology and the understanding of the complexity of human biology really helps us to sort of reframe how we're going to do things. You know, talking to my daughter, who's about to start medical school in a month, and, you know, she's, she's going into what I would say is a sort of Dark Ages view of biology. And, it's frightening to me because I know she's going to get indoctrinated in a paradigm that is really outdated. And, it's not that it's irrelevant, it's relevant, but it's just a surface of what's happening. And, so I'd love to hear from you how with January AI and AI in general in medicine you see the change happening where all this data gets put into machine learning system and AI system and how that will change medical care, how it will change what people do, how it will change what doctors do. Because as an individual doctor, you know, I'm lucky. I basically had the privilege of seeing, you know, tens of thousands of patients literally doing millions and millions of data points. So I have, I have like a pattern recognition machine in my head, but it's certainly not good as, as a machine. But I can tell you after doing this for so many decades that in my patients, I see patterns over and over that no one subscribed. I'll give you an example. Let's say someone's got high levels of mercury. Well, often they have oxidative stress, they have low amino acids, they have problems with methylation, B vitamins, they may have low zinc and minerals, and they have low glutathione because their body trying to detoxify. I see this whole pattern and I'm like, I know what's going on. But that's never in the center, not in a textbook. It's just something that I noticed. And what's true about what you're doing is that there are millions of these kinds of examples of undiscovered archaeological excavation of human biology in ways we've never done before that is only now beginning to be possible.

So yeah, so basically what AI has done, I think most people, your listeners are familiar with what AI has done for, say, driving. For, you know, a lot of people have some self-driving features in their cars, some folks do, that warn you when a car is getting close to you, for example, or if you're getting out of your lane or you're getting too close to the car in front of you. We know that AI is working in fraud protection. It's, you know, Google's correcting your spelling, right? People are, they're using that all the time. You mentioned radiology, right? I think AI can democratize medicine. Today, the difference between an expert, absolute first-rate radiologist with someone with not great expertise, not many years of expertise, is essentially there is a difference right now. There is a difference in the care that people get right now. But I think with AI, we can democratize it. How? Well, AI is fantastic for pattern recognition, right? So it can bring kind of the genius of the finest doctor to all, much like, you know, we were able to bring the best classroom professors to the world by, you know, democratizing education. I think AI can democratize medicine by sort of, so to give you an example, there are 122 million people are believed to have pre-diabetes or diabetes in America. There are 7,000 endocrinologists in this whole country. So the so if medical doctors good luck with that. So if people are not taught to look at novel markers like CGM data and heart rate data, if they are not taught, if doctors are not taught nutrition in school, you tell me how they're going to how well equipped are they for for for observing metabolic dysfunction early and for detecting it and pointing it out to people. So, I think what does AI do for us? Well, AI can interpolate huge amounts of data that is not human interpretable by, it's just not human interpretable. So, let's use the simple example of CGM. If you're wearing a CGM, maybe you remember the food you ate yesterday or the day before, or maybe you use a product that tells you what you were eating last Tuesday. Okay, but can you keep in mind what your heart rate was then? Do you remember how many hours you were fasting then? Do you remember what you were eating that gave you this spike? Now, some people can give you a picture of what you were eating. January gives you the fiber, the amount of fiber you had in your stomach. It tells you the macros of your food, that sort of thing. So, the data, this collection of data we're talking about that's growing, which is classic markers from doctors, new consumer health data synthesis of this massive amount of data cannot be done without AI. It just can't. It's not human interpretable. It just isn't. So AI can give us that. The other thing AI can do is can predict for us. Once it learns patterns, it can predict. Mark works out every Tuesday and Thursday, and when he does, he has his protein shake that he always has. Let me suggest a better time for the protein shake. Or I can see that when he runs before the protein shake, he has a better response than when he runs after the protein shake for Mark. Or I noticed that the time window between his last meal and his sleep is too close. It's like one hour. I'm going to over time, this is what January does today. So we use these interventions to give you examples. One intervention we use is fasting. We help people understand their eating period and their fasting period. And you know the ultimate test of metabolic fitness is how well you go from fast to fed and all of that, but that will be in the future. But at this moment we can tell you what's your fasting period, what's your eating period. Sometimes people come into our program and they've been eating 16 hours a day, fasting eight hours. Over time, we slowly change this 15 minutes at a time. We get them to fast, say, 12 hours, eat 12 hours. And over time, we encourage them, like myself. Now I eat over an eight-hour period. I fast for 16 hours, no problem at all. I've done it over a long period of time. So that's one lever. The second lever, as you know, glycemic response is highly impacted by the amount of fiber in your body. How much fiber you have in your body from last 24 hours impacts how you might spike on the next thing that you eat. January measures the amount of fiber that's in your body that you've had in your diet. So that's another one. Another one is just to understand glycemic load, glycemic index of what you're eating, offering you alternatives to those foods that taste the same but have lower GI/GL. The other one is to essentially help you figure out how much to work out, when to work out, to keep your blood sugar in a healthy range. It also helps you manage your calories because as you know, intermittent fasting in combination with calorie management has shown to improve insulin sensitivity. And the two are very powerful together. Dari always reminds me, not intermittent fasting alone, not calorie management alone, together they are very powerful. So these are things that we can do with AI that we could not do before. Your doctor can't follow you around, they're not in your body, they're not a sensor. They can't, no coach or no one can see inside your body, but a combination of wearables. By the way, wearables are going to be so exciting. We have seen, you have seen very little so far. Wait till we get to continuous lactate sensors.

Tell us about the future. Yeah, because with the wearables, they're interesting and they're novelty for a little while. People get tired of them. I've had different wearables over the time. And it's interesting. entertaining, but then I kind of get bored of it. How do you address that?

Yeah, so what are they telling us today? Sensors are telling us a lot today. They're telling us, as you know, about VO2 max, which some people say is the ultimate measure of longevity. They're telling us about gait. They're telling us about blood pressure, caloric burn. They're telling us about quality of sleep, the amount of sleep, resting heart rate, or heart rate variability. They're telling us a lot today. But What's coming? First of all, as I said, that's not enough because we need to understand food to make sense out of all of these things. But what's coming down the road? Well, continuous ketone monitors, they're going to be fantastic. You've seen the announcement from Abbott where potentially the same sensor that tells you your blood glucose can also tell you your ketone level, which is great because it will today, we look at, we have invasive ketone monitors today in the hospital, end of life in the ER, Yeah. We are using very invasive, but actually that's for lactate, but for ketone, it will warn of impending ketoacidosis, you know, for type 1s, this is really, really important. If their blood sugar is going too high and they kind of need insulin, that would be really useful. Also, people that are on super low calorie, low carb diets can see how their metabolism is faring once they have continuous ketone monitoring. And another thing that's coming in is the continuous lactate monitors. This will essentially, this is what's used, like I was saying earlier in hospitals end of life, they use invasive lactate monitors to see, for example, severity of sepsis, like how bad it is when someone is close to death. But But also athletes, we will love to have continuous lactate monitors for athletic performance. Another thing, of course, and the ultimate, is the continuous cortisol monitors. Once we get cortisol monitors, we will understand chronic stress, which impacts, of course, immune and inflammatory systems, which, as you know, increases susceptibility to diseases like autoimmune disease, cardiometabolic disease, mental diseases, and cancer. So really, really looking forward to ketone and lactate and cortisol monitors coming down the road. This will also be transformative. Is there any insulin monitors coming? We need continuous insulin monitors. The fact that we look at glucose and we don't look at insulin is insane. Yeah, that's right.

It's insane. It's true. I mean, glucose is a big thing, but it's maybe hard to measure insulin.

But it's such a more relevant marker. It is so relevant. It is utterly relevant. And so, there are some technologies way more relevant, way more relevant. I think there are some technologies that I know of in some labs that may be approaching that, but nothing that is going to be commercially available as far as I know today.

Tell us why insulin is more important than sugar.

Insulin, so there are two sides of a coin. Insulin is a hormone that lets your cells take up glucose. People with type 1 diabetes are born without the ability to even produce insulin, so they are lacking insulin. Some people, like my co-founder, Mike Snyder, produce insulin but produce it slowly. So he's a type 2, so he's a person with type 2 diabetes. So if you don't produce enough insulin or insulin is lacking, you have all this extra glucose sitting up in your bloodstream. that then can damage your body in a number of ways. Excess glucose can create microvascular complications, as you know. You can essentially develop retinopathy, neuropathy, you can have diabetic nephropathy, so you can have liver disease. Essentially, when, so let's go through it one by one. Basically, if you have- the insulin is.

The first thing that goes wrong.

Insulin is the first thing.

And actually, if your sugar can be perfectly normal and your insulin can be all over the place to try to regulate your sugar, and it's the first clue that something's wrong. Absolutely. It's just one of those things that's stunning to me that one doctors don't measure because it's cheap and easy to do. Absolutely. And two, that we haven't developed a continuous monitor for that. That to me is a game-changer because when people see that number, that's going to be more predictive than even their blood sugar and what's going on with their metabolic health.

100%, yes. If you need to understand your pancreas function, we don't know how much someone's pancreas capacity is right now. We don't, that would be hugely, hugely useful. Because like diabetic retinopathy is the most common cause of blindness right now for working adults in the developed world. As you know, in diabetic neuropathy, basically high blood sugar can injure nerves throughout the body and damaged nerves in legs and feet, which can lead to foot ulcers and amputation. You know that. And then of course, the kidney disease coming from diabetes is the number one cause of kidney failure. Almost a third of people with diabetes develop nephropathy. So yeah, these are pretty serious and you would think, that if we're solving for continuous glucose, we would solve for continuous insulin to understand the whole picture. And for some reason, we are not.

Can you just distinguish between artificial intelligence and machine learning for people? Because I'm not sure everybody understands that.

Well, just machine learning is the act of developing models and training models with the data that you get. And that's really what we were trying to do is essentially create a model that could You could create a model that could guesstimate the nutritional values of foods, for example. That's a model. You get to artificial intelligence when things start working on their own to some degree, and you kind of create this loop where the machine is learning by itself. You kind of have the model and it is going, collecting data, brings it back, goes back into the model, re-trains and relearns. and uses it. So they're used interchangeably. But for example, machine learning and data science, you hear a lot about data science. Those are completely different. Data science is basically looking at data and looking for patterns where machine learning is actually model development. You actually develop a model and you continue to refine your model for accuracy, for-- It was learning.

It's literally learning from data.

It's literally learning. Yeah, exactly right. Exactly right. So the problems we set out to solve were first, food labeling. So we aggregated the most gigantic 16 million foods, the most biggest database of curated databases, so this mega database of databases of recipes and menu items from local restaurants, not chain restaurants, local restaurants, your mom and pop shops. and grocery items, everything. Everything that was possible to get in America, we aggregated into one database.

Really? That's incredible.

It is. It is incredible. It's a lot of work. It's expensive. And then we used machine learning to guesstimate its nutritional values. Anything that didn't have nutritional values. And that is a very complicated process we went through. We have a lot of intellectual property files around our abilities, pending patents around how we did this. Then we realized that glycemic response was better associated with glycemic index and glycemic load than with just carbs. Because you don't just eat pure carbs, right? You eat carbs with other stuff. With fiber, there's water in the food. So we felt that it was really important that we understand GI/GL foods. So our next machine learning project was to estimate glycemic index and glycemic loads for these 60 million foods.

Which is very interesting because when you start to understand about personalized medicine and nutrition, you realize that even the same food in different people can have profoundly different effects because of their microbiome, their immune system, food sensitivities, their genetics, there's so many factors. So, you know, glycemic load isn't glycemic load. It's like it really depends on the person.

Which is why we ran a trial. So we ran an observational trial, put it out there, 23,000 people applied to it, took 1,022 people, including healthy individuals, people with pre-diabetes, and people with type 2 diabetes, 250 of them were those, and associated people's glycemic response to the GI/GL of the foods they were eating. We turned that into a prediction model. And that prediction model is what we presented at the American Diabetes Association in June of 2020 as a poster. This was Mike Snyder and our scientists together presenting that poster. That poster really described how we had done our work in Silico, using AI first, kind of using machine learning, and then had put it into humans to vet it. And then how our models were accurate 33 hours into the future with a high level of accuracy. So what are the uses of these models? What does it do? What's it good for? Well, when you can predict someone's glycemic response, you can let them compare foods like any two grocery items, any two restaurant menus. They can compare any two recipes. Say you want to make pancakes for your kids this weekend, but you don't want it to be very high high glycemic index. You want it to be something lower. You can compare recipes, you can decide which one you wanna make. Maybe you want to know in essentially in the price of what you're eating in minutes of walking. Like, hey, if you eat this fried chicken, you're gonna have to walk 36 minutes. Do you have 36 minutes to walk to put your blood sugar back in a healthy blood glucose range? You don't have 36 minutes? Maybe save this for the weekend. How about, you know, so it essentially tells people that truly closes the human behavior loop because then you know what this is going to do. And then it also allows us to-.

Oh, you have to track it because, right, so part of what you do is you help people track their blood sugar.

Yes.

And so that gives them real-time feedback. So while you may have determined through your machine learning model that this or that food has this or that glycemic index, or load, when they eat it, it might be different for them.

Actually, the prediction is for them. The prediction is for them. So, based on what though?

Based on their-- because it's not based on the food.

It's not based on the actual food. It's based on the actual food. But after four days of training, that's what the poster presented. After four days of training with January using food data, glucose data, and heart rate data, January can predict your glycemic response to any food in the 16 million strong.

Wow, that's amazing.

So, you don't have to put foods through your body. You can put them through the AI. Let the machine tell you how you're going to respond.

So, if you eat a plum, it might be fine. If I eat a plum, it might not be fine, right?

Exactly. But your prediction and my prediction are going to be different. They're not going to be the same.

Although, it's complicated because this gets even to a meta level of complexity with what the power of AI may be able to do. When you think about the 16 million foods, they're not all the same. So a carrot grown on a regenerative farm is different than a carrot grown in a commercial farm. If you have, I mean, there was just a study published today, nature looked at the metabolomic analysis, literally the metabolic features on the molecules in grass-fed meat versus plant-based meat. and even though the nutrition facts label was identical, they were profoundly different in so many qualities. Absolutely. And I think that's, it's something that people really have trouble with because, you know, if you say, well, yes, meat, you're gonna do meat, but what if you do grass-fed meat or kangaroo meat, for example, we talked on the podcast, lowers inflammation, wild meat, whereas feedlot meat increases inflammation, same amount of volume of protein or grams of protein, but profoundly different biological effects.

How do you account for all of that? That seems like a nightmare. That's our dream. That's our dream. It's our dream to be able to, I mean, what you just said is what Christopher Gardner told us four years ago. He said, look, tomatoes grown in Sonoma are gonna be different than tomatoes grown in Chile. But we are heading to that level of precision. We will be able to tell you your carbon footprint eventually of where your foods came from and what kind of nutrients were in the soil, where they were grown, the method of cooking, and everything else you just said. I mean, precision, we mean precision, right? So some of that can come from companies like us. But some of it has to come from government and government policy, which is why I support Tufts School of Nutrition, because it has such a strong policy program. Government policy, don't underestimate the power. Government can outspend private sector, as you know, tremendously. It can outspend philanthropy. Government policy is critically important. They're part of the solution and we need to bring them to the table. But you're right, that's the precision we're headed for. and January has just started this, but we dream of greater precision.

So, what you're describing, machine, is a radical transformation in our thinking about medicine, which has pretty much been one size fits all. You have diabetes, you have the diabetes drugs. You have rheumatoid arthritis, you have the rheumatoid arthritis drugs. And, it's so uniform and there's really very little accounting for the massive complexity of human biology. and there are, and not only the complexity in biology, but the inter-individual differences in our biology. So no two people are the same. So you're talking about personalized medicine, personalized nutrition, Precision nutrition, Precision Health, Precision medicine. All these terms are floating around, and they're speaking to the fact that, that, that we've just really barely begun to scratch the surface of understanding how our bodies work and, and, and also we're so focused on the end stage diseases that we haven't really taken a step back to go, what is happening in our biology like decades before our first symptom? Now we know, for example, on an MRI through functional MRIs we can see inflammation and changes in brains and Alzheimer's patients 20 or 30 years before they get a problem. Or we can start to look at patterns in your blood tests that as a child that predict your risk of getting diabetes in the future.

100%.

And we just, patient I had years ago who came in to see me Canyon Ranch where I worked and their blood sugar was like 115, which is on the way to diabetes. It's like 126 is diabetes. I said, gee, your sugar is pretty high. You've seen your doctor about it. He goes, oh yeah. I said, what did your doctor say? Well, the doctor said, we're gonna wait and watch it. I said, watch for what?

I said, wait until I get diabetes.

And then I'll give me a drug. And I'm like, whoa. So I think what you're talking about is a real revolution or thinking about medicine, particularly traditional health care is very dogmatic and very structured and rigid and And you're kind of coming out from left field with an idea that has the potential to transform how doctors address disease and how people themselves can be empowered to transform their health. So how did you start to go, oh, this is something I want to do?

Yeah, absolutely. So kind of how I got started is I decided in 2016 to start my own company after many years of, like you said, running a family office, investing and serving on boards of companies and nonprofits. I had early success at Oracle, where I rose from the bottom of the organization in 1985 to vice president at age 27, taking the company from 25 million to 3 billion. So it was quite a ride. So I decided I was an operator and I was going to go hands-on and bet on myself as good old Larry Ellison has always taught all of us to do. And so I started a match-- Bet on yourself.

Bet on yourself.

Bet on yourself. He always bets on himself. That's a good one.

That's a good one, yeah.

Yeah, so I started a massive search into the thesis that were getting a lot of attention, kind of big trends over the next decade, and most importantly, what I knew and loved, you know, the classic kind of Ikigai. So I happened to attend this White House Stanford University conference on societal benefit of AI and how to integrate the ever evolving AI into the real world. there was a healthcare panel that was phenomenal. Fei-Fei Li, who had organized this conference with Russ Altman, suggested that those of us who were interested in health and machine learning go to this conference in LA in two weeks. So, booked my flight that night with a friend who was doing research with me, went off to this conference. There we met Larry Smarr. I don't know if you know him or not, but he's one of the most self-quantified people on Earth. And he was keynoting that conference. he had diagnosed his own Crohn's disease way before symptoms had manifested. And kind of the common theme between all the presentations at the conference were that machine learning could essentially fill in the gaps for missing variables in research and not just going forward, but going backwards. So you could go back and look at research that had been done before and fill in with machine learning for missing variables. And this was this really lit a light in my head because I'd been talking to Andrew Lo at MIT about how little research there was in medical research. And, you know, ever since Nixon- How.

Little research there was on AI, you mean?

No, how little research there was on human body. Like we basically, 50 years ago, Nixon declared a war on cancer. And after 50 years, we have a few therapies and now we think of cancer as chronic condition. We have nothing for neurological diseases. We have nothing for aging. And it kind of became obvious that A, we needed everyone to contribute data. Everyone needed to be part of research. Doing 25 people in this university and 85 people in that university and 1,000 people here and 2,000 people there wasn't going to solve for health in America. It wasn't going to. In fact, China is taking this with gumption. That's a whole other story if we want to get to it. But so the answer was to get everyone contributing to research, like 40 million people have taken genetic tests now. We have that information. We can use that information. But secondly, it was to apply AI and machine learning to be able to see patterns and to be able to see how people are different. And I sort of became obsessed with this idea of this kind of collecting multi-omic data. And I don't know if you remember human longevity at the time, a company called Arevale, these companies who were trying to collect data, but not a lot of that data was actionable. full genome sequence, your full microbiome and.

All that wasn't immediately actionable. No, you're right. I mean, I saw a lot of the patients who went through those programs at Human Longevity Institute in Arival. And there was a lot of actionable data. The problem is that the data was not in the realm of traditional medicine. And so a traditional doctor would look at it and go, I don't know.

What to do with this.

I don't know what to do with this. And they'd send them to me as a functional medicine doctor who knows how to understand systems and networks and patterns in the data. And that was really insightful to me because you can generate all this data, but it's not knowledge.

It's just information.

It's just a bunch of noise. But within the noise, there's patterns and there's data that actually help you decide what to do. 100%. That's the missing piece in medicine right now. And I think, just to kind of piggyback on what you were saying, you're talking about deviating from a medical system that has been focused on a reductionist approach to disease that eliminates all unnecessary variables. It basically ignores all the important data that's in the, in the quote noise, which is actually signal. And this is the opportunity to understand the complexity of human biology in real time. And that, that's what functional medicine for me is. It's really a way of having a container for the complexity of human biology, understanding the body as this complex adaptive system, a network of networks. And when you just kind of think about this, I'm just going to sort of talk, just going to talk about something that's, I think, is going to be relevant for people to understand why what you're doing is so important and why it's, it's so unique. And that is that, you know, when you go to the doctor, you get a lab test and a checkup. and they'll do, I don't know, 30, 40, 50 different, you know, analytes. That's nothing, right? You have literally hundreds of thousands of molecules in your body. You have billions and billions of chemical reactions happening every second. Right. And to think that we actually understand what's going on by just looking at a few numbers is just so silly. And yet it is what we do. And so what you're talking about with your company is to take, you know, massive amounts of data. from massive amounts of people and try to look for the relevant data that teaches us stuff that we never even saw before.

Absolutely. So you're describing what we call in our company the multiomic approach. So I was looking for exactly everything you just said, which was a more nuanced, more complex answer as opposed to just reducing Mark to his, you know, Dr. Hyman to his cholesterol and you to her A1C and the other person to their blood blood pressure. And that's how I found Mike Snyder, the ultimate multi-omic man.

So, so what is multi-omics?

People don't know what that is. Multi-omics is exactly what you were saying. It's multivariate decision making. So, you take in the A1C as one data point. You take the cholesterol as a data point. But you also take wearables as a data point. You also take how much the person is sleeping as a data point. You also take how active they are as a data point, which, I mean, there's no classic medical biomarker for activity, you know. So you take novel markers, you take classic markers, you take essentially all the data that you can get your hands on. I mean, you can go to, you know, genomics, microbiome data, you can go to metabolomics. We are just beginning to understand proteomics. There are a few very promising companies, too, that just went public, Ser and Nautilus, who are going to be understanding the entire human proteome, which is new. It will be revolutionizing biology as genomics revolutionized health for us. So we understand so little so far. But anyway, the interesting thing is that Mike had diagnosed his own diabetes. He has type 2 diabetes. And while Larry had used supercomputers to understand his own data, Mike has applied this to other people. So he had run this multiomic study four years long at at Stanford University where he had taken genomics and microbiome and people's wearables data and their food data and a number of other data and to try to understand how people go from health to disease. So it was kind of love at first sight. We met and we decided to start this through. So I think that what you just said about systems biology, we call it multiomics because I think it's a better representation of what we're trying to do. So Mike and I looked at what was possible and we said, how can we bring this multivariate understanding understanding of human health to everyone, to the masses, not just to the few people that can afford, you know, human longevity, Inc., or could have afforded some of these other Solutions. So we said, what's the least expensive way that we could get started? And we said the basic was basically heart rate monitors who were readily available, like Apple watch and Fitbit, and then continuous glucose monitors, which he had been using for eight years before we even got and he was an absolute, he felt like it was an inflection point in health. We were going from using wearables from fitness to actually for health. And so we started our work and we said, what can we learn from heart rate and CGM? And we quickly learned that- Wait, wait.

CGM is continuous glucose monitor.

So what we realized is that we actually had a missing piece of data. That missing piece of data was food. we needed to deeply understand food. What?

Food is relevant to health? What do you mean?

Yes. Really? Yes. So we thought in order to help.

People, what an epiphany, like Eureka, right?

Exactly. So we were obsessed. So I was kind of obsessed with prevention, use of AI to help us in this new field. And I thought, okay, continuous modalities of data like heart rate and glucose, are going to be superbly interesting for doing machine learning because continuous data, opportunity for machine learning, great. But food is missing. And so we started with trying to understand food and of course the issues that you have with food is that food data is, as you know, food data is imprecise. Even USDA data can be 20% wrong. Food laborers are lacking only grocery items and chain restaurants have food labeling. so you have food journaling is really full of friction, you know, hand entering stuff. There's no passive way, like just wearing something, it just automatically can track your food. Crowdsource data, like data from MyFitnessPal and those are faulty. They have low integrity. So it's hard to machine learn off of crowdsourced data because it's so imprecise. You know, so and then there's a lot of generic advice there. Just like, yeah, just don't eat refined sugar. no refined flour, lots of vegetables, walk 10,000 steps, okay? That will take care of everything. So we realized we need to bring precision to a very imprecise world. So that's how with machine learning we set out to tackle these gigantic problems.

So talk to me about how January AI, your company, is trying to enter this field of AI medicine, which is really not anywhere near clinical application in traditional healthcare right now. But it's happening on the margins, which is where you're working and usually the changes come from the outside and they go on the inside. So I think I'd love to hear your vision of what you're doing with your company and how it can start to address some of these challenges we're seeing with healthcare.

Okay, sure. Well, first of all, let's make sure we're not a clinical solution. So we don't diagnose, we don't treat, we don't manage. We help people understand their bodies and become fiercely self-aware and be able to act on the data that they are provided with in conjunction with their clinical care. But let's talk about a few things. The current standard of care, as you said, is that doctors visiting patients 15 minutes a year, testing people when they complain about symptoms, not before then, use classic markers to manage problems, which are very, very limited. They give them, like you said, reductive approach, give them one marker at a time. They don't think that Mark can handle a high cholesterol and high I1C at the same time. So they give you one, like the biggest problem to deal with. And like, okay, let's just get them working on this. They suggest weight loss is kind of the end all to everything. It's like, you know what, if you just lose 25 pounds, all your markers are going to improve. And by the way, it's really easy. Just go home, no refined sugar, no refined flour, as I said, lots of vegetables, walk 10,000 steps. Eat less and exercise more. Yeah, that's all. Eat less and exercise more. It's all going to work out. The future is looking very, very different. We are looking at continuous health monitoring, continuous health monitoring, and wearables play a role in that. But so does frequent testing and some of the unconventional tests you mentioned, like functional MRIs. And you can see testing companies are just growing like crazy. So the amount of consumer health data outside of traditional clinical setting is growing tremendously. whether it's genomics, microbiome tests, whether it's food sensitivity tests from, let's say, EverlyWell, other companies such as like Ixlayer and other that are now food testing companies. So that's all grown. Wearables data exploding. CGM data will explode. Absolutely. You know about 12 CGMs out there today. They're going to be 47 new CGMs coming online in the next three to five years.

So continuous glucose monitors, right?

That's right. So you will have an explosion of wearables. proteomics will go be transformed completely and we will understand the human proteome for the first time in the next several years. So they, we are- what is the proteome?

Tell people what is a proteome? What is that?

Well, the proteome is basically the, the, the, our full profile, protein profile of humans. And currently we can understand a few proteins we can essentially to sort of understand a few proteins for tens of thousands of dollars. And I think in the future we're going to be able to understand the whole proteome for not many, very many dollars. Kind of what happened to genomics, you know, when we started looking at genomic sequencing, you remember it was $100,000 per person, and it's, you know, went down to $1,000 per person. The whole genome sequencing went down quite a bit, and we have different levels of it. that's been very, very informative for us. But anyway, so the real issue is.

Before you get going, I want to just go a little bit more into the proteome. So for people listening, why do I care about my proteins? And who cares? Like, you have 20, 000 genes, right? You probably have five to seven million variations in those genes, and they're all producing proteins.

That'S their job.

So your body, literally your genes are expressing when they're being expressed are creating proteins. And those proteins are the messenger molecules in your body. And we're, what we're, what you're saying is that we have very little insight into all these proteins and these protein networks and how they work and what they do and how they influence things. And in a very reductionist way, we might understand it, but we really don't.

Understand the complexity of the interaction. And that's what you're talking about figuring out. Yeah, we don't understand basically the function of the, in order for us to understand the function of the cells, we need to understand the proteome. We need to basically understand, it's going to inform disease identification. It's going to tell us so much about our bodies that we don't understand today because frankly it's too expensive and we know so few. And there's quite a few companies that attempt this but there are new methodologies coming in that are going to be transformative. So I'm really excited about those.

So beyond the glucose part, which I think everybody can understand, there's many companies doing the glucose monitoring. It seems to me what January AI is doing is a little bit different in that it's really trying to look at the sort of holistic view of human biology with CGM as a start, but not the whole story.

CGM heart rate and food to start, and then hopefully adding other omics over time. But yes, so definitely you're right. The future, I think people will go to see the doctors for acute diseases in the future. Probably most everyday things will be handled by AI at some point. But, you know, people will go when they have liver disease or something like that, when they have acute disease. And of course we will have personalized health. And you mentioned kind of personalized medicine, personalized health, personalized nutrition. So maybe we can talk about that a little bit. if you're up for it.

Yeah, what does that mean?

For sure. Yeah, exactly. So personalized medicine came about when people were looking at specific therapies for specific cancer. So they started sequencing, for example, the tumor to see what kind of chemotherapy would impact it the best, that could address it the best. So it was like matching medicine to disease, specifically chemotherapy or immunotherapy, with a particular tumor. Precision health is upstream from that. It's way more preventative minded. So we think of, okay, let's collect a lot of data, understand this person's physiology and frankly, psychology, and to then help them kind of know, predict, you know, help them see patterns, basically collect a lot of data and see blips. and trends over time so that hopefully we could say, oh, this person is headed for diabetes. This is what we can do about it. Or this person has hypertension. They don't know yet. Much like Larry Smarr and Mike Snyder self-diagnosed themselves kind of understand, oh, okay, this is where I'm headed. This is what I can do. Precision nutrition is about, you know, foods specifically to keep us healthy, but also address certain diseases. So if you are in if you have ESRD or if you have heart failure, you may want to be on a particular diet.

What is age-related renal disease?

You mean age-related renal disease or exactly. Or if you have IBS, maybe you're on a low FODMAP diet. So you want to have precise nutrition for your situation. If you are in disease, if you are healthy and you want to prevent disease. So, yes. So the future looks to us, future is multiomic, the future is personalized. in the future, people have a lot more data and companies like January will help them make sense out of that data, synthesize the data, analyze the data, and then turn it into simple things like walk this much, eat that. Because why? Because generic advice doesn't work. Personalized advice works. People are infinitely interested in what works for them, not what works for everyone else.

Absolutely. It reminds me of Leroy Hood and the Institute for Systems Biology and his P4 medicine concept, which is preventive, predictive, personalized, and participatory, meaning you have to participate in your well-being. And it's really breaking down the old ideas of disease as we know them. And I think for me, what's really exciting and I'd love to hear how you see AI playing a role and who's doing the work. You want to know my secret for Living a long and happy and healthy life. Well, all I have to do is check out my weekly newsletter, Mark's Picks, where I share my favorite tips for health, longevity, well-being and lots more. Check it out and the link below. So, so really what you're talking about is that the Quantified Self movement combined with the omnics revolution and the superposition of AI and machine learning and big data on top of it is just going to revolution every everything about how we take care of people.

Absolutely. As we develop the omics better, like proteomics and others, as the wearables become more telling, these continuous modalities enable us to do machine learning. And when we do machine learning, we see patterns. And so I do think that, you know, I think China has been very, very smart about, you know, measuring health and trying to understand what we do. And they're able to apply it to to a billion three people and they can solve for health. I believe one of the greatest consequences of China being more advanced in AI than we are is solving for health. And I mean, that's excellent. I'm very, very happy as a global citizen. I want everyone to be healthy. I think that's phenomenal. I just want us to also be running after health in the same way. They're doing the right thing. We need to do the right thing by our population as well.

And so what are they doing differently than we are?

Well, they have a lot of state-sponsored research. They approach a lot of the best scientists in the world. They can set up a lab for you in 48 hours, give you a dozen most intelligent fellows to work for you and to do research. They take winners, as you know, when they know some private sector have a very close public-private sector partnership. If they see a company succeeding, they quickly put their, you know, put their finger on the scale and the company, you know, is propelled into the whole population. So, as you know, they record everything. They have, they have, they have, you know, videos recording the whole population. So technically, you should be able to look at how often people are, you know, sleeping, getting up whenever they're eating, how many times are they eating there? And they have They have, I don't want to say reverse engineered our genome sequencing, but I think that is a belief they are basically all the advances we're making in omic by omic, I think they are picking those up very quickly and taking a much faster path to it than we are. So they're definitely going to have continuous life monitoring. They will have omics, they will have wearables, they will know more about their population. I mean, just look at Mark, Mark, they built a hospital for 1000 people in 10 days. Wow.

It's kind of crazy, yeah, it's true.

And the tracing, and the virus tracing, yeah. I mean, they take, like I said, one of the consequences of AI supremacy of China over US will be solving for health, and we should, hustle. We should run fast. We should solve. I agree.

My biggest honestly, I've been struggling with this for decades, literally, because it's been clear to me that given the massive shift in the paradigm of medicine from a reductionist disease-oriented approach to a systems network approach, aka functional medicine, that there's no way that this can scale unless we have the AI machine learning help and support because the complexity of human biology is just staggering. I mean, there's 37 billion billion chemical reactions every second in your body.

Five billion years of evolution. Absolutely.

How can anybody kind of think through that as a provider in a way that makes a huge difference for people? So, I feel like we're in this moment where I'm sort of waiting around for somebody who's got the money, the resources, the science to actually start to do this. It seems like you're heading on that track. It's not as grand as I would like to see it, which is essentially encompassing this really new paradigm and starting to collect data based on this new kind of network theories of medicine. Because Watson is interesting, but that's the IBM computer that basically went to medical school and can digest all the literature, but it's all from the old paradigm. So, it's like putting a I'd rocket on a horse and buggy. You know, it's like, well, what? Well, we don't want a horse and buggy.

We want a Tesla, you know, or we want. Right. And we need to understand exactly how two people with diabetes are different. For example, someone could have diabetes in their musculature or somebody's it's in their liver. We will be able to understand these differences with AI. I think I agree with you. We're in a, we're a very critical, we're in an inflection point. One, wearables are going exploding as I mentioned. We're in the golden age of machine learning. We can learn a lot from sparse data. You know, COVID accelerated telemedicine. I think telemedicine is here to stay. That is not going away. It unmasks chronic disease. It showed how many people have underlying conditions because people started dying. 40% of people who died had had diabetes or other chronic condition, the cost of healthcare is just not sustainable. So we're spending $3.8 trillion on healthcare today. It's supposed to go to almost $12 trillion by 2040. That's in 19 years. In 19 years, we're supposed to go to 12, you know, 11.8 trillion according to this report that just came out. And only 10% of Only 2.9% of this is going to prevention. Only 2.9% of what we're spending is going to prevention. 10% of it is going to end of life care. It's insane. It's insane. Don't you think those need to be swapped? That's insane. 2.9% for prevention, 10% for end of life care. I think food is medicine is gaining traction. Definitely. I think it's people like, you know, Dari should be minister of nutrition in this country or, you know, secretary of nutrition seriously.

I told him, but he doesn't want the job.

I don't know. Seriously, food as medicine. You know, food is medicine is the medicine we take three to five times a day, you know, and we need a lot more tailored, medically tailored Foods, functional Foods, both obviously natural, but also bread for added Health. So I hope to see government taking a, a much, much stronger step. And also, I think we see a tipping point in consumerization of Health. Right. Just look at the testing companies, Everlywell. Look at retail clinics. You're going to get your healthcare at where you are at. You're not going to go to the doctor's office. You're going to go to Walmart, you're going to go to Target, to CVS.

You'Re going to go to Forward and.

Parsley and One Medical and you're going to go to them. They're going to come to you. Corporate clinics, yes, or they'll come to.

You, or you'll go to the doctor at your work. I think people will get care where and where they want it. So great. So let's talk more about January AI and what distinguishes it from other CGM glucose monitoring companies? And you have an interesting platform called the Season of Me, which is essentially a personalized approach to understanding your body and blood sugar. So tell us, what is the technology using? How's it different? And what is the sort of the framework of how you work with people to drive behavior change? Because that's the hardest part, right? We all know what to do, but most of us don't do it. So tell us about those two pieces.

I would say the most defining difference between us and everyone else in the sector is our ability to predict. It's our AI. That's the biggest differentiator. The fact that we can produce glycemic predictions for people. Like I said, you want to know if you can drink three glasses of Chardonnay or you want to drink three glasses of Chardonnay, all right, ask the AI. What happens to my blood sugar if I have three glasses of Chardonnay? Don't drink three glasses of Chardonnay to find out how it impacts your body. Don't use your body as the place of experiment. Use the-- you do it in silico. Don't do it in vivo. Do it in silico. So let the machine tell you. Why put it through your body? You want to eat this thing you're sitting at your pizzeria and you want to have this pizza, you want to see how it's going to impact you. Look at it ahead of time. Don't look at your glycemic response after it has already happened. Look at it before it can happen. use the machines, not your body to run experiments, make smart choices. So that's the single most important differentiator between us and everyone else. Our ability to predict our AI is our superpower. And in terms of how, like I mentioned, we, the Season of Me program is a curated program that we also have other products in the works that are going to come out. But Season of Me is a curated program. that over 30 days teaches people how to increase their fasting period, how to be cognizant of how much fiber they have in their body to increase fiber, how to manage their calories, how to dial their activity to keep their blood sugar in a healthy range, to become aware of lower glycemic index foods, so, you know, while maintaining. So these are levers we offer them. Some people take up some of those. Some people take up all of those. Some people pick up fasting and nothing else. Some people do all of those. They lose weight. They manage their blood sugar. Their timing range improves. As you recall, A1C has been the kind of golden measure of blood sugar for a while. But that's an average number over 90 days. It's not so actionable. You don't wake up and say, I want to improve my A1C this morning. But timing range, you can manage that. You're like, okay, how am I doing? and that's what January helps you do.

And what are you seeing from the clients that you're using the product and how are their outcomes different and what are the changes they're seeing and how successful have they been at changing their behavior?

They're reporting everything from a lower A1C, lower weight, more energy, fasting, you know, if you don't eat so much so close to your bedtime, you're able to have better sleep, more restfulness, Somebody told us that their focus improved so much they actually sent us a video and they said, feel free to share this with other people. Completely unprompted. They said, I'm keeping my job because of January because I lacked focus so much.

So.

We'Re getting very positive feedback from them. I think we still want to dial our user experience a lot more. We're a science first company. We focused on science first and our research first, and we worried about developing the app and the app user experience and all of that second. We're trying to make sure that we also create a magical user experience for people, and we're working on that now. But very, very positive feedback.

That's so exciting. And how does it work? How do people get connected to January 1st and how do they learn about it?

So they go to our website. Sure, sure. They go to our website, they go through a brief questionnaire to go through telemedicine. They are given a, if they qualify, they are given a prescription for CGMs and they pay and they conclude their transaction on the website. Then they download the app, we ship them what they need for the program and then they download the app, they hook up their their heart rate monitor, whether it's Apple Watch or Fitbit, they hook up their CGM, then they start logging their food, and within four days, voila, they start getting predictions of their glycemic response. We tell them they're spiking foods, what foods are spiking them, what foods are like those foods they could have that wouldn't spike them. We give them what we call activity counterfactuals, like you ate fish and chips. Had you walked 10 minutes, this would have been your response. Had you walked 20 minutes, this would have been your response. So we give them food counterfactuals. You ate fried chicken. Had you had salmon, this would have been your response. We tell them, this is your fasting period. We start increasing their fasting period. We give them education through content about what is insulin resistance, what is intermittent fasting, why is it effective. how does it work on your body? So it's a combination of education, small nudges, and specific behavioral levers such as fasting, calorie management, eating lower GL foods, increasing activity to try to get them to a better metabolic state.

And is it something that, you know, they're able to adhere to that people find sticky and they're willing to sort of do the work around it?

Yes. People come to January. We've tried to simplify certain things over time. They absolutely don't want that. They're like, no, I want to know how much. We're like, Do you want to just give us a picture and a few words? They're like, no, I want to know exactly what macros are in my food. I want to know exactly how much fiber is in my food. You know, we're coming to you for precision. Don't remove the precision. So it's an interesting balance. with precision comes some friction. So a lot of people are willing to do that because they are optimizers. They really want to optimize.

Yeah, and I think we're all going to have to sort of take control of our own health and realize that health doesn't happen in the hospital or the doctor's office. It happens at home. And that all of us being able to be more empowered with these tools, technologies, the sort of decentralized access to our health data and the superimposition of AI and machine learning to help people make sense of it and act on it is just huge. It's a game changer in healthcare. I'm just thrilled that you're doing this work and I'm thrilled that you're putting the science behind it because a lot of times it's easy to do something fun and goofy, but the science part is hard. The 16 million foods you tracked and analyze the ways in which you're integrating all these different data sets, the way you're looking at other analytes and things to do. To me, it's not just a blood sugar monitoring company. It's really more looking at how do we use quantified self metrics and AI to solve our big health crises.

Absolutely. I felt like I had the capacity, the mental space, and the opportunity to look at some really hard problems. And when you have that opportunity, you must take it. So I feel very privileged to actually be working on these and the sort of Frontiers of of health today.

Yeah.

Combining medicine and machine learning and science. Because you know, what we do in a lab takes 15 years to get to the doctor's office. So how do we combine? How do we how do we close the gap? How do we how do we help people become really aware of their state? Those who want to know some people don't want to know. That's okay. Not everyone needs to know, but the ones who want to know how can we empower them, equip them to know, and then how can we help them synthesize the data because it's not human interpretable?

It's right. I think that that's a key message for people to understand that we go to the doctor, we think, you know, the doctor knows everything, measures everything, tests everything. So I went to the doctor, they did my checkup, they did my labs, everything's fine. I know what's wrong. Like, well, that's just like skimming the surface of what's really going on. And I, I mean, that's what excites me so much about medicine today is that we're, we're in this really revolutionary Paradigm of one. the science shifting around what disease is and how health happens in the science of health. And two, the technologies allow us to, like yours, like January AI, to start to kind of dive deep into our own biology and create personalized approaches to, to creating health for ourselves.

So it's super. Absolutely. So our dream is to, to mathematically model all the human functions. We can figure out what's happening, gastric emptying, figure out sort of Wow. How is food moving through your body? We have a lot of dreams. We have a 10-year roadmap ahead of us with very interesting research.

I just interviewed a guy for the podcast, Fred Provenza. He was talking about this work that's being started up looking at metabolomic analysis of people eating different foods, not just in vitro, but actually in their bodies. Absolutely. And say, what happens when you feed someone a grass-fed steak or a feedlot steak? Is it different? What happens when you eat this or that? It's very fascinating and we've really not done that even work, which you think we would have done by now, but it's really and then the question how does it affect different people and all that's going to come into focus with the work that you're doing with AI and understanding how to really disrupt healthcare. So, I'm super excited anyway I can help you guys. I think it's just a tremendous moment in healthcare. and for those listening, you know, a lot of this is still abstract for people, but the truth is with these new metrics, we're able to learn a lot about our bodies. I had the Oura ring for a while, I was wearing it, it was fascinating. When I drank some wine or had some alcohol, I noticed my sleep was screwed up. And I mean, if I wasn't aware of it, or my heart rate variability changed, or, you know, I noticed that when I sort of stopped traveling and being on airplanes three or four times a week, my heart rate variability improved. so my stress response, even though I didn't feel stressed in my head, my body was registering the experience as stress. And that was such an incredible piece of feedback. So I think we really have, we really have just sort of begun to scratch the surface of what we're going to be doing in medicine. And it's not unfortunately, it doesn't really come from the center. I work in healthcare institutions and they do their best and there are a lot of progressive people and really open minds. but it's often challenging with the inertia of the system in terms of reimbursement, what gets paid for, what people are doing. I mean, we know that, you know, food is the best treatment for diabetes, and yet it's not paid for. So doctors do amputations, they do heart surgery, they do lots of medications. It's like bazillion dollars. Diabetes is the most expensive condition in healthcare today and costing America the most money. It's one in three Medicare dollars and probably Two out of three if you count pre-diabetes and all its complications. And yet we don't even treat them with the right treatment that we know works. And we were talking about it just whole next level.

CGMs are covered today by insurance companies only for insulin-intense use population. That's three and a half million people out of 122 million people who are believed to have diabetes or pre-diabetes. But the beauty of AI, again, with AI, you don't have to wear a CGM 365 days out of the year. Payers, employers, put CGMs, use January for your populations with intermittent use. Use it four times a year. Use it two times a year. We build the models. From there on, we predict they don't need to wear it in order to get the benefit. They just need to wear it a few times because their bodies change, they age, they travel, they get pregnant, et cetera. But again, AI solves AI democratizes medicine because we can predict our ability to predict reduces the need for these devices and makes them available and accessible by a much larger population.

Well, thank you so much for the work you're doing and the advance you're creating in healthcare and medicine and the thinking behind this because it's not easy what you've taken on. And for those listening, I think, you know, this whole conversation about AI and medicine and science it's a lot to take in. But I think the bottom line is that, you know, the body is extraordinary complex. We've just been scratching the surface. And with new tools and Technologies of assessment of our biology, both in the traditional Healthcare and also within the disruptive models like January, I were going to be having insights that we never had before. We're going to be able to provide tools for people that they've never had before to optimize and enhance their health and reverse disease. Hey, If you like this video, you're gonna love the next one. Click on it to check it out today.

Dietary guidelines were meant to prevent disease. We got pretty bad. How is that progress gone? How's that working for you? Obesity going up, diabetes going up, heart disease still number one killer, cancer going up. So by any outcome measure, they have been a total failure.