Transcript

[0:00:00 - 0:00:16] Intro: I am the experiment. You can write your own story. Stop trying and don't give the name of fear. Before we get started, make sure you're connected wherever you listen. Follow us on Spotify YouTube. X so you never miss an episode. Okay.

[0:00:16 - 0:00:17] Patrick Lilley: Nice to see you.

[0:00:17 - 0:00:20] Justin McMillen: Nice to see you. Thank you so much for coming out today.

[0:00:20 - 0:00:21] Patrick Lilley: Yeah, glad to.

[0:00:21 - 0:00:28] Justin McMillen: And uh, you are Patrick Lily. You're the founder and CEO of Liquid Bioscience. Sciences or science?

[0:00:28 - 0:00:29] Patrick Lilley: Sciences, yeah thank you.

[0:00:29 - 0:00:29] Intro:

[0:00:29 - 0:00:33] Justin McMillen: Sciences, and that company, what does it do?

[0:00:33 - 0:01:16] Patrick Lilley: So we've been around for about 14 years, and we have a quantitative AI platform that we use in health care. So the notion is that if you can predict patient outcomes, what you can do is you can potentially diagnose diseases, you can predict treatment response. You can explore the the sort of molecular dynamics of drugs and how they're actually working in the body, identify safety and toxicity issues and why the drugs are working, which in turn spawns more potentials for newer treatments, and you can do simple things like if a drug works for 30% of the patients. You can use data to predict which 30% applied it in a very precise manner, and then identify why for the other 70%, doesn't it work? And let's go develop a drug to hit those factors.

[0:01:16 - 0:01:28] Justin McMillen: Got it. Got it. We're going to get into. We're going to get into the weeds with all of those things. Sounds great. But I just want to say the reason that I have you here is because you and I got to spend some time together on the East Coast.

[0:01:28 - 0:01:29] Patrick Lilley: Yeah.

[0:01:29 - 0:01:54] Justin McMillen: With some incredible people, and I was totally blown away. We had dinner together, and you started talking, and you were talking about developing these tests for, predicting, potential SUD substance use disorders, and you had your partners there and, I just couldn't believe it when you were talking about this. You've, you've used in, like, an evolutionary approach to computer science.

[0:01:55 - 0:01:55] Patrick Lilley: Right.

[0:01:55 - 0:02:10] Justin McMillen: So, I mean, you're just blown my mind, and I was like, I have to know more about this guy, and then, and then we did a panel together, and that was pretty interesting, and I think that went well, and then we've since talked. It turns out we live in the same neighborhood.

[0:02:10 - 0:02:10] Patrick Lilley: Yeah.

[0:02:10 - 0:02:41] Justin McMillen: So. Yeah. So I'm I again, glad you're here, and we jumped right into liquid bioscience. But I think it's important to say why you're on here. So I want to explore, first, I want to learn all about what you do, and I want to, and I'm going to dig into it with you because, I think some of it's pretty complicated, and I think with, that brain of yours, I think it can be hard for people to understand. So I'm going to ask you a lot of questions and get into it with you, and then, and then we'll learn together.

[0:02:41 - 0:02:46] Patrick Lilley: Yeah. The good news is, when we're innovating in technology, it's actually helpful to go back to the simplicity.

[0:02:46 - 0:02:47] Justin McMillen: Okay.

[0:02:47 - 0:03:01] Patrick Lilley: And think about a problem in that way, because so many people involved in a field, both of us, probably when we're involved in a field, we see all the details and complexity. It's only when you pull yourself out of it and go back to first principles that you can really innovate and change things.

[0:03:01 - 0:03:06] Justin McMillen: Yeah. That's true, that's true. How did you even how long have you been doing this?

[0:03:06 - 0:03:17] Patrick Lilley: This particular company for 14 years. My first use of evolutionary computing, which is literally evolving solutions in servers to solve problems, was actually in 2001, in the wireless industry.

[0:03:17 - 0:03:18] Justin McMillen: Okay.

[0:03:18 - 0:03:33] Patrick Lilley: And so we use the we use a very simple genetic algorithm to optimize how we updated a phone over the air, which we were the first to do it back then, and then we used it for messaging security in my second company, and then my third company was this one where we're literally building predictive models.

[0:03:33 - 0:03:40] Justin McMillen: How what? Why? So how does a person go? I mean, in your life, how did you get into this? Like, why?

[0:03:40 - 0:03:41] Patrick Lilley: That's a good question.

[0:03:41 - 0:03:41] Justin McMillen: Yeah.

[0:03:42 - 0:05:50] Patrick Lilley: Well, so so my educational background is, is a couple of years of computer science, and then I switched to economics, which is another complicated mathematical system. Right, and then I did my MBA and I specialized in some derivative securities and sort of the game theory of strategy. So I had a lot of quantitative background, and I spent my career in big companies working in every single functional area operations, finance, marketing, strategy, you know, customer service, eventually engineering and so on, and so I got a really nice, broad perspective, and one of the things that happened was, I remember in 1995, I was asked to come out and give a guest talk at MIT, and the internet was a new thing and nobody knew where it was going to go. So we had some really luminary people there, small group of about 25, 26 people, and, you know, the head of R&D at General Motors, the president of the Washington Post, the guy who invented the ATM, you know, some really interesting people, and I gave this little talk on the internet, and where it was going. But on the way there, remember, I was I was reading an article in Scientific American when we were approaching Logan Airport, and there was a guy who had taken Programable chips, and he had discovered this thing in the literature called Genetic Algorithms, and he thought, well, I should try that, because maybe I could program the chips not by explicitly thinking of the program in writing it, but by using artificial evolution to program it, to program itself, to do what I want, and here was the really interesting thing. He had a very simple exercises, and I'm going to put 1000Hz signal in. I wanted to give me a 2000Hz signal out and it worked. So that's all great. So he had colleagues around the world that he wanted to duplicate it so that he could publish and say, yes, the works replicated. Well, it didn't work for any of them, even with his own software, and he couldn't figure out what was going on, and so eventually he called Xilinx, who is the maker of the FPGA Programable chip, and they looked into it and they called him back, and they said, by the way, what's going on here is your chip was in a lot number that has a manufacturing flaw. There's a current leak in one of the gates, and it turned out when he looked into the code, the genetic algorithm found the current leak and was exploiting the flaw to solve the problem.

[0:05:50 - 0:05:51] Justin McMillen: Holy Cow.

[0:05:51 - 0:06:08] Patrick Lilley: And it hit me that, Holy smoke. It can know something the expert doesn't and use it, but it's a fully blind process, just like in nature, and I remember thinking on the approach to Logan, I have to use this somehow commercially, and it wasn't until six years later I found an opportunity to do it.

[0:06:08 - 0:06:11] Justin McMillen: Have you always been this guy when you were a kid. Were you like this too?

[0:06:12 - 0:06:13] Patrick Lilley: Like curious?

[0:06:13 - 0:06:15] Justin McMillen: Well, just I mean, to even think of...

[0:06:15 - 0:06:18] Patrick Lilley: From what my mother says. Yes.

[0:06:18 - 0:06:23] Justin McMillen: Like always into -- Are you are you into science? Are you into...

[0:06:23 - 0:06:54] Patrick Lilley: Yeah, I mean I was, I was, I was into every subject in school. I loved them all, and I used to read a lot of science fiction, particularly the, the harder science fiction that was backed by real science, and that was that was a surprising level of education, but also sort of a way of thinking about things critically, and when you when you see potential futures evolving in the books, then it gives you a sense of curiosity about, well, what could we be doing with technology to make things get to those futures? The good ones at least, right?

[0:06:54 - 0:07:01] Justin McMillen: Geez man, that's amazing. So okay, so you're reading Scientific American. You're going to meet some of like the top in the world in this.

[0:07:01 - 0:07:05] Patrick Lilley: Yeah, and I was such a junior guy at the time. Right. You know, I was 30 years old, you know?

[0:07:05 - 0:07:24] Justin McMillen: You’re speaking at this thing with all these. I mean, that's I think that says something, and so explain to me more so the chip there's a, there's a, manufac.. manufacturing flaw on the chip, and I think what you're saying is that the computer itself is, for lack of a better word, it's thinking and it's using this as part of...

[0:07:24 - 0:07:41] Patrick Lilley: it's just processing inputs and turning them into outputs. Right, and just like any computer does. But he wanted something very specific. It didn't do general computing. It just was a signal processor. But he was doing it as a proof of concept to say, well, could I use genetic algorithms to solve other problems?

[0:07:41 - 0:07:48] Justin McMillen: But how how did the flaw in the chip produce the solution he was looking for?

[0:07:48 - 0:07:48] Patrick Lilley: Ah, okay.

[0:07:48 - 0:07:49] Justin McMillen: Yeah. Explain more about that.

[0:07:49 - 0:10:11] Patrick Lilley: So let's think of evolution in the real world for a moment. So? So no species, no animal, no bacterium. No elephant, you know, other than maybe human beings, and we can speculate about other beings having some level of intelligence. But evolution itself has no mind. It has no consciousness, and its only goal is to propagate the species, to give an advantage to this, to each species, and not at the individual level. This notion of survival of the fittest applies to the species, not an individual. But what happens is you get randomness in the DNA of organisms, plants, animals, whatever they are. Sometimes that randomness there may be a version of a gene that because the environment has changed a little, that let's say that bacteria can reproduce more and it can it's not survival of the fittest, it's that let's say it has the ability to reproduce 10x what the former version of it has. So it quickly outnumbers the old one in the population. That's really how evolution happens right through mutations and eventually sexual reproduction crossover in DNA, you know all of that. But there are chance changes in environment or systematic changes in environment that are exploited by random differences in the variation of our DNA, and if those are good, you know, good by that sense of, okay, Justin can have more kids because he's a half inch taller than Patrick, so his kids will get more apples from the tree. They have three kids each. Patrick's have two. Within a few generations, that little genetic advantage of a half inch in height will mean that you're 90% of the population within, let's say, ten generations, and it's a very simple, small thing, and it's the system is more complex than that. But that's the basic notion. In the same way, if you've got an evolutionary program that's literally got the DNA being the coding for the computer program that goes on the chip. What's happening is there are many versions of that little program that are created by the evolutionary software, let's say 1000 or 1 million different versions. Each of them is tried on the chip, and those small random differences where one is exploiting that particular gate that had the flaw and one that isn't well, that one will start to solve the problem, it will survive, the other ones won't, and it will get used more and more in the population of the algorithms.

[0:10:11 - 0:10:16] Justin McMillen: So in some strange way, the chip being it's sort of like a random mutation of a DNA.

[0:10:16 - 0:10:28] Patrick Lilley: It is, and in our process we're evolving a particular form of simple algorithm, and we're using two different kinds of mutation. We're using sexual reproduction and mating, and we're using migration.

[0:10:28 - 0:10:29] Justin McMillen: before I'm I just want to go

[0:10:29 - 0:10:30] Patrick Lilley: yeah it's a lot.

[0:10:30 - 0:10:45] Justin McMillen: Scientific American thing. So so as you're reading this in your head, is the article written in such a way where it's clear that this thing mimics DNA and even a more real way because of the malfunction? Or did you draw that conclusion from reading the article? Was the article like you said?

[0:10:45 - 0:11:01] Patrick Lilley: I think at the time I didn't understand, like there is a version of sort of digital DNA. I didn't know how a genetic algorithm achieved it. In practice, I understood the mechanism, but not the implementation, and so I wouldn't say I knew that a string of bits was the digital DNA. Right?

[0:11:01 - 0:11:02] Justin McMillen: Okay.

[0:11:02 - 0:11:20] Patrick Lilley: But it was very clear to me that the program is blind and it operates on its own, and it's trying a bunch of things, and it's guiding itself toward the objective rather than just a bunch of random tries that because of the fitness pressure, use of the algorithm survives because it does well, it doesn't survive if it doesn't.

[0:11:20 - 0:11:39] Patrick Lilley: So just like in the real world, the bacteria survives If it gets passed the medicine, we give it, right? And so it what was clear to me was this blind program with no intelligence found something the human professor didn't even know about, exploited it and used it to reach the goal that he set for it.

[0:11:39 - 0:11:41] Justin McMillen: That's incredible.

[0:11:41 - 0:11:45] Patrick Lilley: And that was enough, right? Right there to say, man, I have to look into this.

[0:11:45 - 0:11:53] Justin McMillen: And this is this is incredible, too, because, I mean, we're in the age of, natural language processing models, which are totally different.

[0:11:53 - 0:11:53] Patrick Lilley: Yeah.

[0:11:53 - 0:11:55] Justin McMillen: And you talk to me about this a little bit before.

[0:11:55 - 0:11:55] Patrick Lilley: Yeah,

[0:11:55 - 0:12:01] Justin McMillen: But it sounds like so, so, so there was evolution. How do you say evolutionary...

[0:12:01 - 0:12:03] Patrick Lilley: evolutionary computing.

[0:12:03 - 0:12:05] Justin McMillen: Computing. That was happening before you.

[0:12:05 - 0:12:11] Patrick Lilley: Yeah. In 1969 John Holland invented genetic algorithms, and and it's largely been ignored.

[0:12:11 - 0:12:20] Justin McMillen: Interesting, and genetic algorithms are I don't know how you explain that simply. I mean, you've already talked about evolution, but can you break that?

[0:12:20 - 0:12:23] Patrick Lilley: Let's think of an algorithm very simply. So think of a bread recipe.

[0:12:23 - 0:12:24] Justin McMillen: Okay.

[0:12:24 - 0:13:02] Patrick Lilley: It's a set of ingredients and a set of steps over time, and attributes like how much heat in the oven or whatever. So you mix things, you let them sit, they rise. You, you know, you do these various things, any anything where you do a series of steps to get to an outcome is an algorithm. The selection criteria for a clinical trial or an algorithm. The way you decide how to treat the people that come to your facilities is technically an algorithm. Now, there's there's judgment around it, right? You have both objective and subjective means that you use because that's who you guys are, and it's effective. But they're all algorithms in the sense that there is some knowledge and there's a set of steps to achieve something.

[0:13:02 - 0:13:12] Justin McMillen: Do you have to have in order for it to be an algorithm? Do you have to have more information with each step that then takes you to the the subsequent step or...

[0:13:12 - 0:13:45] Patrick Lilley: Or you have progress. You don't necessarily need more information because again, let's think of the bread recipe. I can lay out all the ingredients on the counter, the spoons and the wheat and, you know, the yeast and everything else. I can have the oven in the kitchen. Everything is there. So I'm gaining nothing new in terms of information or things. But as it proceeds, stuff is happening. Chemical reactions, things, rising bubbles forming. Then we put it in the oven and cooking causes other reactions to happen. So yes, there is something being gained it at every step toward the objective of having a nice warm loaf of bread that we smell and taste, right?

[0:13:45 - 0:13:55] Justin McMillen: Gotcha. So a series of steps in some sort of order. Produces some set outcome, and those series of steps along the way is an algorithm. Right. Okay. So then take that to...

[0:13:55 - 0:14:12] Patrick Lilley: And a football player is an algorithm okay. Right. All of these are algorithms. So the question is okay so what is it. What does it really mean to evolve an algorithm. Well if you think so, your DNA is the instructions for building Justin.

[0:14:12 - 0:14:13] Justin McMillen: Sure.

[0:14:13 - 0:14:15] Patrick Lilley: And only you. Right. Unless you had an identical twin.

[0:14:15 - 0:14:16] Justin McMillen: Yeah. No, not that I know of.

[0:14:16 - 0:14:35] Patrick Lilley: And and then you're subjected to the real world and there's a certain you know, how many kids do you have? How do you thrive in the world or not, etc.? And how much do you reproduce? And you thereby contribute your DNA to the species in proportion to you know, how well your DNA encoded you for the environment in which you live?

[0:14:35 - 0:14:35] Justin McMillen: Got it.

[0:14:35 - 0:15:44] Patrick Lilley: Okay, just like bacteria evading drugs or being killed by drugs, there's an environment in which every organism lives. In the same way. If you have a computer algorithm, which is, let's say, a simple series of math steps saying, oh, well, let's take height and weight and some other things and try to predict, is this person going to get type two diabetes? That that simple set of math steps is an algorithm. Now there are many things known about every patient. There are many different kinds of math or many different ways to combine those. So if we create millions of algorithms out of different kinds of math and different things, we know about the patient's, some of those are going to be predictive, some are not going to be very good at it. Just like the population of any organism in the world, they're going to be good for the current environment or not, and the consequence of inside the computer, if an algorithm is better than others, those die and it survives and it reproduces, literally reproduces. I'm not making a metaphor. I'm saying literally the algorithms are reproducing. They make a clone of themselves which undergoes mutation, or they mate with another algorithm that was also good

[0:15:44 - 0:15:45] Justin McMillen: That’s insane.

[0:15:45 - 0:15:46] Patrick Lilley: It's wild to watch it work.

[0:15:46 - 0:15:56] Justin McMillen: Yeah. That's so if so you basically just turn everything loose, you throw in a bunch of different variables, and then it's trying to get through the steps.

[0:15:56 - 0:15:56] Patrick Lilley: Yeah.

[0:15:56 - 0:16:02] Justin McMillen: Which is the steps is really like you're, you're mimicking evolution. So it's the trajectory of the species.

[0:16:02 - 0:16:02] Patrick Lilley: Yeah.

[0:16:02 - 0:16:03] Justin McMillen: Yeah. Or in a way I guess?

[0:16:04 - 0:16:04] Patrick Lilley: It is.

[0:16:04 - 0:16:11] Justin McMillen: And then it's going through steps and then what works, gets the benefit of being able to live and what doesn't work dies.

[0:16:11 - 0:16:12] Patrick Lilley: Right.

[0:16:12 - 0:16:14] Justin McMillen: But that's computation.

[0:16:14 - 0:16:14] Patrick Lilley: Correct.

[0:16:14 - 0:16:17] Justin McMillen: Okay. So but instead of DNA which is computation anyway.

[0:16:17 - 0:17:22] Patrick Lilley: Yeah. We have a string of bits which is the digital DNA of the algorithm, and that string of bits is oh, the first instruction uses addition and it uses these two variables height plus weight. Right, and then it puts it into a memory location. That's that's a step in an algorithm. The next one might use cosine. Who knows. Right. There's a whole set of different math that could be used, and here's what's interesting is nowhere have I said that the human beings running this system are determining which variables you use, which math is used, which numbers are used, it's emerging on its own, and the real benefit of that is you have no biases, no assumptions, no biases. So it's a wonderful complement to the real experts in the field because it's discovering things. It's looking places they wouldn't consider to look, and then handing them a very focused area to focus their big brains and expertise on. But if you look at the way industry works, particularly in diagnostics and pharma, let's take prostate cancer. Perfect example. So we have a prostate cancer we just created, and we're going to build a blood test.

[0:17:22 - 0:17:22] Justin McMillen: Test for it.

[0:17:22 - 0:17:51] Patrick Lilley: A test for it. Yeah, and we're going to find therapeutic targets for new treatments for it as well. So one of the interesting things about that is if you go search the last ten years of literature on Google Scholar for prostate cancer diagnostic biomarker, you're going to get tons and tons of papers, 15,500 papers in the last ten years, all peer reviewed, all declaring victory, on we found protein or RNA or other biomarkers to diagnose prostate cancer. There are no good tests on the market.

[0:17:51 - 0:17:51] Justin McMillen: Really?

[0:17:51 - 0:19:15] Patrick Lilley: This is true for every disease. I mean that very there's a handful exceptions, right? Exact science has their Cologuard test, which in its second generation is a pretty good test. The accuracy is better than the first generation, and it's based on a series of biomarkers that's very rare. Very few diseases have those kinds of multiple biomarker tests that are highly accurate. People don't realize that most diagnostic tests and processes today are highly inaccurate. The point I was making, though, is that the experts in the field who write those papers are wrong. Why are they wrong? They're wrong not because they're not intelligent, not because they lack knowledge, but for a very simple reason, and here's where we'll use a nice, straightforward analogy. The reason is they assume that when a molecule on the body is involved in a disease, that measuring it would help you diagnose the disease. In other words, it would be discriminative between people with or without the disease. We can separate them into groups. But let's take your car for for a moment. So let's say that you're driving a combustion car, a traditional car, and you start getting a banging noise in the engine and you think, man, I've got to take it to the mechanic. But, you know, cars, and so you're going to make some comments to the mechanic about where he should look. Right. But let's say you don't really know cars. You just think you do.

[0:19:15 - 0:19:15] Justin McMillen: Like everyone.

[0:19:15 - 0:20:22] Patrick Lilley: And you know that one of the most important things for a car is where the rubber meets the road. Literally, the air pressure in the tire. So you tell the mechanic, you know what? Tire pressure is super important. It's involved in the functioning of the car. I want you to use the tire pressure measurements in the four tires to diagnose that banging noise, and the mechanic looks at you like this has nothing whatsoever to do with the engine. The assumption, and I think a lot of researchers in medicine are making is that, again, if a biomarker, if a molecule is involved in a disease, it can help diagnose. But that's the same assumption. If you notice that the air pressure in the tire could diagnose the specific problem in the engine, I don't think they realize they're making this assumption. So what we're trying to do is say, look, guys, you're unbelievably intelligent, knowledgeable. Let us first look where you won't look because you have biases and predispositions, and you focus your PhD thesis in a very narrow area. We're going to cast a super wide net. We're going to find what's actually involved, and discriminative. Then we're going to show it to you, and that will help you focus your wonderful time and attention on the right areas.

[0:20:22 - 0:20:24] Justin McMillen: You deal with a lot of egos.

[0:20:24 - 0:20:25] Patrick Lilley: We do, we do.

[0:20:25 - 0:20:26] Justin McMillen: And I bet I'm sure you do.

[0:20:26 - 0:20:28] Patrick Lilley: And we have to set them aside to right?

[0:20:28 - 0:20:34] Justin McMillen: Of course. But but I mean, the way the way you said that even like, you're intelligent and I mean, that's a huge issue right now.

[0:20:34 - 0:20:34] Patrick Lilley: Yeah.

[0:20:34 - 0:20:42] Justin McMillen: Because here we have super computation and AI and all these things that are clearly there's just no human being capable.

[0:20:42 - 0:20:43] Patrick Lilley: Right? Yeah.

[0:20:43 - 0:20:49] Justin McMillen: Producing the same they don't have the same perfect memory and they don't have the same ability to process information.

[0:20:49 - 0:20:49] Patrick Lilley: Yeah.

[0:20:49 - 0:20:53] Justin McMillen: I'm so this is now I'm starting to understand this more.

[0:20:53 - 0:22:06] Patrick Lilley: Yeah. Can I hit the large language models? Because there's a good contrast right now. So, one of the challenges is that the large language models, ChatGPT and all these that have become so famous in the last year or two, they are not about math. They are about words and sentences and paragraphs and concepts. Okay. Where are they learning about medicine? They're learning from the same publications that I just referenced. Those publications are very good at describing the functions of each molecule in the body. They are not good at saying those molecules are discriminative or can be used to diagnose. Right, and so this is I'm red green colorblind. Right. So this is about 1 in 11 Western European men are red green colorblind. Do you think I sat down and taught my kids their colors? No. That would have been a mistake. Right. So you find somebody who doesn't have the color blindness? The point is that those AI tools are using the very literature that isn't correct about what is discriminative, what diagnosis is disease, and there's another problem when you think about prediction, we think about prediction is I'm going to make a prediction like it's going to rain tomorrow, but there's a lot more to it.

[0:22:06 - 0:22:39] Patrick Lilley: As you know, in your business unit, you're trying to predict what's going to work for this person to get them to the state I want them to be in after interacting with us right when I when I was growing up here in Southern California, you know, Vin Scully was the voice of the Dodgers on the radio, and man, I used to sit with my dad by the radio and listen to the games because they wouldn't put them on television when they were home games, and he said high fly ball to right field okay. So we can picture that. Do those words tell the outfielder where to run to catch the ball.

[0:22:39 - 0:22:40] Justin McMillen: No.

[0:22:40 - 0:23:11] Patrick Lilley: Not at all. The outfielder does have in his brain because he's played baseball since he was five years old. He actually has a mathematical formula in his head embedded in his neurons about the arc of that ball, and he runs to where he knows the ball is going to go. He's actually using a mathematical predictive algorithm in his brain. The words can't do that. The math in his brain can, and so what we're saying is the the large language models can't make predictions that are specific.

[0:23:11 - 0:23:29] Justin McMillen: He's not doing that. They just keep just taking in words and like almost it's almost like an echo chamber because it's just continually reiterate or distilling down or reorganizing the same thoughts, and you you're talking about an entirely different there's a process.

[0:23:29 - 0:23:30] Patrick Lilley: Right.

[0:23:30 - 0:23:38] Justin McMillen: It's a, it's a, it's a, that's, it's a system for sure. But it's yeah, it's computation. It's not the same thing. Let's just say that.

[0:23:38 - 0:23:42] Patrick Lilley: Yeah, and I can tell you the distinction one point further, if you like.

[0:23:42 - 0:23:43] Justin McMillen: Please, please.

[0:23:43 - 0:25:24] Patrick Lilley: Yeah. So just think of in the past how did scientists work. So Galileo, for example, you know, one of the things he studied was what is the formula for objects falling, and, and he also took these, these spheres and he rolled them down, these incline planes, these wooden, you know, wedges of different slopes, and he recorded how long they took and what angle, you know, effect had on that, and he ended up with a mathematical formula from those observations that describe the, you know, the acceleration under gravity depending on the slope of these things. So he took observations of the circumstances, the height of the wedge, the angle of the wedge, and the Earth's gravity itself and the outcome, which is how long did the ball take to go from the top to the bottom? So those circumstances and outcomes he used and God knows how he did it without calculators compute. I mean, the people in the past, it's astonishing to think how brilliant they were instantly. Right. So so he came up with a formula that says, okay, this is how these behave when they roll, and it's a mathematical formula. Why is that important? And by the way of formulas and algorithm also it's important because then you can use that formula anywhere and you can say, oh, if I need a ball to go this fast in this far, what angle do I need and what height do I need? You can derive it from the formula so you can make predictions. You can test the predictions, you can build new systems because you have a model for the real world physics. So our goal in building this software was to say the algorithms that we evolve, they have to be models of the real world. So we have to take in let's say 50,000 RNA measurements per person.

[0:25:24 - 0:25:52] Patrick Lilley: And we have to say here's 48 different kinds of math and logic functions that might characterize what's going on, plus a bunch of numbers. But what we're after is to create a mathematical formula or algorithm that predicts the patient outcome, and that model can be used for prediction, for simulation, for diagnosis, and for insight into the mechanism of the disease. That was our goal is to model the world like the past scientists did, but on problems that no human being could ever deal with because there were too many variables.

[0:25:52 - 0:25:59] Justin McMillen: What is this? Are you a are you sorry that I don't know this, but are you a pioneer in this space?

[0:25:59 - 0:26:04] Patrick Lilley: I don't know about a pioneer because John Holland invented it. In 69. There was a fellow named John Kozy

[0:26:04 - 0:26:06] Justin McMillen: Was he using it for for these...

[0:26:06 - 0:26:23] Patrick Lilley: Very primitive, no very primitive genetic algorithms? There is a real luminary in the field, a guy named John Koza, who was at Stanford. I think he might be retired now, and he built tree based evolutionary computing, which I used in my second company. What we used in my first company was like John Holland's original algorithms. This one...

[0:26:23 - 0:26:25] Justin McMillen: What is tree based. What is that?

[0:26:25 - 0:27:04] Patrick Lilley: Oh, the idea of sorry if you if you think of an algorithm as a series of steps that work up in a tree like I have two variables height and weight, and I add them in a node that says addition is the node, and those are the two inputs. You can imagine a tree of all the the variables, right. We do it somewhat differently. We use what's called a Turing algorithm built out of Alan Turing's work. Before World War two. So because Turing algorithms are the only algorithmic form that is rigorously proven to be able to mimic any system in nature. No other algorithmic form can do that.

[0:27:04 - 0:27:14] Justin McMillen: How in the hell did you go from reading that article to did you already have your brain wrapped around this kind of stuff, or did you have to learn it?

[0:27:14 - 0:27:18] Patrick Lilley: Pieces here and there and then? Yes, I had to learn a bunch, and we keep learning all the time.

[0:27:18 - 0:27:23] Justin McMillen: Right, and then. But did you have did you start bringing in where you like? I need to find the guy. That's the masters.

[0:27:23 - 0:27:28] Patrick Lilley: Yeah. Or you find courses online or you. Yeah. You reach out and you talk to people and you know.

[0:27:28 - 0:27:46] Justin McMillen: So but but and and and so you know, I'm, I'm a business person too, and so there's an entrepreneurial side to me. So are you zoomed out and like have a vision of but we can take this and apply this and it can have a use case and these. Yeah, and so then you're like how do I do this, and now you're studying. Yeah.

[0:27:46 - 0:28:00] Patrick Lilley: Justin the beautiful secret to this innovation. We have 130 trade secrets and patents around the core technology itself. Right. Most of them are trade secrets because they're fairly simple things. The beautiful truth of it is that most of them are borrowed from some other domain.

[0:28:00 - 0:28:01] Justin McMillen: Sure.

[0:28:01 - 0:28:36] Patrick Lilley: And I'll give you a brief example. So one of the things that's important is when we evolve these algorithms on a portion of the patients, it's very possible for them to be so fitted to those patients that they don't work on new patients. So imagine all the prostate cancer patients we had in the data set in in the training group, let's say a third of the patients we used to train the algorithms in evolve, let's say all the prostate cancer patients wore blue shirts, and that was in the data, and none of the non-cancer patients had blue shirts. Well, the algorithm would pick up on that and it would say blue shirts are a diagnostic for prostate cancer.

[0:28:36 - 0:28:37] Justin McMillen: Okay.

[0:28:37 - 0:29:24] Patrick Lilley: That's overfitting, right? That's that's exploiting something by chance. It shouldn't be there. Well, if you then subject that algorithm to another third of the patients or beyond, it will fail because that the chances of blue shirts corresponding to prostate cancer are pretty much zero. Right. So when you solve that problem, how do you pick the right algorithm out of an evolved population so that it holds up on new patients? That's the central problem we spent three years on at the beginning of the company. How do we make sure these algorithms work on new patients so we don't kill people? Right, and believe me, it's the thorniest intellectual problem I've ever tackled. Well, one of the components of it is you have to look at the errors and, and, and how they sort of look mathematically, the errors for each algorithm.

[0:29:24 - 0:30:50] Patrick Lilley: Now, what's really interesting about this is that the errors and the way they're distributed mathematically corresponds to something called Brownian motion, which is the way smoke particles or gas particles move in a room. What's really interesting is why are they called Brownian motion? Well, because in the mid 1800s there was a botanist named Brown, and he was famous for going everyplace and doing these nice sort of colored drawings of, of all the plants and naming them, and the famous example was Clarkia pulchella and I have an original lithograph from, I don't know if it's a lithograph or drawing from him. I bought it at auction because he took the pollen from it and put it in the microscope, and he noticed the pollen was moving. You know, where they put it? In between two slides with water and the pollen was vibrating, and he thought the pollen was alive initially. Later it came out that know what's going on is the water molecules, because there's heat in them, are jostling the pollen and under a microscope you can see that, and that motion became known as Brownian motion because his name is Brown. Well, what's interesting is in 1905, Einstein wrote four papers, published. One of them, which he won the Nobel Prize for was about Brownian motion and the equations for particle based Brownian motion and the transfer of heat in physics, and so his mathematics describe the jostling of the particles that Brown observed.

[0:30:50 - 0:31:38] Patrick Lilley: Now, here's what's interesting. It turned out that if you sort of march forward to World War Two, the British were having a problem, which was the German planes were getting through too much, and they were they were, I think, expanding about 2500 anti-aircraft shells per plane to bring them down. It was unsustainable, and they did a number of things. They adopted the American proximity fuze. They got radar. But one of the things they did was they had a mathematician who came from elsewhere in Europe, Norbert Wiener, and he took a he made a model of the German planes, physics and said, okay, every quarter second or so, here's the cone of probabilities of where they are, and the Gunners need not to try to lead the plane, but put a couple of 2 or 3 or four shells into that cone of probability, and he radically reduced the number of shells. Well, his math was Brownian motion based.

[0:31:38 - 0:31:38] Justin McMillen: Oh my gosh.

[0:31:38 - 0:32:21] Patrick Lilley: Now then you go forward to finance and derivative securities options, futures and so forth, and there's a very famous now known as slightly incomplete model. But it was very groundbreaking at the at the time, the black Sholes option pricing model, and it's an equation that allows you to model a call option or a put option on stocks, the right to buy or sell it at a future time, and that's based on the potential movements of the stock going forward. Well the movements of stock are Brownian motion, so that equation has elements that that look like particle physics and look like Einsteins Brownian motion equations, and you can carry this forward into statistics. So it happened to be that I was exposed to a number of things like this.

[0:32:22 - 0:32:22] Justin McMillen: Because of your economics?

[0:32:23 - 0:32:51] Patrick Lilley: Right, and just reading a bunch and working on derivative securities in graduate school and so on, and talking to my former statistics professor who said, you can use the information in errors to help your model as long as you do certain things, and it struck me that that's one of the things we could do, and so we tested and we found. So this is what I mean by we're borrowing ideas from these great people that came before us and repurposing them to a totally different domain, because they turn out to be applicable.

[0:32:51 - 0:32:52] Justin McMillen: That's insane.

[0:32:52 - 0:33:00] Patrick Lilley: There are people before us Justin that the work they did was staggering, and again, it is there to be farmed. It's it's it's gorgeous.

[0:33:01 - 0:33:21] Justin McMillen: It's so mind blowing though just to think of your head space in starting a company or having a vision to even begin to apply this information, and then what were you always obsessed with? All of the founding great thought leaders of the Western or the Western world?

[0:33:21 - 0:33:29] Patrick Lilley: Yeah, my grandfather used to give me biographies of people and all sorts of people. I mean, Wilt Chamberlain and Einstein and just different people.

[0:33:29 - 0:33:34] Justin McMillen: Was Brownian motion the first one where you were like, we can use this to apply to this, or

[0:33:34 - 0:33:42] Patrick Lilley: It’s hard to say whether it was the first one, but it was one I was pretty familiar with because of the things I had done academically.

[0:33:42 - 0:33:51] Justin McMillen: And then also, were you. So you're going to work every day and you already had a company at the time, or was this the beginning of a company?

[0:33:51 - 0:34:05] Patrick Lilley: So, we had gotten out of our former company and I'd been consulting for a couple of years. I was working for a guy in the in the overseas remittance space, helping helping him help people send money home to their families, basically.

[0:34:05 - 0:34:05] Justin McMillen: Okay.

[0:34:06 - 0:34:21] Patrick Lilley: So I was consulting to him on a variety of things, for a while, and my co-founder and I ended up thinking, okay, well, it's time for us to do something again because we had started the prior two companies and he said, let's, let's use evolution again, but let's build a broader platform, right?

[0:34:21 - 0:34:23] Patrick Lilley: Wow.

[0:34:23 - 0:34:31] Justin McMillen: That's crazy, and so then what does it look like? So is this like, you guys are hanging out in a room with a whiteboard and I'm serious. Like..

[0:34:31 - 0:34:32] Patrick Lilley: Yeah, yeah.

[0:34:32 - 0:34:35] Justin McMillen: Is it, you're just like, spitballing ideas and uses.

[0:34:35 - 0:35:04] Patrick Lilley: Yeah, yeah, on the phone and in person. Not so much in office. I mean, our first company, we started at the at the Starbucks down by the Ritz-Carlton. Right. We were there, like, every day, and then talking until 2 a.m., you know, most nights, and so it was a very similar pattern for us, and you know, we architected, designed. I would look into, okay, what's the market? Where should we go? How do we get funding or the right people in the network. Right, and it all sort of, you know, builds until there's a moment where you get a check and you, you start actually doing.

[0:35:04 - 0:35:08] Justin McMillen: So you did, you did like a raise and you did that, and yeah, it's like this is what we're going to tackle, and you went..

[0:35:08 - 0:35:13] Patrick Lilley: Yeah, we went a couple of years without any pay, and you know, you know the drill.

[0:35:13 - 0:35:14] Justin McMillen: Full bootstrapped. Yeah.

[0:35:14 - 0:35:20] Patrick Lilley: And then we got some seed round investment from some guys in Indiana. Wonderful guys, and one of the local angel groups.

[0:35:20 - 0:35:29] Justin McMillen: And they did they know. Were you able to clearly define what the hell was that you were trying to do or you didn't tell them and they just listen to you and go, sounds like you.

[0:35:29 - 0:36:04] Patrick Lilley: Yeah, I think the the technology was it just sounded like magic to people and crazy, but it was more that we had a track record as entrepreneurs building highly scalable software, doing certain things, like our first company with the over the year firmware updates is now in use, and probably two 3 billion cars and and phones and and other devices because it formed the basis of the standard for mobile device management that’s used today. So there was at least a track record of returning good dollars to shareholders, and I think there were a few people that did that on us.

[0:36:04 - 0:36:12] Justin McMillen: Yeah, you bet on the horse. That's one before, right? Yeah. So but you at this point in your career, you're like evolutionary computation. Did I say it.

[0:36:12 - 0:36:13] Patrick Lilley: Yeah. Perfect.

[0:36:13 - 0:36:15] Justin McMillen: That's your that's your thing.

[0:36:15 - 0:36:15] Patrick Lilley: Yeah.

[0:36:15 - 0:36:18] Justin McMillen: You’re like I am. It's a good hammer. Let's hit some.

[0:36:18 - 0:36:18] Patrick Lilley: Yeah.

[0:36:18 - 0:36:41] Justin McMillen: And you're like there's a ton of ways we can do this, and so you're running with it and you and your partner build, you know, creating patents, and you're, you're creating new innovations and you're applying it first to cell phone technology, not cell phone, but updating cell phone software updates. Then you start to move over to health care. Meanwhile, you're pulling in math and science from all over the place. Every time you encounter a problem, you are giants.

[0:36:41 - 0:36:42] Patrick Lilley: Yeah, yeah.

[0:36:42 - 0:36:53] Justin McMillen: Einstein, and and so in each scenario where you did that, are you taking the time to really learn all of the theory and like, how did you even know where to go look?

[0:36:53 - 0:37:44] Patrick Lilley: Yeah, I don't mean to imply that I become an expert in all these areas. Yeah. I mean, I always joke that I'm a dilettante, right? And we used to have this, this, you know, thing I would put on the whiteboard in front of new employees, particularly the assumption of maximum ignorance. Right. Let's just assume we know nothing, and so I would learn enough, and, you know, my my son asked me this the other day, do you read all these books? Well, no, I read the parts I need right? Later. I might read them, but I read the parts. I need to understand that concept enough to be able to repurpose it and see how it maps on to what I'm trying to solve, and I'll do the Wikipedia, you know, wormhole thing and dive into all the concepts until I get some idea, and again, those sometimes you get into an idea, see how people are approaching a problem. You don't borrow so much of their idea, but the way they thought about it, apply that way of thinking to what you're doing, and you get to an answer.

[0:37:44 - 0:37:45] Justin McMillen: How do you feel about intuition?

[0:37:45 - 0:38:04] Patrick Lilley: Oh, I think it's it's real. It's, you cannot assume that you can deductively logic your way through everything. Intuition is a huge deal, and that's why I like to read widely and talk to people from a variety of professions and interests, because you never know where your idea is going to come from. You have no idea.

[0:38:04 - 0:39:10] Justin McMillen: Yeah. That's a as you were talking, for some reason, I just had in my head, I imagine you driving and being like, I got it and it showed it happened more of being like, hey, here's how I think we could try to do this. Let's, let's mess with this right now, and then playing with a different idea. So I'm starting to get all of this now. So, so you're applying evolutionary computation to solving all of these problems and that carries you forward, through lots of different business ventures, and, and then that leads you to where you are now, which is kind of where we started. We started we went right in. But yeah, this stuff, it is important to get a sense of kind of, how things work, and I think what's really interesting about you is that, you're clearly super bright and somehow you've met there's a lot of people in this world who have good brains. I mean, you're an outlier for sure, but but translating the ability to think the way that you have into a moneymaking endeavor, like, do you ever think about, like, how in the hell did you figure that out?

[0:39:10 - 0:40:32] Patrick Lilley: Yeah. So it's it's interesting. I'll get to the moneymaking in a moment, too, because there was a particular conversation I had with one of my daughters about this. But we all have to eat. Yeah, and we all have to amass resources to provide security for our families, and so, you know, just as you can, you know, bring ingenuity to problem solving, you can bring ingenuity to this problem of, how do I take that and turn it into income for my family, but also for another reason, which is to amass the resources to do greater good. I had a conversation. My oldest, second, oldest daughter is 29 and when she was in sixth grade, I remember she asked me, why don't you just give away the technologies that you produce? And I said, well, it's interesting because if I did, what I could do is I could establish a foundation and I could raise money every year, donations, build something, give it away for free, and then next year, start the process fresh again and try to raise money and a foundation. On the other hand, if I build a product and I charge people, big companies and others for the benefits that we produce, I can make a profit, and by the second year I might be able to, let's say, work on two diseases instead of one. By the third year might be eight diseases instead of one. So by the time I have a ten year track record, I've worked on vastly more through compounding than I could at the beginning.

[0:40:32 - 0:40:33] Justin McMillen: There’s nothing wrong with making money?

[0:40:33 - 0:41:56] Patrick Lilley: Not at all. I'm a libertarian economist by education, right? And I think there's another thing, though, that I know you share. So, you know, it's interesting there was something several months ago that I think reflected a large reason about why we do this. There was a, a picture I saw that to me was very, very disturbing, not only for the sort of surface reason, but for a reason I'll tell you. There was it was a picture of, London during World War two, and there was a little girl dressed nicely, 2 or 3 years old, and her legs were under some rubble, and she was up on her arms and you could see she was trying not to cry. But the pleading look in her eye, straight into the camera was, you know, as a father, it really hits you, and the thing about it was, you can think of the horror of what Hitler was doing. You can think of the horrors of war, you can think of all of this. But you know what hung around with me for 2 or 3 weeks and I, before I finally process it. Why the hell wasn't the photographer helping that girl? Why was he taking a picture instead of helping her? And one of the reasons I do what I do is because I feel that when we have certain skills, there's a lot I don't have. But what I do have, I want to apply it to help people at the biggest scale possible, and I think that's some not only a moral imperative, but it's truly evil not to do so.

[0:41:56 - 0:42:07] Justin McMillen: Yeah, right. Yeah, we share that for sure. I have a weird thing for since I was young too, where I'm drawn towards like where there's the most amount of pain. Yeah.

[0:42:08 - 0:42:17] Patrick Lilley: And particularly when people are helpless. Yeah, right. You see, they can't pull themselves out of the situation. So you have an obligation to do what you can.

[0:42:17 - 0:43:01] Justin McMillen: Yeah for sure, and that's a. Yeah. When people's brains don't work right and therefore their thinking isn't working right. So for me it's like I have. Elderly people, children like anything in the world that I could possibly do to help them, and then people that are impaired that can't think or their their mentally have no their faculties aren't there. It's like, how do you expect this person to make a good decision and not to destroy themselves? And so those are like my areas of focus, and then outside of that realm, I'm a big, big on personal responsibility. Like, yeah, and actually that feeds into the way that our models are for treating addiction. But

[0:43:01 - 0:43:31] Patrick Lilley: Well, one of the things it's interesting that we talked about before when we met was one of the biggest problems I see in medicine and health care and health and wellness and all of it is how do we get people to do the right things for themselves, and all of us struggle with this, right? I get up in the morning and I look at the workout equipment in the garage, and my mind is already starting to make excuses. Well, I gotta look at an email. I need to do this and you have to do the right thing. That's something that I'm very curious about in your endeavors how you do that.

[0:43:31 - 0:44:03] Justin McMillen: Well I'll tell you. We were talking about this this morning, and I had a call with these PR people to do some ad campaign, well not ad campaign, but some campaign to talk about this project we're doing, and I was explaining that if we saw so someone, someone said there's a lot of stigma around addiction, and I said, well, yeah, of course, because people are making poor choices, and one of the people the call is looking like, and I go, that's the truth, right? They're bad choices. Like, anybody who says they're not bad choices is wrong. But I think you got to go a little bit farther in than that and go, why?

[0:44:03 - 0:44:04] Patrick Lilley: Yeah.

[0:44:04 - 0:44:24] Justin McMillen: And and if that part of their, their humanity that's supposed to make the choice is messed up, that's the problem. So you got to think of what are all the things that come together to make a choice to go left or right, and arguably one of the most extreme versions of that is substance use disorder, because people are making a choice to take a drug that they know will kill them.

[0:44:24 - 0:44:24] Patrick Lilley: Yeah.

[0:44:24 - 0:44:31] Justin McMillen: It's hijacked everything about them. So if you solve that to any degree, you're never going to solve it perfectly. It's why I'm obsessed with it.

[0:44:31 - 0:44:31] Patrick Lilley: Yeah.

[0:44:31 - 0:44:37] Justin McMillen: But if you can make any dent in that, you can take that information and apply it to everything, you know, eating disorders.

[0:44:37 - 0:44:37] Patrick Lilley: Right?

[0:44:37 - 0:44:59] Justin McMillen: The fact that we live in a time of abundance and that everybody has more than they need, and we all have to understand how we contend with the resources we have. Right. It's great we solved it, right? If our ancestors showed up, they'd be like, Patrick, you did it. Look at everybody. You guys are all everyone's fat. Everyone's like, got everything they need. No one's getting chased. That's a great thing. But now we have new problems.

[0:44:59 - 0:45:00] Patrick Lilley: Yeah.

[0:45:00 - 0:45:06] Justin McMillen: So how do we manage this relationship with our dopaminergic system and... Yeah, this is stuff.

[0:45:06 - 0:45:12] Patrick Lilley: And your point is exactly relevant to our comment about evolution, right? The environment has changed. We changed it.

[0:45:12 - 0:45:12] Justin McMillen: Yeah. Yeah.

[0:45:12 - 0:45:15] Patrick Lilley: Right. Civilization changed it.

[0:45:15 - 0:45:28] Justin McMillen: Yeah. Now it's mismatched which is a whole thing. But I want to I want to keep going on this evolutionary computation because I think what's interesting is everyone wants to know where the world is going.

[0:45:28 - 0:45:29] Patrick Lilley: Yeah.

[0:45:29 - 0:45:53] Justin McMillen: And I think you're standing you're in an interesting position right now, what you said about you know, I saw it in your eyes. You're talking about this girl, and I was thinking about this when I was driving around today listening to your a different podcast that you'd done. I thought, My God, if this guy is doing all the things that he says he's doing and I know you are, it's just you just need more time and resources and people, because..

[0:45:53 - 0:45:55] Patrick Lilley: That’s been the struggle. Absolutely.

[0:45:55 - 0:45:59] Justin McMillen: Yeah. Because you could you could find answers to all of these questions.

[0:45:59 - 0:46:35] Patrick Lilley: Yeah. Well, the journey was very funny because, you know, when we when we took the first few years to really be sure the technology was working and we did it on a lot of projects where human lives weren't really involved in the same, you know, way, you know, finance, marketing, optimism, optimization, things like that. We then started doing academic collaborations and we would walk in the door with this magic thing nobody ever heard of. If you look, the number of professionals involved in neural networks and deep learning and the current kinds of AI is millions of professionals worldwide, and the special interest groups maybe have 5 or 600 people in my field total worldwide.

[0:46:35 - 0:46:35] Justin McMillen: Wow.

[0:46:35 - 0:46:50] Patrick Lilley: It's like a fraction of a percent, okay. It's just been ignored. Even though it is the most powerful computation engine known to mankind because it is what nature uses to compute millions and millions and billions of organisms.

[0:46:50 - 0:46:51] Justin McMillen: It's how we got here.

[0:46:51 - 0:49:10] Patrick Lilley: It's how we got here, right? I mean, in fact, I had a conversation with IBM years ago and they said our system is based on the human brain, that's why it's superior. I said, well, what produced the human brain? You know, so so the point was, is we started to do these. We had to very carefully build, you know, academic free work as a track record to then show companies to do analytic service work for Big Pharma, for Mayo, for the FDA. We forecasted Covid for the government, etc. and as we built that credibility, then we got to a place where we could at least attract investors, not into our own company, but to spin off or partner with new JVs that would pursue individual disease diagnostics or clusters of diseases, and that sort of joint venture model has been very good. It's uneven. Some companies do well, some don't. It better than, I would say, a standard VC portfolio. But we're investing the intellectual property there. But as you pointed out, the number one impediment is getting the resources, the right people, the right capital, and if you look, the the nature of scientific and medical advancement is in fits and starts and there's tons of resistance. The guy who won the Nobel Prize, Japanese fellow who’s name I forget, for how T-cells actually work. He spent a very long period of time fighting against the dogma and outright criticism and public humiliation by existing scientists who said, no, no, no, no, it works this way. The guy who in the mid 1800s, what was his name? Angle vice or something like a Swiss obstetrician. So he figured out he noticed that when he washed his hands before delivering babies, the mothers wouldn't die as much, and he found this out, and he used it. He washed with, like, a chlorine solution, and he told all the other obstetricians and surgeons and everybody else, it was 25 years before it became common practice, and he was so aware of the burden of suffering. He died in a mental institution. Right. That's not where I want to go. That's not where you want to go. But my point is, the resistance is tremendous, even when you have the track record, and because we spent that time at the beginning of the company solving the reproducibility problem, we've had no prospective failures at all in any of the projects we've done in terms of the biomarkers working, and that's interesting.

[0:49:10 - 0:49:11] Justin McMillen: When you say biomarkers working.

[0:49:11 - 0:49:13] Patrick Lilley: I mean, they work in a test, they work on new patients.

[0:49:13 - 0:49:27] Justin McMillen: I know what you're talking about, and we didn't really we said in the very beginning, but I want to make sure the audience gets it. So we're now talking about you took the company to practically apply evolutionary computation to healthcare and testing diagnostics. Right?

[0:49:27 - 0:49:34] Patrick Lilley: Right, right. Finding which molecules to measure in the blood that you could combine mathematically to diagnose. Do you have the disease or not.

[0:49:34 - 0:49:35] Justin McMillen: And so if you had...

[0:49:35 - 0:49:36] Patrick Lilley: A simple blood test.

[0:49:36 - 0:49:45] Justin McMillen: If you could do whatever you wanted in this, you had this is your tool to do it. You could predict most any disease before it happens. Well, before it happens.

[0:49:45 - 0:49:48] Patrick Lilley: I, I want a separate prediction from diagnosis.

[0:49:48 - 0:49:49] Justin McMillen: So diagnosis.

[0:49:49 - 0:49:50] Patrick Lilley: Yeah, we have had...

[0:49:50 - 0:49:51] Justin McMillen: Early diagnosis.

[0:49:51 - 0:49:59] Patrick Lilley: And we have had some success in prediction as well. But but most people don't do longitudinal studies the right way. So it's the data is not there but diagnosis, absolutely.

[0:49:59 - 0:50:00] Justin McMillen: Got it.

[0:50:00 - 0:50:04] Patrick Lilley: And prediction of treatment response when we have the right data.

[0:50:04 - 0:50:15] Justin McMillen: And and one of the huge advantages is that you can predict, and so when you say success rate, you're not family meaning that you're you're you're not getting false positives or false negatives.

[0:50:15 - 0:50:53] Patrick Lilley: So it's not just accuracy. So I'll give you an example. So so last year one of our clients mine's bio out of Germany in the US they were claiming colorectal cancer. We provided them sort of better math for their biomarkers to improve the accuracy of their test. They then said, okay, do you want to work on pancreatic cancer together? And I said, yes, but we'll also do the biomarker discovery for you. So we'll look at the the RNA in blood in pancreatic tissue, find the right biomarkers, and then you develop them into a test will license them to you. So we found the set of biomarkers. We licensed them to mines. They recently just a few weeks ago announced they did an independent validation study, and the accuracy was equal or greater to then what we did when we discovered.

[0:50:53 - 0:50:54] Justin McMillen: Wow.

[0:50:54 - 0:51:50] Patrick Lilley: So it reproduced, and that's been true of everything we've done because we spent the time at the beginning of the company. So we have that really nice track record. But just in the challenges I walk in the door and I say, Justin, imagine you didn't know me and you didn't know that Vahan invited me out to that that conference in D.C. I'm just some guy on the street. You meet with a software company in Orange County, and I say, well, I can I can diagnose, I can find the biomarkers to diagnose substance use disorder in like, yeah, listen, man, I know the disease. It's not that simple. It won't be, you know, but I actually have the track record and we've done it, and we did it across six substances and found what's in common without respect to the substance itself. I have that track record, but the first time I said it to you, it still could appear as a one off. You're not looking at the fact that we built the Henry Ford assembly line for automobiles and systematized a reliable process of biomarker discovery and test development.

[0:51:50 - 0:51:51] Justin McMillen: I still need to learn more about it because I'm.

[0:51:51 - 0:51:52] Patrick Lilley: Yeah.

[0:51:52 - 0:52:20] Justin McMillen: Yeah, you have to teach me more about that because I, I the nature of addiction is so interesting because the real idea is like you say, it's it. I don't I don't know if we should get into this particular section of things right now, right now. But we could, it's about making a decision. So let's say we have every kind of drug of choice out here, and you're saying, I know the biomarker that's going to predispose somebody to addiction, to crystal methamphetamine.

[0:52:20 - 0:52:22] Patrick Lilley: Yeah.

[0:52:23 - 0:52:33] Justin McMillen: The only way that I'm going to become addicted to crystal meth is if I choose to pick it up and put it in my body. So can you predict whether or not I'm going to make a choice to do that?

[0:52:33 - 0:53:03] Patrick Lilley: Yeah. We have a bit of an open question in the biomarkers right now. So you met Stuart Gitlow who who's the psychiatrist on staff at ignite, our partner in substance use testing. So when we look at the biomarkers and the nature of them, some of the biomarkers are related to impulsivity. Some are, you know, things you would kind of expect. Some are related to inflammation. So some of them appear to be the effects of using a substance, but not the particular substance, just the results generally of being addicted.

[0:53:03 - 0:53:04] Justin McMillen: Okay.

[0:53:04 - 0:53:33] Patrick Lilley: But others appear to be related to predispositions, which is exactly what you're asking, and because these data sets were not longitudinal, they were just snapshots of people with and without substance use disorder. With one exception I'll get into in meth. It's very difficult to say whether those biomarkers would predict that you have a predisposition. I think they would, and I think Stu thinks they would as well, because of the biological nature we discovered after the math gave them to us.

[0:53:33 - 0:53:40] Justin McMillen: Yeah. There's an intersection between psychology and there's there's I think you could do it.

[0:53:40 - 0:53:40] Patrick Lilley: Yeah.

[0:53:40 - 0:54:35] Justin McMillen: I think you could. But it would be, for example, somebody who has is more sensitive. Right. So there are certain genes related to sensitivity. So we talked about this before dandelion versus the orchid. You take that and then you combine a traumatic upbringing. Then you could say that this world is painful for this person. So their baseline level of just basic comfort is pain, whether it be psychological or physical. So there's a constant state of dis, you know, discomfort. So then you have and you have that person next to somebody who doesn't have that, and the two of them use some sort of substance for the first time. One person goes from their baseline, which is relative comfort or comfort to euphoria. The other person goes from total discomfort to feeling normal to euphoria. He distances farther. That person's more likely to want to use that drug again.

[0:54:35 - 0:54:36] Patrick Lilley: Yep.

[0:54:36 - 0:54:39] Justin McMillen: That would. So you could find the stuff that we relate to.

[0:54:39 - 0:55:17] Patrick Lilley: And so what's so interesting about both of you and Steven Lloyd talk about this exact combination of genes and environmental influences and events and trauma. Right. So what's interesting about it is both of you have tremendous experience with people undergoing the condition and people who don't, and you you mentioned intuition. Well, there's there's probably not a math formula in your brain that says, oh, Bob is going to do this, or Pete is not going to do this, but your intuition when you meet people, I bet you anything versus mine, for example, is going to be highly enhanced with respect to predicting whether somebody is likely to become addicted, right?

[0:55:17 - 0:55:20] Justin McMillen: For sure, it's to a point of being haunting. Haunting.

[0:55:20 - 0:56:05] Patrick Lilley: Yeah, exactly. Now, what's interesting is what we have discovered by doing this work, and we've done now 320 projects across more than 60 diseases. So the patterns become clear about the kinds of things we're able to see, and what we are able to see is that data about human behavior, data about the genes, data about RNA and proteins floating around right now in the blood. All of those can contribute to predictions. They all contribute different levels, and some are better at discriminating. Today. But what's really interesting about it is given that trauma creates emotional response, has physical effects, trauma and its after effects are going to be reflected in the biology of the patient.

[0:56:05 - 0:56:09] Justin McMillen: Yeah, they're going to have sympathetic nervous system dominance. They're going to have more cortisol in their blood.

[0:56:09 - 0:56:09] Patrick Lilley: Right.

[0:56:09 - 0:56:11] Justin McMillen: Chronically stressed, and you can measure all of that.

[0:56:11 - 0:57:45] Patrick Lilley: Right. So it shouldn't be surprising that we can discover biomarkers that not only tell you do you have substance use disorder or are you likely to develop it, and here's the interesting study of the six data sets we used to discover the substance use disorder biomarkers. One was a methamphetamine data set, and there were three classes of patients people who were never never users, people who were former users and diagnosed with MUD. You know, methamphetamine use disorder, and people currently with methamphetamine use disorder, and these people that are former were several years out of it. Now, remember that the biomarkers we discovered were across all six substances. So we were not finding metabolites in the body of methamphetamine itself. So the biomarkers diagnosed and we said, okay, we're curious. Let's do the analysis a couple different ways. Let's put the former users with the current and say as a group, how accurate is the discrimination between them and people who never used, and then because they're former, let's assume they don't have substance use disorder anymore. Let's put them with the the controls who never use which accuracy is better. Well, turned out the accuracy was better when you grouped the two MUD and former, which implies that the biology of the patients, those two groups of people was more similar than the former users, was to people who never used. One conclusion you could surmise from that is okay well, that either means that there are long term effects of substance use disorder on the body that we're detecting, and they're just persistent.

[0:57:45 - 0:57:45] Justin McMillen: It's possible.

[0:57:45 - 0:57:53] Patrick Lilley: Or there was a baseline difference between the biology of those people and people who never used, and we're detecting that, and it might be both.

[0:57:53 - 0:58:20] Justin McMillen: It probably is both I think that the the baseline that the I think there's a genetic there's genetic factors. For example, the DRD2 and DRD4 that I've talked about. There's a lot of literature around that. If you have less sensitive dopamine than your basic level of satisfaction is going to be lower, that amphetamine creates 1,000% increase in dopamine in the brain. That's going to feel a lot better for some people than others. Yeah, that's so interesting. I

[0:58:20 - 0:58:47] Patrick Lilley: So now you see why it's so important to us to say we have a perfectly agnostic mathematical method for discovery, but we desperately need the experts, not only the scientists, but the people who deal with the people with the condition. Because, you know, it's so easy. In science, we talk about the ivory tower to never understand what it's really like, and we've all sat on the couch and told the NFL team what to do next. But that's not like being on the field.

[0:58:47 - 0:58:52] Justin McMillen: Yeah, I think I mean, I certainly have biases for sure. I mean, I can't not.

[0:58:52 - 0:58:54] Patrick Lilley: Well, you have to because you've been practicing successfully, right?

[0:58:54 - 0:59:33] Justin McMillen: Yeah. It's I mean I have my my own thoughts and there's is a lot of intuition. I'm just so excited to learn whether where I'm wrong and where I'm right, and I want to beat up all of these ideas to get to some higher form of truth, which is where Vahon and I have, like, bonded over that idea. What I'm so interested in, though, is there's this there's the future of health care. As far as I can tell, everything is moving towards, computers and computation just as a broad term, AI and everything in between is certainly going to be a part of all the future of health care without it, without a doubt.

[0:59:33 - 0:59:34] Patrick Lilley: Right.

[0:59:34 - 0:59:43] Justin McMillen: And you've got this way of diagnosing things and ideally early. Right. That's if you're seeing a biomarkers you can see this stuff.

[0:59:43 - 0:59:43] Patrick Lilley: Yep.

[0:59:43 - 0:59:44] Justin McMillen: Way before it's detectable in us.

[0:59:44 - 0:59:53] Patrick Lilley: Yeah. We just announced autism with Ignite Biomedical. The same folks that we did a SUD, and we're detecting autism with biomarkers in kids well under two years.

[0:59:53 - 0:59:53] Justin McMillen: Really?

[0:59:53 - 0:59:55] Patrick Lilley: Yeah. Yeah.

[0:59:55 - 0:59:57] Justin McMillen: Do you have a degree of accuracy on that?

[0:59:57 - 1:00:04] Patrick Lilley: It's a it's greater than 90% sensitivity and specificity, and we replicated across 13 data sets 13 different patient groups.

[1:00:04 - 1:00:08] Justin McMillen: How early could you could you pull that from a from a fetus.

[1:00:08 - 1:00:14] Patrick Lilley: Well it's interesting because one of the data sets is, cord blood.

[1:00:14 - 1:00:15] Justin McMillen: Okay.

[1:00:15 - 1:01:15] Patrick Lilley: Now is the big debate how much of autism is genetic versus environment? And there's two kinds of environment. There's the motherly environment of the womb, and then there's post-birth. Well, the cord blood presumably reflects genetics of the baby. Plus the, the you know, the nine months in the womb. So there is some form of environment. You know, RFK is talking about acetaminophen. I you know, I don't agree. But, the question is how accurate can how accurately can we diagnose autism from cord blood? Because they detect they had to wait. They took the cord blood. Then they waited for the kids to get older, and when they got diagnosed or not, after a certain period of time, then they mark the database with yes or no, and our job was to basically predict, are they going to develop autism? What's interesting is the accuracy on all the other data sets is above 90% on that one. It's around 75. The input, and let's think about this. The best accuracy you can get 100%.

[1:01:15 - 1:01:15] Justin McMillen: Sure.

[1:01:15 - 1:01:20] Patrick Lilley: The worst accuracy is not actually zero. It's 50%. It's a random guess.

[1:01:20 - 1:01:20] Justin McMillen: Yeah, sure.

[1:01:20 - 1:01:49] Patrick Lilley: Okay. So the difference between 50% and where you are all the way up to 100 is how much is genetics in the mother's womb. So it looks like about half of autism is a combination of genes and the mother environment. Now that could be all genes or all motherly environment. We know there are some genes involved, so it's not all mothers, but fully half of autism is outside the womb. It's environmental. So that's a that's a nice exercise.

[1:01:49 - 1:01:58] Justin McMillen: But Patrick, could you could you have the certain genes set up to let's say that every human being is going to get exposed to the same environmental condition.

[1:01:58 - 1:02:05] Patrick Lilley: Right? yes, in your some genes will predispose you that environment giving you autism in some way.

[1:02:05 - 1:02:16] Justin McMillen: Yeah. So it's like the genes could be there, but perhaps it's because I'm not going to we're not going to go down the vaccine route. But perhaps one particular vaccine that everybody. Yeah, it's you can buy.

[1:02:16 - 1:02:44] Patrick Lilley: Yeah. Let's just say lead paint, right? Because nobody is going to get triggered by lead paint. Right. So, you know, I was born in 65. In all likelihood, I was exposed to lead paint growing up. He was just used in various dishes and houses and everything else, and somehow I didn't develop autism. That may mean that either I don't have sufficient genetic components to be triggered by the lead or that there was something else that helped me not get it right, but there are definitely both right there.

[1:02:44 - 1:02:50] Justin McMillen: When you when you find this stuff, do you publish it and make it public, or is it like your trade secrets?

[1:02:51 - 1:03:30] Patrick Lilley: We do publish some papers, but peer reviewed papers are extremely hard to get published these days when you're using something like evolutionary computing, because the reviewers don't understand it at all. What we've done is we've got a paper out now that explains the technology, and it's for the TNF inhibitors we licensed also to Ignite Biomedical, and that we're using now as a foundation for the credibility for other papers. So we can just refer to it. Anything new. It's like I said, the dogma in medicine in science, what's really funny is why should our stuff be considered new when life has been around for, what, 4 billion years on Earth and evolution's been operating it?

[1:03:30 - 1:03:33] Justin McMillen: Yeah, it shouldn't be, but I don't I don't think people get it.

[1:03:33 - 1:03:42] Midtro: If you'd like to see more of this content, please subscribe to us on our channel on YouTube. You can follow us on Spotify or on X.

[1:03:42 - 1:03:55] Justin McMillen: I mean, I, I'm the average person, can't even like even natural language processing models. Like, we're at a point now where most people can't wrap their heads around the concepts that are driving moving forward.

[1:03:55 - 1:03:59] Patrick Lilley: Yeah, and that's understandable, right? I can't throw 105 mile an hour pitch.

[1:03:59 - 1:03:59] Justin McMillen: Right, right.

[1:03:59 - 1:04:00] Patrick Lilley: Right, so...

[1:04:00 - 1:04:20] Justin McMillen: But but I think I think it could get there. I think I think it starts with the potential outcomes. Like if the public sees that, that if you had free reign, you could do whatever you want. You could develop a test for everything. You could be testing people from an early age for all sorts of different sicknesses and illnesses, and early, early detection is the key to survive.

[1:04:20 - 1:04:34] Patrick Lilley: Right, and we can do genetic testing, right? We've worked on genetic data sets as well. It doesn't need to be RNA or proteins, doesn't matter to us as long as the data are measurements or categories we can use it.

[1:04:34 - 1:04:49] Justin McMillen: So the average person sitting here listening to this, if you had the keys to the kingdom, you could produce tests, they could be tested, insurance pays for it, and they would know way ahead of time if they have any kind of the big scary cancers, cardiovascular, I mean, yeah, how many?

[1:04:49 - 1:05:33] Patrick Lilley: I mean, well, we have the ones that we have that are being commercialized by partners right now. We're in pancreatic cancer, breast cancer, lung cancer with a with a mayo clinic. We have a joint venture in kidney cancer, and not only diagnosis but staging and recurrence. So yeah, it doesn't, and we've done Alzheimer's, Parkinson's, ALS, we're about to do frontotemporal dementia, which nobody wants to develop drugs for, because when you recruit, you accidentally pull in. A lot of patients with other dementias should are biologically different. So the drug doesn't look like it works. Right. So I would love to take the diseases and just say biggest prevalence to least prevalence is in many in parallels. We can just start going down the data sets and handling them. I would love to do that.

[1:05:33 - 1:05:36] Justin McMillen: That's amazing. Is that how you've chosen what you've gone after? So far?

[1:05:36 - 1:05:49] Patrick Lilley: Yeah. Is it big? Are people suffering right. The girl again, you know, are they suffering because of this? And are there people who have an economic interest in solving the problem that can partner with us, bring resources. Right.

[1:05:49 - 1:05:53] Patrick Lilley: And so then you produce the test that lets people know.

[1:05:53 - 1:05:59] Justin McMillen: But we're now we're only talking about diagnostics right now. There's also the application of this to discover

[1:06:00 - 1:06:00] Patrick Lilley: New treatments.

[1:06:00 - 1:06:13] Justin McMillen: Yeah. Yeah. So let's we'll move from diagnostics to treatments. But I want to also see if I can explain the evolutionary computation idea in a simple way, and I'm going to oversimplify the hell out of this.

[1:06:13 - 1:06:13] Patrick Lilley: Yeah. Good.

[1:06:13 - 1:06:38] Justin McMillen: But you create you essentially create an evolutionary system of computation, computation being a bunch of processing, parts of a computer that processes information, in the way that information flows through that computer is that it tries things and things that work move forward and things that don't work die off.

[1:06:38 - 1:06:38] Patrick Lilley: Right.

[1:06:38 - 1:06:39] Justin McMillen: Just like evolution.

[1:06:39 - 1:06:51] Patrick Lilley: In my.. So I have one of my daughters works for me. She actually does our literature, scientific research. She's 26 and, you know, she's, she had like four patents by the time she was 19. She’s a brilliant..

[1:06:51 - 1:06:52] Justin McMillen: Nice.

[1:06:52 - 1:07:30] Patrick Lilley: Brilliant person. But, what's really interesting is when she was really little and we started this, she said, dad, when you describe the technology, it's like little math people, and the idea is each one of the algorithms is like a little guy with some math formulas in his head that predict things, and there's a million of those guys running around and they're running around a patients predicting things. Do you have cancer? You don't. The more they're right, the more they survive and they reproduce, and so those people are literally dying off. If they don't predict well, they're surviving and having their own mathematical kids. If they do well, you can really think of it like evolution. It it really is...

[1:07:30 - 1:07:31] Justin McMillen: That's a really beautiful picture.

[1:07:31 - 1:07:32] Patrick Lilley: Yeah.

[1:07:32 - 1:07:34] Justin McMillen: Now I'm picturing the table being covered with a bunch of little geniuses.

[1:07:34 - 1:07:48] Patrick Lilley: It's literally like what happens in nature. It is not a it's not metaphorical. We have literally produced evolution inside the computer for these little organisms, whose sole job is to predict what happens to the patient.

[1:07:48 - 1:08:07] Justin McMillen: So I'm gonna go really out there for a second and say, could could you take somebody if you had enough data about a human being in the way they lived, could you could each human have their own little version of this that's able to constantly through the data gathering, provide like alongside of them.

[1:08:07 - 1:08:08] Patrick Lilley: Yeah.

[1:08:08 - 1:08:08] Justin McMillen: Like as they’re growing up?

[1:08:08 - 1:09:58] Patrick Lilley: So think think about it this way. So, you know, when you go to Amazon.com, there's a set of things they present to you because of your history and what you've done on other sites and so forth, and we all think of those as being absolutely unique and personalized at the same time. You might actually fall into a group of 1.2 million men worldwide that have the same interests and habits. I don't know how many it is, but with respect to shopping on Amazon, you might be very, very similar to them. So you have the appearance of it being individualized, even though it isn't in the same way. What we see is in you need to understand our algorithms. When they come out, they're not like these giant neural networks with billions of tokens. The equation to predict response to treatments for inflammatory bowel disease is a simple algebraic equation with two variables and like three math functions. Right. That we licensed to ignite. So if somebody is going to take a drug for inflammatory bowel disease, there's two inputs the dosage and a particular RNA measurement in their blood. They're combining that simple formula, and it says yes or no. This drug is going to work for you. It's very simple. It's not complicated at all. What that means is that underneath all of the complexity and all the individuality, there are common patterns to diseases, and human beings work in similar ways inside our biology. Well, of course we do. We're the same species. If we didn't have that kind of order and lack of chaos, we'd just be puddles of goo on the floor. Right? There's real order here, and remember, I talked about Galileo. He's looking at observations of the real world and trying to uncover the math formula that describes the real world system of gravity and the spheres and the the the the inclined planes, in the same way we're trying to uncover the law of physics inside you and every other person with your afflictions.

[1:09:59 - 1:10:06] Justin McMillen: Wow. It's mind blowing, and it's the key to personalized medicine and precision medicine.

[1:10:06 - 1:10:35] Patrick Lilley: Yeah, because then we can take each patient and we can say, okay, he's male, he's 40 years old. He's had this history with these drugs. He's got this disease and he's got these six biomarker measurements, feed those inputs into this model, and it says, okay, infliximab is going to work for him. Entyvio is not going to work for him. This other drug may work for him, and now the physician and the patient have choices where they say, okay, now we know that these are the viable choices and these are the ones we can eliminate and they can make the decision.

[1:10:35 - 1:10:43] Justin McMillen: So is this why isn't this widely supported so that all doctors are using this and insurance companies are demanding it?

[1:10:43 - 1:10:44] Patrick Lilley: Well because it's...

[1:10:44 - 1:10:45] Justin McMillen: Or is it moving in that direction?

[1:10:45 - 1:12:17] Patrick Lilley: Remember what I said about the way the industry approached this was they assumed their domain knowledge would lead them to the right biomarkers for tests. There's more than a 99% failure rate in papers trying to reproduce their findings on biomarkers. Further, this notion of being able to measure multiple RNAs and proteins in a single test reliably and in a low cost fashion is relatively new. It's over the last, certainly less than a decade, and the regulatory and payer processes are not used to biomarker tests with multiple molecules in them. So there's not that many that have been approved. Exact Sciences was a, you know, a pioneer leading the way in this in this particular area in colorectal cancer. So, you know, there the world is just not used to the complexity of the test, even though it's actually not as complex as they think. The minute somebody hears math or algorithms, their eyes glaze over and you can see the fear in them. Right? But everything is an algorithm. Even the simple choice of, oh, Justin, your LDL is over 140. We need to give you a statin. That's an algorithm leading to a decision. All I'm saying is you might have six measurements of different molecules. A little more complicated math. Not much, but it's packaged as a blood test that tells the physician yes or no. So I don't think it's going to be that much longer. When you see a whole set of tests. We know that we've got several tests coming out in the next couple of years that are very much like this. Their RNA or protein tests. But, you know, things take time in medicine.

[1:12:17 - 1:12:18] Justin McMillen: Yeah.

[1:12:18 - 1:12:35] Patrick Lilley: And the by the way, that the insurance companies and the government are extraordinarily slow, I think the FDA is very slow because they're worried about having to be careful. The payers are slow because insurance companies don't have a lot of incentive to be fast. Right.

[1:12:35 - 1:12:39] Justin McMillen: Well I think the insurance companies could be highly incentivized here.

[1:12:39 - 1:12:53] Patrick Lilley: Well like you said, if you can detect early what treatment decisions would you make that lower a lot of downstream costs, and then what to them is an ancillary benefit into you and me as a primary benefit is the reduction of the suffering in the meantime.

[1:12:53 - 1:13:00] Justin McMillen: Yeah. I mean but I mean I think early early detection could cut so much health care spend.

[1:13:00 - 1:13:41] Patrick Lilley: You know what's interesting there. When we first started doing work in Alzheimer's years ago because we've done 40 projects in Alzheimer's, is that when we started the work in Alzheimer's, the prevalent view among consumers was, if I have Alzheimer's, I don't want to know, and the reason they didn't want to know was there were no treatments. But here's the thing. If there are treatments under development and you haven't been diagnosed, you can't get into the clinical trial. If you don't know you can't do proper estate planning, you can't warn your family. You can't do any kind of advance planning. It's amazing to me that people think this way, that they don't want to know information because I'd rather know, have the burden of the knowledge and figure out, okay, what do I do about the problem?

[1:13:41 - 1:13:46] Justin McMillen: Yeah. Have you tested yourself for everything?

[1:13:46 - 1:13:46] Patrick Lilley: Not yet.

[1:13:46 - 1:13:47] Justin McMillen: That's a personal question.

[1:13:47 - 1:13:54] Patrick Lilley: Yeah, I is, as you can imagine, I take all the normal blood panels and stuff every six months, and I monitor and. Yeah.

[1:13:54 - 1:13:56] Justin McMillen: Yeah, you said you do carnivore. You're still doing that?

[1:13:56 - 1:14:26] Patrick Lilley: I'm very close to carnivore. I'm one of those lean mass hyper responders. So my LDL spikes okay, and I'm very comfortable with it being 130-140 as you've seen the recent studies. That's great. But it went to close to 200, and that's still okay from a mortality care perspective. But I wanted to drop it down, and so I have maybe another 20 or 30g of vegetables or something a day to bring it back down to 130. But other than that, yeah, I've been paleo since 2015, carnivore for over a year. With this recent addition back.

[1:14:27 - 1:14:30] Justin McMillen: You had fit in in our community over here. Yeah, everybody's on some kind different...

[1:14:30 - 1:14:46] Patrick Lilley: By the way, a guy you should chat with sometime is right here. Michael Rose. He was the distinguished professor of evolutionary biology at UC Irvine, and he's got a I won't call it a longevity company, but it's a longevity company that he's starting, and he has a podcast. He probably want to talk to you as well.

[1:14:46 - 1:14:47] Intro: Yeah, we should do each other's, that’d be great.

[1:14:47 - 1:15:33] Patrick Lilley: But he's a brilliant speaker and a brilliant guy, and he was you know, whenever you see people talk about dietary things, it's always, oh, it's lectins and it's this or lectins or you know, Gondry and there's there's always a good logical story. But then you ask for the evidence. Not there, and sometimes people have some evidence, but they don't have a theory about why, and so you're wondering, every time I argued with Michael and he said something. Right. Because I'm prone and I don't mean argue, I mean ask questions. Yeah, he had both the evidence and the theory, and I would say, well, wait a minute, why can't I eat beans anymore on paleo? And he talked about the evolutionary adaptations out of Africa and all this, and he'd say they're pro-inflammatory, but where's your evidence? And he walked me over to a poster in his lab and oh, yeah, one of my students did this three years ago, and

[1:15:33 - 1:15:33] Justin McMillen: That's awesome.

[1:15:33 - 1:15:42] Patrick Lilley: You know? And we analyzed his data on longevity with his fruit fly experiments, which are astonishing, cleanest and strongest signal I've ever seen in biology.

[1:15:42 - 1:15:43] Justin McMillen: What's he doing?

[1:15:43 - 1:15:51] Patrick Lilley: So he's he's now retired from UCI very recently and he's creating a company basically related to health spin.

[1:15:51 - 1:15:54] Justin McMillen: What how does he making fruit flies live longer.

[1:15:54 - 1:16:38] Patrick Lilley: So this is very interesting. So this the way we met was by chance. But we're using computers to do artificial evolution. He's evolving large 1000 fruit fly populations replicated 30 to 40 times with selective pressure, where, for example, he says, I've got a thousand fruit flies in a container. I'm feeding them. They're living. I only harvest the eggs from the longest live fruit flies, and they're the only ones that reproduce, and he started doing this in the 70s, and after some number of generations, I forget what it was 50 or 70 generations. You see, big effects in the average lifespan of the fruit flies. He got them to live 5x their life span, 5x.

[1:16:38 - 1:16:40] Justin McMillen: So now he has strong signals around...

[1:16:40 - 1:16:41] Patrick Lilley: Right? That's how I guess living 500 years.

[1:16:41 - 1:16:42] Justin McMillen: Yeah.

[1:16:42 - 1:16:43] Patrick Lilley: That's astonishing.

[1:16:43 - 1:16:44] Justin McMillen: That's crazy.

[1:16:44 - 1:16:52] Patrick Lilley: And when we analyzed the RNA and the DNA from his fruit flies and the large replicated populations, unbelievable signal.

[1:16:53 - 1:16:56] Justin McMillen: So they're they're genetically predisposed to live longer.

[1:16:56 - 1:16:57] Patrick Lilley: Correct.

[1:16:57 - 1:17:05] Justin McMillen: And through that, you're now able to look at the difference between the ones that are living normally compared to 5x, and you can see the difference which points out exactly what's causing it?

[1:17:05 - 1:17:22] Patrick Lilley: Yeah. But here's what's interesting, and this was the shock to me is I had this naive notion, never having taken a biology class before I started this business, that, oh, there would be like probably a dozen genes involved in longevity, right? Sounds reasonable. No, no, it's fully one tenth of the genome.

[1:17:22 - 1:17:22] Justin McMillen: Really?

[1:17:23 - 1:17:38] Patrick Lilley: It's a huge proportion, and it's maybe even bigger than that. So it so for the fruit flies, they have 14,000 genes as opposed to our 20,000. We found 1400 highly, highly significant genes both through the RNA and direct genetic analysis.

[1:17:38 - 1:17:46] Justin McMillen: If you were to throw out the the how strong of a signal like, if they had that genes that are 100% going to live 5x.

[1:17:46 - 1:18:05] Patrick Lilley: Let's put it this way, we did a discriminative problem, almost like diagnostics. We said, okay, let's take the RNA measurements, or the genes. We did both, and let's create a classifier, an algorithm that puts the flies into the short lived group. You know, the 28 day flies versus the 140 day. Long live ones. 100% accurate.

[1:18:06 - 1:18:06] Justin McMillen: Wow.

[1:18:06 - 1:18:19] Patrick Lilley: And by the way, the discriminative signal, if you looked at the math and the distribution of the scores, they were so far apart it was like LA and New York. It's like there was no overlap between them.

[1:18:19 - 1:18:27] Justin McMillen: That's crazy. That's is it possible to see what what what are the biggest drivers?

[1:18:27 - 1:18:56] Patrick Lilley: Yeah, and this is where Michael's research on the paleo diet and a number of other factors is, is key. But he's also looking to develop therapeutics, and, you know, there's certain things I can't say that I know about how he's going to deliver them, but he is not looking for the typical pharma, one drug hit, one protein target in the body. He's looking to deliver a package to the body that hits multiple systems at once, to then lift everything.

[1:18:57 - 1:19:04] Justin McMillen: That's so interesting. Aging is fascinating because you got degradation of DNA, right? Methylation and all that.

[1:19:04 - 1:19:04] Patrick Lilley: Yeah.

[1:19:04 - 1:19:10] Justin McMillen: And then yeah, and I mean, you're you're just losing information, you know, from what David Sinclair?

[1:19:10 - 1:19:10] Patrick Lilley: Yeah.

[1:19:10 - 1:19:16] Justin McMillen: Yeah, and you know, he's said basically that he thinks he can get back to the original blueprint.

[1:19:16 - 1:19:16] Patrick Lilley: Yeah.

[1:19:17 - 1:19:21] Justin McMillen: Well, his big claim is that he found the original blueprint and that it exists, and we thought it didn't.

[1:19:21 - 1:19:22] Patrick Lilley: And he's a reductionist.

[1:19:22 - 1:19:23] Justin McMillen: Yeah.

[1:19:23 - 1:19:23] Patrick Lilley: Yeah.

[1:19:23 - 1:19:24] Justin McMillen: Yeah. Do you think he's correct?

[1:19:24 - 1:19:41] Patrick Lilley: I don't think he's correct. I think he's one of these guys that has very logical, very scientific sounding things that he says that touch on some realities, but they are a fraction of the total story, and he doesn't really understand the real system underneath.

[1:19:41 - 1:19:54] Justin McMillen: That's interesting. I think people in your position that work in science and technology, but science, the big issue is that the people that are communicating to the world have to speak that way..

[1:19:54 - 1:19:55] Patrick Lilley: Right.

[1:19:55 - 1:19:56] Justin McMillen: To get the information out.

[1:19:56 - 1:20:47] Patrick Lilley: But think about the definition of biomarker, right. Because it's such an abstract word and most people have never even heard of it. The idea of cholesterol as a biomarker, even that's broken down into LDL and HDL and VLDL and so on, it can get complicated fast. But I will tell you, in all of our work, there's not a single disease of any kind other than some extremely rare childhood diseases like Friedreich’s Ataxia, that a single biomarker of any kind, genetic RNA or protein, can diagnose a disease. It's always a portfolio of them, and it's very much like when you go into a museum, we can all stand in front of a Van Gogh painting from one perspective and look at it and all perceive because it's flat. But if you go to a sculpture, you have to walk around it to see the full shape of it. The biomarkers are like that. You have to measure multiple things in the blood to get a full picture of the sort of shape of the disease.

[1:20:47 - 1:20:47] Justin McMillen: Sure.

[1:20:47 - 1:20:57] Patrick Lilley: But what that means is that diseases and the body's response to them are systems, and they're not reductionist. They're not a single target for a single pill.

[1:20:58 - 1:21:00] Justin McMillen: Gotcha.

[1:21:00 - 1:21:09] Patrick Lilley: All of this was a surprise to me because, again, I had one biology class in high school and I didn't know anything about biology, and I had to learn it coming in from the side.

[1:21:09 - 1:21:31] Justin McMillen: Well, I, I mean, you've done an amazing-hell of a job figuring that out. I think what's what's interesting about what you've done and what our, our work is that to try to effectively treat addiction, you have to you have to say you have to have something to aim for. The problem with health care is not having anything to aim for are not health care. It's like cancer. You're aiming for no cancer detection,

[1:21:31 - 1:21:31] Patrick Lilley: Right.

[1:21:31 - 1:22:07] Justin McMillen: Substance use disorder, chronic disease. You know it's bio-psych-social is the behavioral health framework. So then you go okay what is healthy biology look like. What is healthy psychology. What is healthy. What is being socially healthy look like. So you have to establish what these things are and when I started with biology, that immediately brought me to evolution because I thought, well, there's all this, bro science. There's the latest, there's the supplement here, the this the getting, you know, I'm super big and jacked and the different things. But the truth is, is that the best information we have is how natural selection got us...

[1:22:07 - 1:22:07] Patrick Lilley: Right.

[1:22:08 - 1:22:21] Justin McMillen: To where we are. Yeah, and I mean, if you look at the early writings from when I was trying to create the trans model, everything was how are we built to live? And then that was where the environmental stuff came into play, where it was like environment changed so much and we didn't.

[1:22:21 - 1:22:21] Patrick Lilley: Yeah.

[1:22:21 - 1:22:43] Justin McMillen: I mean, if we were studying us in a lab and you change the environment rapidly, that would be a huge factor to consider. Or if you stopped moving, if you said, I'm going to make this, I'm going to make this rat not move anymore, knowing full well that all the other rats that for all the history of rats have run around for many hours a day, hunter gatherer style, you stopped moving. That would be a very big thing to look at. Why is the rat not healthy?

[1:22:43 - 1:22:43] Patrick Lilley: Yeah.

[1:22:43 - 1:22:44] Justin McMillen: Well, obviously...

[1:22:44 - 1:22:48] Patrick Lilley: So Justin is part of what you're doing, trying to bring back elements of the ancient environment?

[1:22:48 - 1:22:49] Justin McMillen: Absolutely.

[1:22:49 - 1:22:50] Patrick Lilley: That's interesting.

[1:22:50 - 1:22:56] Justin McMillen: Yeah, absolutely. I think everything we want to know about the health of our species can be answered by looking at the world through and evolutionary lens.

[1:22:56 - 1:23:20] Patrick Lilley: It's so analogous to what Michael Rose was saying about diet, which is under 40 or 45 years old. You can eat your recent ancestral diet. You know, mine. Mine were from Ireland and France. But when you're over that age, you need to go back to our deep ancestral diet in Africa. Right, and so his point was he's trying to bring elements of the old ancestral environment to bear upon today's people.

[1:23:20 - 1:23:21] Patrick Lilley: That's exactly what you just said.

[1:23:21 - 1:23:22] Justin McMillen: It makes sense, right?

[1:23:22 - 1:23:22] Patrick Lilley: Yeah. It does.

[1:23:22 - 1:23:46] Justin McMillen: You know, there's there's all these weird phenom- there's all kinds of different things to look at with this. So, for example, I teach a book in my class, I teach a class here and I teach a book called Tribe, and the whole theme in the book is that when human beings experience an external threat, an existential threat, that they come together and it's, you know, contrary to all belief, it's like wartime makes people more happy.

[1:23:46 - 1:23:46] Patrick Lilley: Yeah.

[1:23:46 - 1:24:00] Justin McMillen: Right? So when we have these threats on the outside, we suddenly come together and mental health gets better after 9/11 everyone’s united, I think Reagan said that if we ever had if we want world peace, we just need an alien to show up on the white House lawn.

[1:24:00 - 1:24:00] Patrick Lilley: Yep.

[1:24:00 - 1:24:08] Justin McMillen: And suddenly the human race. So why is that? Well, it could be that we're all built for love, and you can make it woo-woo but really, it's there's actually an evolutionary reason for this..

[1:24:08 - 1:24:09] Patrick Lilley: There is.

[1:24:09 - 1:24:14] Justin McMillen: Which is that when things got hard and there was something bad going on, we had to collaborate more.

[1:24:14 - 1:24:14] Patrick Lilley: Yeah.

[1:24:14 - 1:24:23] Justin McMillen: And anyone who didn't did do that died off and their genes didn't get spread, and so we are built to form interdependent bonds. So if you want to know what socially healthy looks like...

[1:24:23 - 1:24:23] Patrick Lilley: Yeah.

[1:24:23 - 1:24:38] Justin McMillen: You got to know what you contribute, and you got to know what you need, and you have to be able to to be clear about how you're contributing to your environment, and people have to collectively align on something, and I think purpose even sits like right square in that. Why humans need purpose so much?

[1:24:38 - 1:24:39] Patrick Lilley: Yeah.

[1:24:39 - 1:24:46] Justin McMillen: It’s because back in the day, it was like I killed the antelope. You made the baskets. I had to have a purpose, and guess what? If I don't have one, that's a signal.

[1:24:46 - 1:24:47] Patrick Lilley: Yep.

[1:24:47 - 1:24:53] Justin McMillen: Something's wrong and I'm going to die because I'm going to be isolated and everyone else is going to eat, and I'm going to be over in the forest by myself, and that means certain death.

[1:24:54 - 1:25:50] Patrick Lilley: Yeah, I mean, what you're describing is absolutely the truth here, because there people I have this funny view of evolution. They keep thinking of individuals and fitness of an individual. But the reality is the mathematics behind cooperation in evolution is the same math that drives large scale civilizations. It's the same math that drives voluntary transactions and free market capitalism, and there's a very interesting guy who's the former head of R&D at Goldman Sachs, Paul Burchard, he's in L.A. brilliant guy, 180 IQ kind of guy. He's actually got patents on emotional AI and moral AI, and he does all this super interesting work that's mathematically grounded, and he points out that morality, a system of morality and collaboration and our human emotions as well, that are tied to those things are an evolutionary imperative and mathematically necessary for civilization

[1:25:50 - 1:25:51] Justin McMillen: 100%.

[1:25:52 - 1:25:54] Patrick Lilley: But but he's rigorous about it in ways you and I couldn't be. Right.

[1:25:54 - 1:25:55] Justin McMillen: Yeah, yeah.

[1:25:55 - 1:25:55] Patrick Lilley: Yeah.

[1:25:55 - 1:26:01] Justin McMillen: Well, and you're, you're rigorous about in ways that I haven't been, and I want to learn from you, Patrick. Yeah, I really...

[1:26:01 - 1:26:01] Patrick Lilley: Thank you.

[1:26:01 - 1:26:03] Justin McMillen: I hope we can work more on things cause..

[1:26:03 - 1:26:04] Patrick Lilley: We'll find a way.

[1:26:04 - 1:26:15] Justin McMillen: Yeah. Because, so I'll give you an idea, like where I'm thinking about this is that, I don't know how you familiar with the project that Vahan and I are playing with Carenet. That, you actually you are.

[1:26:15 - 1:26:16] Patrick Lilley: He gave me some of the. Yeah.

[1:26:16 - 1:26:24] Justin McMillen: Okay, so it all comes down to. So the end result of it is any chronic disease you want better remission.

[1:26:24 - 1:26:24] Patrick Lilley: Yep.

[1:26:24 - 1:27:02] Justin McMillen: Okay. So, how do you do that? Well, he thinks, personalized or precision medicine approach to treatment based on data is the key. Right. So how do we collect enough data to be able to know that when this person shows up with this set of genetics and epigenetics that we give them, we start here on this treatment baseline, we start treating them, and then across a period of time, we measure again based on the new information. Then we make pivots and changes and then we have a better outcome, and then what I've been playing with recently, where I realized we had a whole in our proposal, was that I think we need to study what successful remission looks like.

[1:27:02 - 1:27:06] Patrick Lilley: Yeah. You talked about this in DC. What's the definition?

[1:27:06 - 1:27:31] Justin McMillen: Yeah. So like what would be interesting is if you if we measured thousands of people that are that have had SUD but have remained in remission and you look at a set of biomarkers from them and then you could, you could establish a target and you could say, here's what we're aiming for, and it may be that they need to be a little bit anxious or that they need to be lazy. Or it might be counterintuitive.

[1:27:31 - 1:27:31] Patrick Lilley: Yeah. Who knows?

[1:27:31 - 1:27:55] Justin McMillen: Break everything. Right? We know. But then you take that set of biomarkers and then you bring it back here. You measure people and you see how they're they're the same substance, and you go, okay, how different are their biomarkers from Joe's biomarkers. Biomarkers back here, and then you start to focus your treatment on addressing the things that would affect those biomarkers, and that that would be an approach to.

[1:27:55 - 1:27:56] Patrick Lilley: Right.

[1:27:56 - 1:28:07] Justin McMillen: You know, good news is if we don't do anything and we already produce about a 20 or probably that's conservative percent better post-treatment remission rate. So even it just we learn what our model is.

[1:28:08 - 1:28:21] Patrick Lilley: Well and don't forget what I said that we don't necessarily need biomarkers in the omics sense. We can take your medical records, your demographics, the things you did, all of that information, and yes, we've done that before.

[1:28:21 - 1:28:25] Justin McMillen: And how much it is. Is Vahan really familiar with your.

[1:28:25 - 1:28:26] Patrick Lilley: I think he's pretty familiar.

[1:28:26 - 1:28:27] Justin McMillen: Yeah.

[1:28:27 - 1:28:51] Patrick Lilley: He sees the way our system would work is we would have data from his system. We would develop predictive or diagnostic algorithms or choice algorithms for treatment, and then our algorithms would be embedded in his system. So when somebody came in, the things you collect about them would fire off at the algorithm that's at diagnose, choose a treatment presented to Justin's team. Right.

[1:28:51 - 1:28:55] Justin McMillen: You could create individualized treatment plans that auto populate in an EMR.

[1:28:55 - 1:28:55] Patrick Lilley: Yeah.

[1:28:55 - 1:29:10] Justin McMillen: So these are the things you should be measuring and these are the things you should be targeting, and of course again my biases, but also through having done it and watched it work is I think most treatment should be focused on getting us back to a health status similar to our ancestors. So...

[1:29:10 - 1:29:11] Patrick Lilley: Right.

[1:29:11 - 1:29:24] Justin McMillen: We're looking for those key things, and the assumption is that that if you do that right, that most of the issues around SUD that result in will either subside or go away, or they'll become harnessed to your advantage as long as they're translated into...

[1:29:24 - 1:29:31] Patrick Lilley: That was a very impressive thought I heard from you in DC was, how do we turn this into an advantage? I had never heard that perspective.

[1:29:31 - 1:29:32] Justin McMillen: Oh, it's I mean, that's how...

[1:29:32 - 1:29:35] Patrick Lilley: It's obvious when you hear it, which is why it's so beautiful. Right.

[1:29:35 - 1:29:52] Justin McMillen: Well, it's I think because addiction really that the pathways that it works on is what drives us forward, and so what I noticed through and you said, you said earlier, I bet you can see people ahead of time. You can kind of capture. So like for example, with you, you might if I talk about you?

[1:29:52 - 1:29:52] Patrick Lilley: No not at all.

[1:29:52 - 1:30:09] Justin McMillen: What I know about you. I think that, and I know this is true. There's a book I just read about it. Higher I- Higher IQ is actually one of the variables that exists in people that have SUD. That's an odd one, but you would think that, contrary to popular belief, it's like you're, so...

[1:30:09 - 1:30:12] Patrick Lilley: We're not smart enough to turn down the drink. Is that right?

[1:30:12 - 1:30:34] Justin McMillen: Well, I mean, but people who are smarter tend to have SUD more. So that's one of the, one of the things that we see. Another thing is an obsessive, obsessive nature. Right. So you saying thinking about it late at night and I'm going, going, going, people who can't sleep in a ruminating that obsessive nature is an inability to be satisfied. There's there's solid reasons in the brain for why that can happen.

[1:30:34 - 1:30:34] Patrick Lilley: Yeah.

[1:30:34 - 1:30:38] Justin McMillen: It's not just because you're awesome and you just are really determined. Right? There's actually there's...

[1:30:38 - 1:30:39] Patrick Lilley: You're built...

[1:30:39 - 1:31:06] Justin McMillen: -Different things around how your dopamine is working in your brain, how sensitive or how dense your neurons are that affect the way that we feel about wanting more. Now, the real issue, though, is and that's just a few. But if you look at the things that predispose people to addiction and they're working in all the same basic behavioral pathways that result in behavioral behaviors, that could, if harness would be the greatest advantage. Absolutely. Obsession with something. It's not a bad thing...

[1:31:06 - 1:31:08] Patrick Lilley: Right. We could see it being linked to purpose, right.

[1:31:08 - 1:31:24] Justin McMillen: Yeah. Like I always say to the guys here, the reason you're here is because you suck at quitting, right? And I'm like, but that's a good thing. It doesn't do you really good at not quitting. That's how you win in life, right? The the difference between people who win and lose oftentimes is that the people who quit early, they give up.

[1:31:25 - 1:31:25] Patrick Lilley: Yeah.

[1:31:25 - 1:31:45] Justin McMillen: Like you're some guy that you're on meth and you're riding your bike down the street at two in the morning with one shoe missing, just so you get to your dealer when everyone else, the normal people that aren't like you, they do cocaine once at a party and they're like, yeah, that was kind of weird and not going to do it again. Why is it that you're different like that? Why can't you quit? And how do we take your not quit nature and turn it to law school?

[1:31:45 - 1:31:46] Patrick Lilley: Yeah.

[1:31:46 - 1:31:50] Justin McMillen: Or so. Yeah. These things that seem obvious to me.

[1:31:50 - 1:31:51] Patrick Lilley: Only after you hear them though, you know.

[1:31:51 - 1:32:02] Justin McMillen: Yeah, yeah. I can't wait to get there. I'm so excited to get the the message out there more, but this is the most I've talked about. Tree House in this podcast.

[1:32:02 - 1:32:03] Patrick Lilley: It's worth talking about, man.

[1:32:03 - 1:32:29] Justin McMillen: Yeah, I like it, but I'm so fascinated with what you're doing and I think, I think that, with where things are going with computation, and I think if you continue to get out in the public and I think you will, I feel like I'm lucky to have captured this interview at this point in your life, because I think the next ten years of your life, I think you're going to become a very important person in this particular field.

[1:32:29 - 1:32:30] Patrick Lilley: Thank you, man.

[1:32:30 - 1:32:44] Justin McMillen: But it's yeah, I believe that. So I'm excited that we probably had one of the first opportunities to sit down with you. But it's the question is how is the public going to understand what you do, and that's really what it is.

[1:32:44 - 1:33:14] Patrick Lilley: Yeah. The key for us, I mean, if, for example, when we help ignite and others raise money for the joint ventures, is to present in very simple terms, there's a group of people who are suffering like this, there's no way to find them and therefore treat them all we're doing. Forget about all the magic of the math and the science and the computation. The fact is, we're going to produce a blood test. You're going to be able to go to a phlebotomist, quest, labore or whatever, take your blood and a couple of days later you have prostate cancer and it's 99% accurate.

[1:33:14 - 1:33:14] Justin McMillen: Yeah.

[1:33:14 - 1:33:50] Patrick Lilley: That's that's the story here. Because that first leads to the decision. The other thing we do is once we have the model of the biology, we can find those biomarkers that are symptomatic like a fever versus the other side, which is the biomarkers in the same group that might be nearer to the beginning of the core of the problem, those you want to target with treatments. So we license some drug targets for, opioid addiction to a company called GTC health. You may have heard of there in Irvine. They produced a drug to treat fentanyl addiction on the basis of the targets we gave them.

[1:33:50 - 1:33:50] Justin McMillen: Wow.

[1:33:50 - 1:34:44] Patrick Lilley: So they have an amazing AI that can synthesize compounds which we can't do, and we are really good at finding what are the what are the things in the body that you want to push on, which are RNAs, which proteins to change the disease. The pairing is powerful because it results in a new way to treat. So we use the diagnostic test to identify the patients. Now there's a treatment for those identified patients. Then if you collect the data we can say, okay, if you collect data on the different treatments, and by the way, you guys are out there because pharma is looking at one drug, one, you're looking at various aspects of the person. But you can look at all of those is interventions. If you record those over time, we can use our system to say based on the attributes of the person when they come in, which are the treatments that are most likely to work. So now you've got a whole thing from detection all the way through treatment, treatment choice and resolution.

[1:34:44 - 1:34:46] Justin McMillen: And then disease management ongoing after.

[1:34:46 - 1:35:08] Patrick Lilley: And disease management. Like for Mayo in our renal cancer we're predicting recurrence. But my goal is to have a blood test that's taken every two years after biopsy or after surgery so that you're monitoring the patient already they're monitored through imaging, but why wouldn't you retake the recurrence test every two years as a blood test, say, are you likely to recur in the next two years?

[1:35:08 - 1:35:16] Justin McMillen: So all these tests who who who owns, is it pharma that owns test testing?

[1:35:16 - 1:35:21] Patrick Lilley: Pharma is typically not at all involved in diagnosis, and they see themselves as distinct, which is a mistake.

[1:35:21 - 1:35:21] Justin McMillen: Yeah..

[1:35:22 - 1:36:06] Patrick Lilley: Right. Because you can't treat a patient unless you identify them as having the disease when take all the indication. Right. So our job is to make sure that undetectable or inaccurately diagnosed diseases are accurately diagnosed, to then get to treatment and then help pharma develop new treatments. We're trying to hit the whole spectrum. So there are a lot of diagnostic companies, and the bigger sort of commodity diagnostic companies like LabCorp and quest, who were not in the complex molecular tests, their venture groups and some of their strategic players, and we know them. They are actually seeking opportunities to acquire companies, smaller companies that have developed these more complicated, newer types of tests.

[1:36:06 - 1:36:11] Justin McMillen: That's interesting. I didn't know that. So LabCorp owns they would license these tests.

[1:36:11 - 1:36:16] Patrick Lilley: Yeah. So the typical way today is they don't know anything about this. Yeah LabCorp offers a lot of various things.

[1:36:16 - 1:36:19] Justin McMillen: But I thought they were just logistics. But somebody else is doing all that.

[1:36:19 - 1:37:14] Patrick Lilley: They do a great job of logistics and sites. But and they offer these tests and they don't necessarily own IP of the tests. Some of these are commodity tests like cholesterol. But increasingly they're saying, well, if we have distribution, if we have patient sites, if we have all of this and there's higher value tests that might be reimbursed at 1500 to $2000 for cancer detection in blood, then why wouldn't we be in that business? Because we can just hook them up to our network. So what they do is, as most big companies do, they say, okay, well, we can either try to develop our own tests or we can look at the farm of potential venture investments or partnerships with smaller companies developing these, and wait till they emerge in an evolutionary fashion to be the winners and losers, and then buy the winners at precisely the inflection point where their test is approved. They've got a reimbursement code, but they lack distribution. Boom. We can get them for a certain amount of money and it blows up.

[1:37:14 - 1:37:18] Justin McMillen: Have you been do you have you personally gotten reimbursement codes or have you done that process?

[1:37:18 - 1:37:21] Patrick Lilley: Our JVs and clients have done that, yes.

[1:37:21 - 1:37:26] Justin McMillen: Have you considered opening like, your own lab that just uses tests that you create?

[1:37:26 - 1:38:00] Patrick Lilley: Yeah. So so this is a perennial discussion with investors, right? Is they say, well, can you vertically integrate but, you know, you think is a friend of mine says, let the singer sing the dancer dance, right or right, and for example, we work with a couple of labs, but our primary lab partner is Cal Biotech down in San Diego, because they have been around for almost 40 years, developed over 200 tests, and they're just good at it, right? For us to learn everything they know about regulatory reimbursement, lab equipment, processes, sourcing reagents, all of that stuff.

[1:38:01 - 1:38:07] Justin McMillen: Well you dont have to you just create a division. But like a CEO, sorta character in there and just be like, don't do that, and then...

[1:38:07 - 1:38:08] Patrick Lilley: But it's already there.

[1:38:08 - 1:38:08] Justin McMillen: Yeah.

[1:38:08 - 1:38:12] Patrick Lilley: Right. So I mean a restaurant doesn't have cows in the back room, so.

[1:38:12 - 1:38:14] Justin McMillen: Farm to table maybe?

[1:38:15 - 1:38:55] Patrick Lilley: Right, yeah, yeah. Some of that. My oldest daughter's a chef actually has worked in some farm to table places. But but that's part of the reason. But but even David Barka the CEO of that lab, is now expanding and adding an ongoing clia lab to process the test, not just develop them. So we're just saying, all right, let's build a virtual factory of discovery, validation out licensing. David, does the development. David, can potentially do the first few years of the testing in his clia lab and then get it to the point where the big guys want to buy it, and the only missing piece that we haven't systematized is the capital race. Everything else is fully systematized.

[1:38:55 - 1:39:26] Justin McMillen: Yeah. That's that's beautiful. I'm wondering what other use cases outside of. I'm still thinking health care, but I'm thinking that there's this concept, you know, the, the Kennedy thing around that we got to go from a health care's sick care system to a health care system. There's a lot of talks, and we talked about this earlier about early detection and prevention and keeping people healthy so they don't ever get sick or being able to stay ahead of it, which is wellness in general.

[1:39:26 - 1:39:26] Patrick Lilley: Yep.

[1:39:27 - 1:39:28] Justin McMillen: I know we're going in that direction...

[1:39:28 - 1:39:29] Patrick Lilley: I hope so.

[1:39:29 - 1:39:42] Justin McMillen: I, we're working with, Blue Cross Blue Shield right now, and they're giving us a bunch of data, and we're, we're going to analyze it, and then we're helping them develop a value based care model for SUD.

[1:39:42 - 1:39:43] Patrick Lilley: Beautiful.

[1:39:43 - 1:39:44] Justin McMillen: It’s gonna be awesome. Yeah.

[1:39:44 - 1:39:45] Patrick Lilley: Yeah.

[1:39:45 - 1:39:47] Justin McMillen: Which is great because we're like, linking arms and...

[1:39:47 - 1:39:47] Patrick Lilley: Yeah.

[1:39:47 - 1:39:49] Justin McMillen: Like adversarial, you know, not that meant...

[1:39:49 - 1:39:50] Patrick Lilley: This blue shield?

[1:39:50 - 1:39:50] Justin McMillen: Yeah. Yeah.

[1:39:50 - 1:39:51] Patrick Lilley: Fantastic.

[1:39:51 - 1:40:05] Justin McMillen: In North Carolina, so shout out to them that they're innovating and doing this, and North Carolina in general is really, really on the forefront. They basically told us they want to be number one in the nation when it comes to provision of effective SUD care.

[1:40:05 - 1:40:05] Patrick Lilley: Wow.

[1:40:05 - 1:40:09] Justin McMillen: And they've like put their money where their mouth is. I mean they're pushing like to.

[1:40:09 - 1:40:10] Patrick Lilley: That's amazing to hear.

[1:40:10 - 1:40:33] Justin McMillen: Yeah it's incredible stuff. So so anyways Blue Cross Blue Shield is doing this. Well I bring that up. Is that there, and the same conversations happening with Aetna. But all these insurance companies are saying we're tired of paying for ten treatments that don't work. We want to pay one person to do it right and then effectively pay the right people to innovate, right?

[1:40:33 - 1:40:34] Patrick Lilley: Yep.

[1:40:34 - 1:41:17] Justin McMillen: And so that happening I feel like monitoring, patient monitoring, remote patient monitoring, wellness, for example, like an insurance company saying, okay, got a pool of a million people more than that usually, but got a million people. If you maintain a certain level of health and wellness, your premiums are lower, you know, as measured by whatever. So you can all be seen by a doctor. You have an annual physical, and this group of people, if they maintain a certain level, a certain BMI, a certain level of fitness, a certain amount of testing, then you're less of a cost burden and the goal would be to shrink to to increase that pool of people across the line and then and then give really good. No out of pocket health care.

[1:41:17 - 1:41:17] Patrick Lilley: Yeah.

[1:41:17 - 1:41:19] Justin McMillen: To the people who genuinely need it.

[1:41:19 - 1:41:19] Patrick Lilley: Right.

[1:41:19 - 1:41:35] Justin McMillen: Right. These are really obvious ideas. So I'm interested in that. But there's a place for what you're talking about, because I think this evolutionary computation model could be used for, I mean, if every person sort of had their ability to dump their own information into that on a regular basis.

[1:41:35 - 1:41:37] Patrick Lilley: Yeah, you could develop algorithms to predict many things.

[1:41:37 - 1:41:39] Justin McMillen: It could just manage health. Right.

[1:41:39 - 1:43:05] Patrick Lilley: I think there's a couple interesting things. One, let's go to the disease side of things and then come back to wellness. So one of the things we tried proposing to insurance companies, particularly the capitated health companies, let's say they're given 20,000 a year per patient. If the costs exceed that, they eat it. If they're less than that, that's their profit, right, and it's often the case that the bigger companies like United, we'll say, okay, for a set of neighborhoods in New York or whatever, we're going to give you that capitated market, and we're going to pay that 20 grand. So they have a lot of incentive to reduce the cost and gain health benefits. There's a bit of a challenge in the downstream cost, because anything they do that's preventive, their patients typically shift networks within two years. So the next guy experiences it and they're missing their collective opportunity to collaborate with each other. But what I proposed to them was very simple. I said, so you you have told us in meetings, all of them said the same thing. 5% of our patients are half of our costs, right. Late stage type two diabetes, chronic kidney disease, you know, end of life. There's all there's a set of things, and they said, we understand very well who they are. I said, okay, but a substantial proportion of those people are dying in that year, and they said most of them said, okay, so if they're dying, but you continue to have 5% next year for a 50% of your cost, somebody is coming in right? So if that's the case, why don't we take the 95% that aren't in the 5%...

[1:43:05 - 1:43:06] Justin McMillen: Figure out who’s gonna...

[1:43:06 - 1:43:52] Patrick Lilley: And predict who's going to be in your group next year? Now you can do early interventions to prevent them from getting into the next year's 5%, and we can do it without biomarkers. We can do it from lab tests, demographics, past meds, everything you have, because the health care system has tremendous medical records and claims data, and we have in the past used that kind of data to predict certain things, like how far out are you going to get knee replacement surgery if you're an osteoarthritis patient? We are about to go four years out versus all other AI and statistical methods being one year out, right? Stuff like that, which allows you to get a bigger patient group to attack with an osteoarthritis drug to be preventive. So this didn't fly with these guys because they just couldn't leave the notion of their standard reports in the idea that we know who these patients are.

[1:43:52 - 1:44:24] Justin McMillen: They're just not that they're not, I'm always shocked by when when I meet with some of those folks, like, just, I think they're moving too fast or they're not thinking. I mean, it's like it's. Or there's just these Leviathan machines, and I don't want to speak poorly. I mean, we we work with lots of payers and they do good work with us, and, but I've noticed, I mean, when I say moving too fast, I don't they're not though, because it's actually takes like a year to get anything done. It's what they have 50 projects and none of it matter and no one's really putting their brainpower on.

[1:44:24 - 1:44:25] Patrick Lilley: Yeah.

[1:44:25 - 1:44:25] Justin McMillen: Any one of them.

[1:44:25 - 1:45:27] Patrick Lilley: Part of the problem is they're big. I've worked in big companies right out in my when I was 30, I was head of strategy for Toshiba's, you know, computer division in the US, about a $4.5 billion business that was a relatively small team of 300 people, but in a much bigger corporation, a $50 billion corporation, and what I noticed was we were all even in our 300 person division. We were highly siloed, and we couldn't talk to the consumer electronics people at Toshiba. We were forbidden from doing so. It was hard to talk to certain people in Japan, only people directly related to our computer business, and that's not an indictment of Toshiba by any means. Big companies are like this. But what I have noticed is pharma and health care in general. People are so siloed that you and I working in cubicles, don't know anybody else. You're my neighbor and I go to lunch with you. I have no idea what you do for a living in the company. We certainly don't collaborate. We don't think of bigger strategic initiatives. We might even have the same innovative idea of pitch it to our two different bosses, be working at the same time and never collaborate.

[1:45:27 - 1:45:40] Justin McMillen: I never want that. Yeah, it's interesting because as a as a person that's bootstrapped a few companies with my family and all that, and like, we strive constantly for developing SOPs and systems and you want to make systems.

[1:45:40 - 1:45:41] Patrick Lilley: Right.

[1:45:41 - 1:45:47] Justin McMillen: But then the fear is that you create systems that just produce absolute people not thinking for themselves.

[1:45:47 - 1:45:47] Patrick Lilley: Yep.

[1:45:47 - 1:45:59] Justin McMillen: And no creativity, and yeah, that's a constant challenge, and I I'm deeply aware of that, and as we grow and we're going to grow much bigger, I think we're going to be very careful about that.

[1:45:59 - 1:46:29] Patrick Lilley: Well you have this nice model also of of replicated sites. Right, and what's interesting about that is rather than try to get there's obviously setting aside the logistic problems of bringing every person to a giant site in Kansas. One of the things you can do is you can, understand and test different protocols and variations on them in different places, and you will have a naturally emerging sort of species in each place of people that are innovating in a different way, and then you can begin to share that knowledge.

[1:46:29 - 1:46:30] Justin McMillen: That’s exactly right, Patrick.

[1:46:30 - 1:46:52] Patrick Lilley: And because they're isolated from each other, there is some benefit that their ideas don't. If I have one site and somebody has an idea to solve a problem A, the company decides to invest in one potential solution to A. If I have two sites, one pursuit pursues one of the solutions, one pursues another one because they independently think of it, and then you get to compare them and see which is best. Right?

[1:46:52 - 1:47:20] Justin McMillen: You're exactly right. That's exactly how it works. We do the same treatment model in the different sites. For the most part. There's a couple tweaks. But it'll never be the same because you have different people delivering it. Yeah. So what you see is like, this person who's been trained in this way or has this background is delivering this information in this way, or this works better for this demographic. But we do have a perfect scenario for sort of doing A/B testing and pulling one thing out and adding something else. But you got to be careful because we're dealing with human beings and...

[1:47:20 - 1:47:21] Patrick Lilley: Yeah.

[1:47:21 - 1:47:22] Justin McMillen: You want to, you know, if if you got something that works...

[1:47:22 - 1:47:24] Patrick Lilley: You can’t be wildman experiments, I get it.

[1:47:24 - 1:47:43] Justin McMillen: You can't be like, oh, we're going to have none of these guys exercise. Well, we know full well that waking up early in the morning and exercising first thing is a key. It's a key factor. Although I'd be super interesting, I would love to see if you just got people in excellent physical condition and had them heavily social, interdependent social relationships and didn't know clinical how they'd fare.

[1:47:44 - 1:47:45] Patrick Lilley: Yeah, that is interesting.

[1:47:45 - 1:48:08] Justin McMillen: I would really because because psychology is so interesting. Yeah. The question is like what drives what? And the chicken or the egg, you know, I mean, is it you got to have neurochemistry, right? And then people think better or do you think better and then behave differently, and then that gets neurochemistry, right. Like what's the most that's been asked so many times. What's the most important dimension.

[1:48:08 - 1:48:11] Patrick Lilley: Yeah. Well I'm personality is heavily genetic.

[1:48:11 - 1:48:11] Justin McMillen: Yeah, for sure.

[1:48:11 - 1:48:13] Patrick Lilley: Right. Especially introversion and extroversion.

[1:48:13 - 1:48:31] Justin McMillen: Absolutely, and that's a huge subject of this to that I think I think that yeah, I think personality, the challenge that we're facing is that I think most people that are looking for answers to the nature of addiction are looking for the way substances metabolize in the body.

[1:48:31 - 1:48:32] Patrick Lilley: Yeah.

[1:48:32 - 1:48:37] Justin McMillen: And I'm more interested in behavioral phenotypes that are driven by genetic switches.

[1:48:37 - 1:50:04] Patrick Lilley: Right, and the body's response I'll tell you something interesting about cancer. That's an analogy to this. So there are companies trying to diagnose cancer, and there's a number of tests out there that try to detect the the DNA shedded from tumors. Now it's floating around in your blood if you have a tumor, but it's floating around in extremely tiny amounts, and the way they detect is because it's different from the person's DNA and they look for particular genetic variants. There's a subset of known genetic variants associated with particular cancers. Well, here's the problem. So if a person has a solid tumor in their body, the thought is, oh, that's all a tumor is comprised of all the same cancer cells. It's not. It's multiple species of cancer cells. There are different genetic variants within the tumor, and what's happening is the body's immune system is an environmental influence on the tumor. So are treatments. So what's happening is artificial selection. The species of cancer within the tumor that are killed by the treatments or the immune system go away. The ones that aren't proliferate, and people think the cancer is evolving, but it's really not that. It's that the various species of cancers are going through, you know, artificial selection. What's interesting about that is those can continue to change in extraordinary ways, and some of the medications may induce further mutations or we don't know, the number of ways that a cancer can mutate is much larger than the number of ways the body can respond to any cancer.

[1:50:04 - 1:50:05] Justin McMillen: Yeah, it's so fast.

[1:50:05 - 1:50:22] Patrick Lilley: So you can measure a smaller number of things in the body's response that are more consistent and reliable. So what we focus on is measuring the RNA and proteins in the blood that are associated with the body's response to the cancer. Now the signal is persistent regardless of the evolution of the cancer under treatment.

[1:50:22 - 1:50:23] Justin McMillen: That makes sense.

[1:50:23 - 1:50:47] Patrick Lilley: And what you said was very similar, right. It's this notion that, okay, I'm doing these particular things to the patient over time. What we need is to find the commonalities in the biology of the patient with respect to their response, their biological response, and their other phenotypic response, as reflected in biology, to the addiction to the treatments.

[1:50:47 - 1:50:48] Justin McMillen: Yeah.

[1:50:48 - 1:50:52] Patrick Lilley: And you're going to have a smaller number of things to examine, which makes the problem easier to solve.

[1:50:52 - 1:51:09] Justin McMillen: Yes, and to, to treat it too. You can't have it has to be winnable. We may discover that there's an exact way to treat addiction, but it's so unbelievably, it's just impossible. It's too resource. There's too many resources needed.

[1:51:09 - 1:51:09] Patrick Lilley: Well...

[1:51:10 - 1:51:11] Justin McMillen: So you have to find that how you capture the most.

[1:51:11 - 1:51:12] Patrick Lilley: Yeah.

[1:51:12 - 1:51:13] Justin McMillen: With. You know what I mean? You really.

[1:51:14 - 1:51:29] Patrick Lilley: There's there's a guy that, I forget his name. He does these, little short, things on YouTube, and he was a, like, a former CIA guy or something. He talks a lot about fooling your mind into doing certain things and how you read people, and it's very interesting guy.

[1:51:29 - 1:51:30] Justin McMillen: Is it Andrew Bustamante?

[1:51:31 - 1:51:47] Patrick Lilley: No. no, I've seen that guy, too. So one of the things he says, it's so interesting is so take a sales script, the world's best sales script, and it's working really well for your sales guys. Now give it to an introvert. I'm sorry, but it's not the world's best sales script anymore.

[1:51:47 - 1:51:48] Justin McMillen: Yeah.

[1:51:48 - 1:51:50] Patrick Lilley: And this is the point.

[1:51:50 - 1:52:28] Justin McMillen: Yeah. You got to tease out what are the base things that like I think of it like an operating system. Like you know they say 1 in 10 people worldwide and this, this the same in other areas of the world, which is the strongest idea that there's some sort of genetics at play. 1 in 10 people has SUD and it's been that way for a long time and go back decades and see this 10% number, and that's alcoholism, poly substance all combined 10%. That- so you could you could argue that that 10% maybe they they are the extroverts, and so this script works perfectly for them. But the idea is you got to start with like, what is this? What is this particular type of human look like.

[1:52:28 - 1:52:28] Patrick Lilley: Right.

[1:52:28 - 1:52:33] Justin McMillen: How do they behave. How do they interact, and I have theories about that. I want to know the answers to that. This is...

[1:52:33 - 1:52:36] Patrick Lilley: The key is to measure right and then we can help analyze and predict.

[1:52:36 - 1:53:10] Justin McMillen: Yeah. Yeah. So you start with like for this particular group of people the baseline treatment should include these things because it addresses like the key key phenotypical behaviors with them. Then then you go let's go one step further and look at how the different, because there's countless different variations of that operating system, operating system being bio-psych-social, all the factors that contribute to it, and let's start adjusting treatment along the way to be more effective, and then all that does that, that's pointless unless you measure it against what happens after treatment. Look and sustain remission. So it's complicated problems.

[1:53:10 - 1:53:33] Patrick Lilley: Yeah. But here's something interesting. You know, I have complimented you before on the notion that, again, you're looking at many aspects of the patient, not single drug, single target kind of reductionist thinking. One of the one of the corollaries of that, though, is how do you know if people are different from each other, if you're looking at them along one dimension? Oh, I'm just looking at people with two arms. Well, that's almost everyone, right?

[1:53:33 - 1:53:34] Justin McMillen: Right.

[1:53:34 - 1:53:49] Patrick Lilley: But but are they male? Are they female? Are they introvert? Are they extroverted? Are they using meth or cocaine or, you know, there's a whole set of things. Are they physically fit or are they not physically fit? If you don't look at all of those, you have no way to know somebody is different, which doesn't lead you to thinking of treating them differently.

[1:53:49 - 1:53:50] Justin McMillen: That's true.

[1:53:50 - 1:53:57] Patrick Lilley: So if you start with a reductionist premised, you never even think about multiple treatments, which is where we are today in pharma.

[1:53:57 - 1:54:04] Justin McMillen: Yeah. Well, it's it it's the scientific method is beautiful. But that's kind of the issue with it is that you really I mean, that's...

[1:54:04 - 1:54:18] Patrick Lilley: You should hear my 26 year old daughter complain about the use of the scientific method. She's she agrees with you. It's a beautiful method. She she says she says, dad, look, we've been working in this industry for a while. She says, I've seen like three people use it.

[1:54:18 - 1:54:19] Justin McMillen: Yeah.

[1:54:19 - 1:54:20] Patrick Lilley: They misuse it all the time.

[1:54:20 - 1:54:21] Justin McMillen: Sure.

[1:54:21 - 1:54:32] Patrick Lilley: They're not the number of studies we see. One of the first autism data sets we had, which was one of the 13 we used, has only boys in it.

[1:54:32 - 1:54:33] Justin McMillen: Wow.

[1:54:33 - 1:54:40] Patrick Lilley: And and apparently the reason is because, well, autism is tough to diagnose in girls. Well okay. So why is this?

[1:54:40 - 1:54:41] Justin McMillen: I’ve heard this before somewhere else.

[1:54:41 - 1:55:02] Patrick Lilley: Yeah. Here's what we found it so interesting there. There's there's multiple examples of this. But there's an RNA in the blood of boys with autism that's elevated. In girls, It goes down, it goes in opposite directions. So if you just had a test to say is it elevated? It would say no girls had autism. Or it might say normal girls.

[1:55:02 - 1:55:06] Justin McMillen: But you have to you have to first tell it that it's a female, and then this is where the...

[1:55:06 - 1:55:36] Patrick Lilley: Right and there are specific things. All of the forms of quantitative analysis out there, they don't. This is going to shock you if you know two things about somebody's two variables, two RNAs or RNA and gender, there can be a mathematical interaction, multiply them, divide them, whatever. Most of the equations in neural networks and regression and all the statistics that are used in machine learning and statistics, they don't allow variables to interact. Have you ever heard the term independent variables?

[1:55:36 - 1:55:38] Justin McMillen: Yeah. Why are they independent?

[1:55:38 - 1:55:41] Patrick Lilley: Because they're treated as mathematically independent and they don't interact.

[1:55:41 - 1:55:42] Justin McMillen: But that's not how the nature works.

[1:55:42 - 1:55:43] Patrick Lilley: of course not.

[1:55:43 - 1:55:44] Justin McMillen: Everything is interdependent.

[1:55:44 - 1:55:54] Patrick Lilley: Right? I mean, thinking of income and golf, the amount of golf you consume, there's an interaction with income, I guarantee you, because it's expensive. That's a really simple example.

[1:55:54 - 1:55:56] Justin McMillen: Yeah that's crazy.

[1:55:56 - 1:56:13] Patrick Lilley: So that's the frustration is that we we need people to also sit back like we talked about, and this gets back to wellness. What's the definition of wellness. Let's define it, and then let's find the aspects of which we can measure, and then we can look at efficacy of methods to hit that in the right way.

[1:56:13 - 1:56:32] Justin McMillen: Yeah you have to start. You will never hit a target. You're not aiming for people don't, and so it blows my mind how, and it's because I think highly intellectual people love tend to like more conscientious people really like processes, and so they forget sometimes to even that there's a goal.

[1:56:32 - 1:56:37] Patrick Lilley: Yeah. Yeah. Let's just turn the handle. The meat grinder. I'm not looking to see if the meat comes out.

[1:56:37 - 1:56:53] Justin McMillen: It's like but and the challenge with that is I think the goals that you haven't achieved yet, you requires vision to have that, and that's very that's kind of opposite to somebody who's a very intellectual because I mean you find people with both. You have both. Certainly. That's why I asked you, like, how did you come up with the..

[1:56:53 - 1:56:55] Patrick Lilley: Right.

[1:56:55 - 1:57:05] Justin McMillen: But you have to have an imagination. You have to be able to, to see something that isn't there to even have a goal or a target, or have data that informs you of what the target should be.

[1:57:05 - 1:57:23] Patrick Lilley: You're so spot on for another reason. I have talked about wellness to people in healthcare, insurers, pharma diagnostician. A lot of people, if you ask them what is wellness, they sit and think about it. Inevitably their answer is not these words, but it means this absence of disease.

[1:57:24 - 1:57:24] Justin McMillen: Yeah.

[1:57:24 - 1:57:28] Patrick Lilley: But if I look at any NBA player, their wellness is above mine.

[1:57:28 - 1:57:29] Justin McMillen: Yeah.

[1:57:29 - 1:57:33] Patrick Lilley: Right, and I'm disease free. That doesn't mean...

[1:57:33 - 1:57:34] Justin McMillen: Right.

[1:57:34 - 1:57:43] Patrick Lilley: So that's a really really important point, and so there's a vision that's required to say what is that thing. It is not an absence of something.

[1:57:43 - 1:58:07] Justin McMillen: It's coming like the internet, and then the the truth being on trial and all the bullshit that's getting slung around in the world today is terrible. But at the same time, what you have is you have this meritocracy sort of thing where people that are discovering things on their own or bringing them to the table, and some of the ideas are bad, but they're popping up and algorithms are bringing them forward, and YouTube and different...

[1:58:07 - 1:58:08] Patrick Lilley: Right, people are trying them. Yep.

[1:58:08 - 1:58:17] Justin McMillen: So so we have this new layer of information moving forward alongside science, which is very slow, and I think a great example is fasting or...

[1:58:17 - 1:58:17] Patrick Lilley: Yeah.

[1:58:17 - 1:58:22] Justin McMillen: Carnivore diet or you know, there's no money in in cold exposure or saunas.

[1:58:22 - 1:58:23] Patrick Lilley: Right.

[1:58:23 - 1:58:46] Justin McMillen: Right? Doctors aren't. But they you know, people are like this makes me feel better then Scandinavia has shows this study was causing and reducing all cause mortality by 50% for 20 minutes a day, five days a week, and it's like we're getting these different things where we're learning and it's starting to shape the direction that we're going, which is cool, I think. Yeah, the future is, some days I feel like it's falling apart, but other days I'm pretty excited.

[1:58:46 - 1:58:54] Patrick Lilley: Yeah, well, I'm glad you have optimism. I've always said if you're an entrepreneur, you must be an optimist because otherwise you're just insane.

[1:58:54 - 1:58:54] Justin McMillen: Yeah.

[1:58:54 - 1:58:56] Patrick Lilley: Or or stubborn, which we all are.

[1:58:56 - 1:59:00] Justin McMillen: Yeah, yeah. You have to be almost delusional sometimes especially early.

[1:59:00 - 1:59:19] Patrick Lilley: Yeah, there's there's some actually quite good evidence that, an optimistic mindset. The benefit of that is that when you're optimistic, you will not only consciously but unconsciously work toward hitting those goals and making that that positive reality happening.

[1:59:19 - 1:59:20] Justin McMillen: Yeah, yeah.

[1:59:20 - 1:59:43] Patrick Lilley: And there's a very funny thing. So I have a strong math background. So you can as you can imagine, I'm not a gambler. Right? I just I don't gamble. I mean, I gamble enough with the business, right. Just being in this business at the same time, about once every 9 or 12 months I will buy a single lottery ticket, and not because I expect to win, but because the feeling of the hope for the moment gives you a little kick of energy.

[1:59:43 - 1:59:44] Justin McMillen: Sure, sure.

[1:59:44 - 1:59:54] Patrick Lilley: Right, and just sort of maintains the optimism and it's not a true reflection of me believing I'm going to win the lottery. It's just the notion of the sliver of possibility gives a little energy.

[1:59:54 - 2:00:26] Justin McMillen: Yeah, yeah, I get it, I get it. But, what you're talking about around belief, I think that's what you mean. Like what somebody believes will help make it come true like that there's, there's a lot to that, and that I think the there could be a quantum physics conversation about that, that it gets either really woo-woo like, you manifest things into becoming real. But I think what we know and Google's represented this with their algorithms is that our preexisting beliefs, because the way the brain works, we spend more time ignoring information than.

[2:00:26 - 2:00:26] Patrick Lilley: Yeah.

[2:00:26 - 2:00:54] Justin McMillen: So so our brain will orient around what we believe to be true. So if you believe that your. So for example, I'll use people that come into treatment, and if you have a belief system based on doing a bunch of terrible things because you've been sick, no excuse, but that explains it right in your belief system, as I'm, I'm a piece of shit and I'm not worthy, and that's the lens you look at the world through and your brain's trying to ignore everything outside of your belief system. You will look for evidence to support that. That's true.

[2:00:54 - 2:00:55] Patrick Lilley: Yeah, yeah.

[2:00:55 - 2:01:03] Justin McMillen: And so belief plays a role. I'd be interesting to find out if some people are more likely to be able to believe things to be true and make them come true than others.

[2:01:04 - 2:01:06] Patrick Lilley: Yeah, let me come back to that in a moment.

[2:01:06 - 2:01:06] Justin McMillen: Yeah.

[2:01:06 - 2:01:28] Patrick Lilley: One of the things I want to tell you first is I mentioned you, we had done a little we've done some work in PTSD, and ignite has asked us to license PTSD biomarkers. So that's our next project with him. When you think of PTSD you think of as a purely psychological thing. But I think we already know that we can get better than 95% sensitivity and specificity on PTSD in the biomarkers

[2:01:28 - 2:01:29] Justin McMillen: Before or after it's happened.

[2:01:29 - 2:01:54] Patrick Lilley: After it's happened, we have a data set where we're going to try to predict who gets it, because that would be valuable. Do you deploy somebody to this kind of environment or not? Right. But we have data sets of first responders, people who were, at the World Trade Center who are not first responders, just civilians, and we have some military data sets and it's quite clear that there is a biological signal from the psychology. So the brain is affecting the biology of the body.

[2:01:54 - 2:01:55] Justin McMillen: Oh, yeah.

[2:01:55 - 2:02:28] Patrick Lilley: You know this because the business you're in, a lot of people don't believe this. What's really interesting, back to your other point is, one of the things I've wanted to do for a very long time is just sort of a side experiment. We have access to all these wonderful data sets from the NIH, tens of thousands of them for different diseases, where they measure tens of thousands of RNAs per patient or proteins or DNA. One of the things I want to do is take a bunch of clinical trial data sets where treatments were tried on people and there was a placebo arm, because in almost all trials, there are placebo responders.

[2:02:28 - 2:02:30] Justin McMillen: What does that mean?

[2:02:30 - 2:02:32] Patrick Lilley: They're not given the drug, but they get better.

[2:02:32 - 2:02:33] Justin McMillen: Got it? Okay.

[2:02:33 - 2:03:10] Patrick Lilley: They improve. This is particularly true in autoimmune diseases. There's a and and Alzheimer's is true also. But I think for a different reason is that social engagement happens in a trial, doesn't typically happen to patients when they're not in a trial, and social engagement is very important for your brain, as you already know, you've talked about it. But autoimmune disease, we find, is very biologically driven, is very immune driven. But there's a pretty high responder rate, 20 to 30% in the placebo arms of the trial. What I'd like to do is throw away all the treatment arms, collect all the placebo arms from all the trials, and predict placebo response from the biomarkers at the beginning.

[2:03:10 - 2:03:11] Justin McMillen: You should do that.

[2:03:11 - 2:03:19] Patrick Lilley: Because you can think, is it is it? For example, is it the RNA for optimism? Is it for gullibility? Is it for positive thinking? What is it exactly?

[2:03:20 - 2:03:21] Justin McMillen: You should absolutely do that.

[2:03:21 - 2:03:23] Patrick Lilley: Now imagine you find the drug targets.

[2:03:24 - 2:03:24] Justin McMillen: Yeah.

[2:03:24 - 2:03:37] Patrick Lilley: Then you say, now I can develop a drug that if I give a small dose to somebody, amplifies the value of every other drug. So you can give a smaller dose of the real drug, have less side effects. But the same efficacy.

[2:03:37 - 2:04:03] Justin McMillen: You should do it. There's there was an article I read about the placebo and it was like, why is why are we not looking at this? The signal is so strong. It's so it's it's there constantly, and it's not it's it's too, too strong for it to be not of importance, and then I think somewhere in the this was like 5 or 6 years ago, but somewhere in the article they had said that they found some sort of genetic factor that contributed to the placebo response. Have you heard about this?

[2:04:03 - 2:04:04] Patrick Lilley: I haven't. I'll look, yeah.

[2:04:04 - 2:04:08] Justin McMillen: I'll see if I can find it, I think I think it was like a big publication, like New York Times or something.

[2:04:08 - 2:04:12] Patrick Lilley: Yeah. Imagine an amplifier for every treatment.

[2:04:12 - 2:04:50] Justin McMillen: Well, I think there's. Yeah, there could be something. So. So when you said, PTSD, you talked about PTSD and and then psychology and then the physical. I mean, there's an intersection, between the way we think and feel in the body, and that's the autonomic nervous system. So arousal, sympathetic and parasympathetic, rest and digest the things that happen when the body's autonomic nervous system shifts from a state of arousal to comfort. I mean, rest and digest, is where all of emotional regulation sort of stems from, and then emotions obviously color how we think. Right?

[2:04:50 - 2:04:52] Patrick Lilley: Right, and there's floods of stuff, yeah.

[2:04:52 - 2:05:20] Justin McMillen: Exactly. So if somebody is I don't know. I mean, it does not surprise me at all. So for example, if you study somebody who had PTSD, you might find biomarkers that are related to them being sympathetic, nervous system dominant. Their Gaba could be lower. So it's not putting the brakes on the sympathetic nervous system. So they're hyper vigilant. Hypervigilance is going to mean that they're not sleeping well that they're having racing thoughts. They're going to be laser focused not notice things, jumpy I mean you would see all of these things. They would be chronically stressed out right.

[2:05:20 - 2:05:20] Patrick Lilley: Yep.

[2:05:20 - 2:05:27] Justin McMillen: Because they're in this constant state. So yeah there's that's the thing is all these things intersect. But we all want to separate them out.

[2:05:27 - 2:05:28] Patrick Lilley: Yeah.

[2:05:28 - 2:05:28] Justin McMillen: But they dance together.

[2:05:28 - 2:05:36] Patrick Lilley: Completely integrated. Right. That's that's again why we want to model everything in systems. Not independent factors.

[2:05:36 - 2:05:49] Justin McMillen: Yeah. If you ever have like a thing that you're playing with and you're like what would be the thing that would affect this or what sort of assessment tool that's validated. Can we play with the data from psychometrics and stuff. You should hit me up because I have.

[2:05:49 - 2:05:52] Patrick Lilley: Absolutely. Yeah, yeah. It sounds like you have a ton of knowledge there.

[2:05:52 - 2:05:55] Justin McMillen: I don't have a ton- I mean I just know where to look. Okay. I have a lot of ideas.

[2:05:55 - 2:05:57] Patrick Lilley: It's good enough. I got a guy who's as good as.

[2:05:57 - 2:06:04] Justin McMillen: Yeah, well, no, I mean, I have, I have this stuff, but I, I'm just so curious about it that I think about this all the time. So my brain would run.

[2:06:04 - 2:06:06] Patrick Lilley: Yeah, yeah.

[2:06:06 - 2:06:14] Justin McMillen: Yeah. Oh, man. I don't know where we go from, how long have we... Luke, how long have we been going?

[2:06:14 - 2:06:15] Producer A little over two hours.

[2:06:16 - 2:06:17] Justin McMillen: This is great. Yeah, it feels like...

[2:06:17 - 2:06:18] Patrick Lilley: Amazing.

[2:06:18 - 2:06:54] Justin McMillen: Yeah, like ten minutes. So, we should probably close things out, but, I think, I think, I think the question I have for you is, where do you think your field is going? And how do you think, medicine is going to change and prediction is going to change? And precision medicine is going to change. Like, where do you think that's all headed, dude, do like your sort of...

[2:06:54 - 2:09:25] Patrick Lilley: Yeah, yeah. Let's say to your optimism or pessimism, we'll just change the timescale, right. Is it five years? Is it ten, is it 20? But my hope is that we arrive in a place where for, you know, every major condition, you know, whether it's substance use disorder, cancer, neurological diseases, autoimmune, whatever it is that, doctors are assisted by AI in the large language models in saying, okay, the patient has described some things I'm measuring, some things. I have enough information now to send them for a simple blood test or three blood tests. I suspect it's one of these three things. I think it's a mess or chronic fatigue syndrome or whatever. So I'll send them for either the most likely or all three of those things, and I'll get a yes or no back, hopefully high accuracy, and then I'll say, okay, of the available treatments, I have some sort of heuristic or I have another blood test I can give. Maybe it was even done at the same time that says which of the treatments are likely to work for this person, and then they prescribe, and I think what happens is two, two fold, one firm is going to resist this a lot because they already resist pay for performance. TNF inhibitors like Humira, for example, work for in the long run after six months for only 30% of the patients. But in the United States, you have to stay on them for a much longer period of time before you're allowed to switch from that first line treatment to an alternative. In the meantime, you've got a progressive disease you're suffering, you've got pain, and this insurance system is still paying for this very expensive drug. Predicting it's not going to work for you takes revenue away from the market leader. So they don't want to shrink their market. The up and comers that have different mechanisms of action for new drugs they should adopt to test like this though, because they should say, well, I can't be first line at all because of the the TNF inhibitors. I've got this Jak1 inhibitor. Right. So what I'm going to do instead is I'm going to partner with the test to predict, and I'm going to switch. Although 70% of people who doesn't work for over to me as first line. So the incumbents I think are going to start to turn over the big pharma companies, because it's always the case that the small and medium sized companies recognize the opportunities to change before the big ones, too.

[2:09:25 - 2:09:28] Justin McMillen: The big ones are like the Titanic trying to steer away from the iceberg and they don’t.

[2:09:29 - 2:09:39] Patrick Lilley: Yeah, look, I watched this happen in the PC industry. I was watching the internet emerge while I was at Toshiba. They didn't recognize it. They were thinking in PR hardware, where are they today? They're zero. They're gone.

[2:09:39 - 2:09:39] Justin McMillen: Yeah.

[2:09:39 - 2:11:29] Patrick Lilley: Right. Polaroid is gone. There's a plenty of examples of this. So one, I think that that notion of pairing the diagnostic with another test for treatment choice is one thing. The other thing is I think we're going to eventually get the message out and people will compete with us and start to develop systemic AI models that are mathematical, that will model the biology, and the consequence of that is we will get better treatments that are very specific to particular patients. Right, and that's crazy. I don't think we're going to lose doctors, although in terms of eliminating doctors, I worry about doctors today because they're being paid less and less and told to do more and more paperwork, and we have a physician on our board who's telling me that when you have a practice of multiple doctors who are in their 50s and 60s, when one of them retires, they can't find any young doctors because the young doctors only want to work for big corporations, work 9 to 5, get their pay. They don't want the upside or the headache, of managing a practice. But doctors are still necessary. Everything you said is involving the human touch in the patient experience and how crucial that is. People are very worried about doctors being pushed aside by AI for diagnosis and everything else, but the doctor still needs to be there for time of interaction with the patient, because there's so much more to a human being, and if we look at let's take the advance that we saw in the 80s, which is antilock braking in cars. Okay. So the thing is that you think anti-lock braking, man, that's going to replace the human being for deciding when to pump the brakes in a, in a slippery situation? Well, that's true for most of us, but none of the professional race drivers has anti-lock braking in the cars. It's the very best in the profession. Will still be the key and they'll still be there.

[2:11:29 - 2:12:46] Justin McMillen: Yeah, yeah, yeah. I think doctors I, I think about this all the time today. Part of my connection to Vahan was when he started telling me what he was doing. I was I started meeting with different people in DC in different areas, and, and all of them were bringing these different ideas to me, and I was like, I the way I work is I have, this exercise I called the Tea leaves, which is I take all the different factors in my life that things that have appeared and that seem random, but they're really crazy that they happen to show up, and I lay them out on the table and I sort of look at them, I use note cards to do it, and I'm like, how does this all fit together? And that helps guide me. That sounds like magic, but it's not. It's just like organizing around what's there in front of me. Right, and through that, I came up with this idea that's not certainly other people have had it as well, that we'll talk about maybe off this, that, off the podcast, but but basically, I'm pretty sure that AI is going to completely transform. I know it will transform the health care system, and it's going to be because each person is going to have way more agency in making their own health decisions, and so that will affect doctors.

[2:12:46 - 2:12:46] Patrick Lilley: Yeah, I agree.

[2:12:46 - 2:12:53] Justin McMillen: It will, and there's no way around it because with wearables and the ability to feed in AI.

[2:12:53 - 2:12:54] Patrick Lilley: Yeah.

[2:12:54 - 2:13:05] Justin McMillen: With all the information that it needs to understand you, and then for that I understand everything that was ever published again. Like you said though, there's a lot of issues with that. But perhaps you even apply this evolutionary model to

[2:13:05 - 2:13:05] Patrick Lilley: Yeah.

[2:13:05 - 2:13:17] Justin McMillen: To it as well, and then you now have this thing that's sort of working beside you as you go through life. It's able to pull all the information about you, knows more about you than any doctor ever has.

[2:13:17 - 2:13:18] Patrick Lilley: Yeah.

[2:13:18 - 2:13:22] Justin McMillen: Or ever will. Has superintelligence not like the version of what we say is superintelligence.

[2:13:22 - 2:13:23] Patrick Lilley: Not Johnny Depp,

[2:13:23 - 2:13:36] Justin McMillen: No. Yeah, but it's it has so much information that it can it can do computation and answer any question, and then with talking AI’s and avatars, you can sit up all night and have a conversation about...

[2:13:36 - 2:13:36] Patrick Lilley: Yeah.

[2:13:36 - 2:13:38] Justin McMillen: Your sleep, and it could ask you how you're feeling.

[2:13:38 - 2:13:41] Justin McMillen: And that could be logged in and that could be correlated to...

[2:13:41 - 2:14:11] Patrick Lilley: Yeah, more assessment, and I think there will be some doctors that resist this. But one of the things we have to remember is it's it's very easy to complain about physicians in the following way. Endometriosis, which affects probably more than 15% of women worldwide. You know, the tissue of the the uterus is growing outside the uterus very painful, and impairs fertility all sorts of things. The average time to diagnosis is ten years, and the only diagnostic method is exploratory surgery, and we're in 2025.

[2:14:11 - 2:14:13] Justin McMillen: Have you figured this one out yet.

[2:14:13 - 2:14:45] Patrick Lilley: We have the biomarkers for it. Yes. Yeah. So we want to do that, and every woman in my family has had it including both the daughters I mentioned. One of the things about it is you can blame the physicians and say, well, they're ignoring the women's pain and saying it's not serious and so forth. Then there's some of that, there are a lot of physicians that don't take women's complaints seriously and mark them down as anxious and give them, you know, some sort of benzodiazepines or something. But, by and large, they just the physicians haven't been given proper tools.

[2:14:45 - 2:14:45] Justin McMillen: Yeah.

[2:14:45 - 2:15:01] Patrick Lilley: This is not a disease that you can easily diagnose from just pure signs and symptoms. You have to do something else, and so our goal was to find the biomarkers in the blood. So once a girl starts missing school because of her periods, you know, it's a precursor.

[2:15:02 - 2:15:02] Justin McMillen: Yeah.

[2:15:02 - 2:15:17] Patrick Lilley: And you get her on a test right away when she's 17 years old and okay, you've got endometriosis, and now you start to think about what are the interventions to get her off of that before she gets into childbearing years is a, you know.

[2:15:17 - 2:15:39] Justin McMillen: Yeah, the tools are missing. I don't I think I mean, I told you about being sick and a doctor was a doctor who developed a stent that, Yeah. The way that I ended or it came out to be was that my main pancreatic biliary duct was closed, and this doctor at Cedars-Sinai, that speaks all over the world on this developed a specific type of stent that went in there

[2:15:39 - 2:15:40] Patrick Lilley: Just for this?

[2:15:40 - 2:15:44] Justin McMillen: Yeah, and it it ended up I was better within 48 hours.

[2:15:44 - 2:15:45] Patrick Lilley: That's crazy.

[2:15:45 - 2:15:46] Justin McMillen: Yeah. My whole life changed.

[2:15:46 - 2:15:48] Patrick Lilley: Such a simple mechanical solution. Right?

[2:15:49 - 2:15:53] Justin McMillen: Yeah, very mechanical. In fact it has like a little pigtail on and it slowly twists itself when you eat.

[2:15:53 - 2:15:54] Patrick Lilley: That's fantastic.

[2:15:54 - 2:16:02] Justin McMillen: I had to have two of them put in because the first one didn't last long enough. But but the point is, the doctors save lives.

[2:16:02 - 2:16:02] Patrick Lilley: Right.

[2:16:03 - 2:16:17] Justin McMillen: My, my dad's heart was stitched up by. They did an open heart surgery, a bypass with on his beating heart in Arizona, and I remember looking at the doctor's hands and I was like, I literally held them because I was like, I can't believe you just.

[2:16:17 - 2:16:18] Patrick Lilley: Yeah, it's it's remarkable.

[2:16:18 - 2:16:19] Justin McMillen: How did you do? You know?

[2:16:19 - 2:16:19] Patrick Lilley: Yeah.

[2:16:19 - 2:16:23] Justin McMillen: He's stitched up his damn heart stitched little hoses on, you know, it's like...

[2:16:23 - 2:16:25] Patrick Lilley: It’s, yeah, you can't even like...

[2:16:25 - 2:16:51] Justin McMillen: No, I can't even it's it, it's beautiful. So we need healers. The question is just where do they going to go and how is it going to work, and and they got to integrate and we need a revamp. Anyway, I think, you know, America has led the world in so much innovation. You know, a lot of people hate the US, but a lot of people came here a long time ago, and it was sort of the seekers and the, the most adventurous folks, and they all bread.

[2:16:51 - 2:16:52] Patrick Lilley: Yep.

[2:16:52 - 2:17:20] Justin McMillen: He ended up with this big explosion of innovation, and we have carried a lot of amazing things forward into the world. We've, you know, a lot of what's been born in this country has resulted in a lot of bad things, too. But, I mean, ending a lot of poverty and a lot of terrible things. Now we're the sickest country, you know, and I think the next thing is that we have to find a solution for that and then innovate and pioneer for the rest of the world, because typically the rest of the world to follow.

[2:17:20 - 2:17:23] Patrick Lilley: Well, they're already following us in our bad habits.

[2:17:23 - 2:17:23] Justin McMillen: Yeah, exactly.

[2:17:23 - 2:17:24] Patrick Lilley: Yeah.

[2:17:24 - 2:17:38] Justin McMillen: So we have to flip the script soon so that we can continue. I mean, we've done it through a lot of other means, but we need to to show the world what it looks like to focus on health and go from being the sickest country in the nation to the healthiest.

[2:17:38 - 2:17:38] Patrick Lilley: Yeah.

[2:17:39 - 2:17:54] Outro: Hey, guys. It would really help us out a lot if you liked and subscribed. We have a lot more episodes on the way, and if you like what you saw here today, put your notifications on and you'll know when a new one comes out, and, we're excited to get them in front of you guys and to hear what you have to say. So please leave your comments down below.