The AI That Caught What Every Doctor Missed — Genome-Powered Medicine Is Here | Dr. Don Brown

This is something I think you should be aware of, that this patient— I think we have one real-world example of a patient who was suffering from increasing leg pain and sleep disturbances that his physician thought was just a worsening of his longstanding restless leg syndrome. And the AI came up and said, test for arsenic.

Well, hey everybody, we're going to get right back to the podcast, but I want to share something with you that is very personal. Uh, you know, for decades I've been a practicing neurologist. I've been asking a single question, and what makes a good brain go bad? More importantly, obviously, what can we do about it? And that question has led me to write my new book, which is called Brain Defenders. Uh, it's now available for preorder. And in this book, I'm going to take a hard look at where we are today and offer up some challenges as to where we are today in terms of what we're being told. You know, despite all the attention on targeting proteins, misfolded proteins like beta amyloid for Alzheimer's, we still don't really have any meaningful treatments for diseases like Alzheimer's and Parkinson's disease per se. And we certainly can treat symptoms, but as it relates to treating the underlying issue, We're falling short. Many of these approaches manage the symptoms at best, but without really addressing what's driving the underlying process. And this science is evolving, but what's really exciting is the emerging research on what I've called the microglia. I didn't call it, but that's what they're called. These are the brain's immune cells. And when they're balanced, they are protecting, they are repairing. But when they're dysregulated, and we can talk about why that happens, and we do talk about it in the book. They can drive inflammation and lead to brain degeneration, often long before symptoms appear. And here's the empowering part. These microglial cells are deeply influenced by your day-to-day lifestyle choices, your nutrition, your sleep, your environment. And in the new book, Brain Defenders, I share both the science and provide a practical roadmap that'll help you calm inflammation, support supportive microglia, And really ultimately take control of your brain's destiny. And if you preorder your copy today at braindefenders.com, you can receive some special bonuses like a sneak peek at chapter 1, the full glossary for the book, and curated offers from my favorite brain health products, many of which I've discussed here on the podcast. This is a new way of thinking about brain health, and I truly believe that this book Hopefully it's going to change everything. I can't wait for you to read it. And I think you're going to be as excited as I am about the information you're about to receive. That said, let's get right back to our podcast. Well, hello everyone. Welcome to The Empowering Neurologist. I am Dr. David Perlmutter. Today we are going to tackle one of the most challenging questions of our time. And that is, what is the role of artificial intelligence in medicine? How artificial intelligence, how can it be used not as a replacement for the doctor, for the clinician, but as a powerful ally in our ability to understand human health and certainly human illness? My guest today is Dr. Don Brown. He is the founder of a company called Bioscope, and Dr. Brown has an educational background that bridges things like biology, data, and computational thinking. So he's in the perfect position to help us understand the role of artificial intelligence that it plays today, and certainly prognosticate in terms of where the future is going to take us. The foundation that he has is critical because the problems that he is addressing simply can't be solved within a single discipline. So it's the confluence of multiple disciplines brought about by the ability of artificial intelligence that really is giving us a leg up in understanding health and illness. Dr. Brown's work is grounded in the understanding that human health and certainly disease are grounded in the notion of interconnected systems, our metabolism, as we've talked about on this program at length, immune function as well, neural signaling, environment, and of course, deeply our genetics. And that these are all interacting over time. The problem is how do we embrace this interaction? We as healthcare providers have a difficult time with that. But Dr. Brown's education and research experience has shaped a perspective that gives him the ability to move on a reductionist view where we look at each of these things individually towards a more integrated view. Which is exactly what led to the creation of the company called Bioscope that we are going to investigate today. Bioscope was designed to make sense of complex and seemingly very divergent biological data in a way that is clinically meaningful and personally relevant. In other words, being able to identify specifically what's going on with the individual person, as opposed to hoping that a particular treatment, for example, that is globally used for a disease process might be effective in an individual. And rather than focusing on individual lab values or a specific diagnosis, this kind of platform uses artificial intelligence to help identify what are called patterns and relationships, and therefore early signals that are often missed in conventional medical models. In our conversation today, we're going to discuss how Dr. Brown's training informed his approach to health technology. And why the notion of fragmented medicine limits prevention and how AI, when it is thoughtfully designed, can support better insight, better questions, and ultimately the goal, lead to better outcomes. So this is a conversation about the intelligent use of AI in the advancement of human health. Let's jump right into our podcast. Well, Dr. Brown, welcome to The Empowering Neurologist. So glad you could be with us today.

Well, thanks for having me.

This is an exciting topic, and I think that there is some trepidation on the part of public at large and certainly in physicians' offices to fully embrace the notion of AI for whatever reason. But I think we're clearly moving in that direction already. So you saw what AI can do, and you decided to create this company. And we'll talk all about that. But first, tell me about your educational background, and that will sort of pave the way for Why a guy like you then gets into a company like, uh, creates this company like Bioscope?

Well, I always wanted to be a scientist. Uh, so after an undergrad in physics, I entered an MD-PhD program in biochemistry, fully expecting to spend the rest of my life working in a basement laboratory someplace. But after my first year of med school, I going into the biochemistry lab, I realized that I loved learning about biochemistry and still do, but I sucked in the lab. I was not cut out to be a laboratory scientist. So, I switched the non-medical part of my studies to the then young field of computer science. I got a graduate degree in computer science along with my MD and accidentally became a high-tech entrepreneur. I started a little software company while I was finishing up medical school. That ended up getting sold to EDS and General Motors. And then I kind of just kept going. I started 2 of the first 3 software companies ever to go public in the state of Indiana, the latter of which eventually grew to about 2,300 people, offices around the world, and was sold in 2016 for $1.4 billion. But strangely, all during those years, I had this sense of lack of fulfillment because I had never done anything in the life sciences with those companies. And so that led me back to grad school. Now, maybe in my mid to upper 50s, I stunned my kids. I've got 8 of them and told them I was going to enroll at Johns Hopkins in a master's program in biotechnology. I was twice the age of the next oldest person in every class. You know, this grandpa, Competing against all these East Coast hotshots. But, I don't know, it just relit the fire in me. I went back through biochemistry and then molecular biology, cell biology, advanced cell biology, and then immunology, advanced immunology, cancer biology, stem cells, genomics, bioinformatics, and just had a blast. So when my company was sold in 2016, I did 2 things. One, I donated $30 million to my med school alma mater, the Indiana University School of Medicine, to form an immunotherapy center. And they're doing great work with cancer immunotherapy, CAR T-cells, and so on. And I started my first real life sciences company called Lyphomic. That was kind of the precursor to this company, but I'll take a breath there and see if you—

No, I'm following the story. Let's keep going till we get to Bioscope because I'd like to get to when there's probably in your mind this convergence of datasets that you realize, and there is a tool. So let's keep going. This is great.

Yeah. Well, so I started a company called Lifomic in 2017, immediately after I announced it the day the transaction closed on my previous company and really wanted to build a cloud platform where we could aggregate everything about a patient. And there we were primarily focused on cancer patients. So we were ingesting everything from the electronic medical record, but also germline and somatic whole genome sequencing. And it started being used for some of the most difficult cancer cases at Indiana University Health. Triple negative breast cancer, pediatric sarcomas, multiple myeloma in particular. But what we were missing was AI. So we were trying to code really complex rules to analyze this stuff. And a couple years ago, we were approached by Tony Robbins and Peter Diamandis, who had started a chain of longevity centers called Fountain Life. And they had a great clinical experience, but their IT infrastructure wasn't so hot. So they bought us. They bought Lymphomic. I didn't take any operating role and thought, you know, I'm done. I'm retired. I live out here in Park City, Utah. I love mountaineering and rock climbing and all that sort of stuff. But in the spring, I was approached by some people I know out of Harvard about buying and relaunching 23andMe. And as a courtesy, I did some due diligence.

Due diligence.

Quickly decided I didn't want any part of that mess. I'm much more of a startup guy than a turnaround artist. But I learned 2 things that set me off. The first was that the cost of whole genome sequencing, which had been stubbornly stuck for us at Lifomic at about $1,000, had fallen to around $300. That set off all sorts of alarm bells. But the second thing I did was take the time to do a deep dive into the state of medical AI. And I was blown away. And it recalled for me kind of an offhand remark I'd made when I'd given the commencement address to the IU School of Medicine back in 2017. And I told these kids graduating, these incipient doctors, that before too long in their careers, they would be working with artificially intelligent colleagues. And it hit me, I could build this now. And by God, I want to. I want to be the one to build it. And so I told my kids who had begged me to stay retired, I said, I can't sit this out. You know, this is just too exciting. And so, you know, one of the benefits of having had a long and, you know, a relatively successful career and having treated people ethically is that if you kind of put out vague signals that you're going to do something, you quickly get flooded with interest.

I know that. I know that drill.

I had that. I mean, it's tremendously gratifying. And so I did, I put out some vague signals and I had wonderful people I had worked with in the past who said, if you're serious about this, I will take a cut in pay to come be part of it. And so it kind of goaded me into it. So I just pinch myself because this has all come together so quickly. So I incorporated Bioscope back in May, and we started coding in July.

Of 2025?

Yes.

Okay.

And we're up to about 30 people now and experts in everything from high-end cloud architectures to bioinformatics. And we got our— lured our security expert back away from Shopify, and she took a huge cut in pay to get back into the startup realm. And we've just cranked. And, you know, first step was kind of rebuilding what we had built at Lifomic, only better, an ultra-secure cloud-based repository where we could ingest everything from a person's electronic medical record, blood tests, images, notes, but also a complete whole genome sequence and a whole microbiome analysis. And, we engineered it. We're now off working on being able to ingest the epigenome, just the complete sequence of sites where a genome has been methylated that controls which genes are switched on and off, the proteome, the transcriptome, all these big datasets. Probably my favorite is the immunome, The set of immune repertoire receptors that our T and B cells make that determine what our immune systems are able to recognize.

Hey everyone, we're going to get right back to the podcast, but I have an important message for you. If you're caring for somebody with Alzheimer's, any other form of dementia, or even what we call mild cognitive impairment, or maybe you've received one of these diagnoses yourself, Or if you're focused on preventing cognitive decline, I want to speak directly to you for just a moment about some serious research that's going on, looking what we call 40 Hz light and sound simulation. We've been actually talking about that on the program. Here's the challenge. Not all 40 Hz light devices are the same. Most use what's called stroboscopic light. And that's the type of light that flashes, then you can see the flashing. And that can cause nausea, it can cause headaches. And if you can't tolerate that, you won't use it. So, there is a company called OptoCeutics. They've solved this problem with a patented technology that still gives you the 40 hertz light flashing, but through light that looks and feels quite normal. It's the device I actually have on my desk when I'm working. And that's why this company sees a 94% adherence rate and significant improvements across various metrics including mood, energy, focus, sleep, and memory. And the light is called the Evy light, E-V-Y. You can use it as I do when you're working on your computer, when you're reading, when you're watching TV, eating breakfast, whatever. It becomes really part of your day, not just another burden. It's kind of passive in the background. So I recommend giving it a try for 90 days. And if you and your family don't see value in it, then you can return the device for a full refund. They cover the shipping, Both ways. And you can learn more about OptoCeutics at optoceutics.com/perlmutter. You can use the code to get a $200 savings. The code is PERLMUTTER, my last name, 26. This is a device that really is risk-free in terms of getting involved with 40-hertz stimulation. Important information. Let's get right back to our podcast. Let me stop you there for just one second. I want to unpack something just a little bit for our viewers who Maybe a little bit overwhelmed with what you just covered. I think one of the most compelling parts is, I mean, most people have had their genome sequenced. They know what they are, they might be at risk for. But what you just mentioned, I think, is a little bit newer technology where we're now able to look at the epigenome. In other words, what are those controlling factors that regulate gene expression or gene suppression? And I think the empowering part of that, and I know we're going to get there, when you tell us what you're able to do with the data, is that we're able to change that based on certain interventions, many of which are actually lifestyle interventions, though there may be pharmacological or supplemental inroads to modifying our genetic, our epigenetic expression. So that's relatively new. The other thing you mentioned is you're inputting data with reference to the microbiome. Is that metabolomic data or is it— 16S RNA? What are you inputting with respect to the microbiome? And then, explain it a little bit more for our listeners.

Well, in each of our cells, we have the DNA that contains the recipes for the genes. That's what genes are, recipes for about— we have about 20,000 genes that code for proteins. But, in our guts and on our other surfaces, We have at least 100 times more genes from the bacteria that are along for the ride. And it's really only been within the last, I don't know, 10, 20 years that we've come to appreciate the role they play in our health. And in fact, it's looking like some of the diseases of our brains, like Alzheimer's and Parkinson's, may originate in our gut, or at least be heavily influenced by the composition of those bacteria in our guts. So now, we're able to inexpensively sequence not just our own nuclear genomes, but the genomes of all these bacteria, and then use that information to characterize them, to analyze what is the relative abundance of different types of organisms because some are good. There are organisms like Akkermansia that help form a protective barrier along our gut lining. And, there are the baddies like C. diff, Clostridium, and others. So, being able to first characterize what bacteria are there and then you alluded to What are they doing? What metabolites are they creating? Because certain chemicals they produce, like butyrate, are beneficial to our health, but others aren't so good. And so, yeah, that's a new thing, being able to take that information along with our information about our nuclear genomes and combine all this and interpret it via AI.

What about mitochondrial genetics?

Well, I mean, that too. I mean, that's hugely important and a passion of mine. Now, fortunately, we get that in doing a 30x or kind of a 30 times sequencing of the whole genome. We can pull out mitochondrial DNA, and that too has a huge influence on our health.

Okay, so Dr. Brown, here I am. You've analyzed me from stem to stern. You've sequenced my genome. You've looked at my microbiome. You've looked at every metabolic parameter known to man. You have this incredible array of deep information about me. So what? What now? What's the benefit of that? I'm teeing it up here for you.

Oh yeah, you are. You're doing a great job, David. Well, Yeah, you're right that if I just handed the average doc, any doc, a USB key with all that information, they would look at me as if I were crazy. Like, what do you expect me to do with this? But what we're able to do with AI is kind of increase the signal-to-noise ratio. We're able to analyze that information along multiple steps and pick out the pieces that are most impactful to our health. And then what we do at Bioscope, and this is kind of my little invention for better or worse, I've flown high-performance aircraft for a number of years, bopping along as a single pilot at 30,000 feet in a nice fast little plane. And they all have what's called a synoptic view. They all have this graphical representation of the aircraft and its systems. And so, at a glance, you can see when something goes wrong. Is it the fuel valve from the left wing? Is it hydraulics, whatever? So, what we've done, well, we kind of took that ethos and have applied it to human health. So, our interface is pretty cool. I encourage your viewers to to visit our website, they can see. But we start off by creating a 3D wireframe from a couple of pictures of a patient, a person. And that alone to a physician conveys a lot of information. But we also characterize 15 different dimensions of health, everything from metabolic and cardiovascular on down to nutritional health, environmental health, And the final one is healthy longevity. And we pass all that information that we've gathered through our AI layer, which is really a sophisticated committee of AIs that are constrained. Well said. We have an API that allows us to access something crazy like 100 million academic papers from all the top journals. And we also constrain them to use highly curated databases like ClinVar and other, MONDO, that are maintained by the NIH and leading universities so that they can't go out and hallucinate or make stuff up. We really try to box them in that you've got to justify what you come back with, but The first thing we ask them to do is look at each of those 15 dimensions of health and just assign a red, yellow, green indicator. So first step, you know, using all the available information as complete or incomplete as it might be. And then for the physician, it makes their job much easier. You know, they can look and see, David's heart is great, but his gut health, it's red and there's something going on. And so, they can click and drill down into each of those dimensions.

But, what AI has allowed you to do is really solve for incompatibility. It used to be differences in electronic medical records, EMR, different lab values, different indices that are used in different countries in terms of how they measure vitamin D levels, for example, or A1C or blood sugar. And this allows you then with AI to input data and let the AI just deal with it and then influence these pattern recognition images that you create.

Yeah, that's exactly right. We tried to do this with just brute force coding at Lifomic and realized this is impossible. You know, this is— it's going to take thousands and thousands of developers to try to do this. And so, it's mind-boggling to me with what has happened in just a few short years. What we are doing would not have been possible a year ago. That's the staggering truth of it that reflects the pace of development of AI.

So tell us what we're looking at right now.

Well, that's a representation of that ultra-secure cloud repository that I mentioned and the way that we can ingest patient data, anything we know about a patient, and then in real time, When the physician goes to bring up a patient's record, what we do is have the AI layer interpret all that information in light of the latest medical knowledge.

So you mentioned a year ago, this probably couldn't have happened. And now you're able to input just all this seemingly disparate type of data. And what is the interface then? Is this something that the patient inputs or the clinician inputs?

It can be both. We have— I mean, you know the importance of patient history. So we have a very thorough sort of interface that allows patients to enter information, all the supplements they might be taking, of course, medications, medical history, family history.

But the narrative of the history, The intake history, the narrative that the doctor writes or the scribe puts down, that gets inputted and analyzed by AI as basically data.

It's critical data.

Yeah,

yeah. Yeah, we were taught that in medical school, the importance of a good history. Correct. And that hasn't diminished.

Well, the other thing we were taught was it's not so important what the disease the patient has, but who is the patient? who has the disease. And I think you solve for the second, uh, statement that this is personalized medicine to the highest degree that has ever been, uh, ever been seen. And the interface that patients have is via an app.

Uh, it's a, a web, uh, interface. Uh, I don't know if you can, uh, bring it up, but, uh, even if you, uh, just bring up our, uh, website, you'll see it.

We're looking at it now. Tell us what we're looking at.

So what you're seeing is that I mentioned that 3D wireframe representation of a patient's body, what we call body habitus in medicine. And then you see the 15 dimensions of health with that sort of stoplight metaphor where each one is marked red, yellow, green across the top. The physician can click in to get more information. Even, it's so cool now that AI can pull out and create a nice timeline view of a patient's medical journey organized by diagnosis. So the physician can, at a glance, see the medications the patient has been tried on, The biomarkers that are pertinent to that dimension of health, say hemoglobin A1C for type 2 diabetes, but also imaging studies. But then I think the coolest thing, if you look over at the right-hand side, you'll see a panel where the AI is surfacing observations. It's saying, Dr. David, This is something I think you should be aware of, that this patient— I think we have one real-world example of a patient who was suffering from increasing leg pain and sleep disturbances that his physician thought was just a worsening of his longstanding restless leg syndrome. And the AI came up and said, test for arsenic. And his physician was like, Why would I test for arsenic? And, the AI is able to explain. This guy is homozygous for a slow-acting version of a gene called AS3MT that methylates arsenic, which is critical for its elimination from the body. Right. And, he lives in an area that's a former mining town where the soil is known to be contaminated with arsenic and cadmium. His symptoms are consistent with neurological problems. And perhaps the coolest thing was that the AI noticed that he was becoming insulin resistant despite being very trim and very active. And his primary care physician was just scratching his head like, I can't explain this. And it's because elevated insulin— elevated arsenic induces insulin resistance.

Hey, we're going to get right back to the podcast, but I do have an important message. You know, for decades we've been told that our genes are basically our destiny as it relates to health and that everything's locked in place predicting even our future brain health. But let's be clear, your DNA isn't a verdict, it's a roadmap. And I'm talking to you today about 3x4 genetics. I actually described them in the new book, Brain Defenders. And this is a company that helps you understand your genetics, what you've inherited, and I think more importantly, how your daily choices can influence how those genes are actually expressed. In other words, how you can control your own genetic expression. And you know, that matters for your health and it really matters for brain health, including things like memory and mood and cognitive resilience. You may carry genetic markers associated with cognitive decline, like As we've talked about APOE4, but possession doesn't equal expression. Your lifestyle, your nutrition and metabolic health, all the things that we talk about here on the podcast can powerfully influence which genes are turned on and which stay quiet. The 3x4 Genetics Test looks at how your genes impact brain function at a cellular level, and which then through the 3x4 Health Journey, which they provide you, guides you through a clear and a science-backed program focused on inflammation, methylation, oxidation, oxidative stress, things that we routinely discuss here on the podcast that are the key drivers of basically what makes a good brain go bad. And we have spoken with the team at 3x4 Genetics, and they were able to get a special offer for our podcast community. So head on over to 3x4genetics.com. /drperlmutter and get that special offer and start really your personalized brain health journey today. Let's get your genes working for you and let's get back to the podcast.

So to have, I mean, it's what I described back then.

Yeah, I fully get it. And here's when it became clear to me. I was outside the lobby of a hotel recently in Los Angeles. And a car, somebody had ordered a taxi and the taxi pulled up. These people got in and went on their merry way. There was no driver in the car and they were as happy as could be. They felt as safe as could be. And at first blush, oh, I need the driver, right? Why? Why do you need a driver? You want to get from A to B in the safest way possible and the quickest. Right. And if you look at the statistics, This driverless car has a driving record. The whole notion of autonomous vehicle has a driving record that is incredibly superior to anything a human could do.

Yeah.

And you get over it very quickly. You realize that if I get in this car, my chances of getting to my destination without getting in an accident are far greater than an Uber driver. I'm not knocking the Uber driver because they're probably a better driver than I am. But that said, This is taking advantage of technology, and I think people get over it very quickly because at every conference I go to now, the opening remarks are about AI and what are we going to do and how are we going to incorporate AI into our medical practices, and is it going to take the place of doctors, on and on, and what are we going to do, and we need to handhold, we need to be compassionate. And this is nothing like that. This is the most incredible tool that we absolutely must embrace today. I can't read the 10,000 journal citations a day on any particular topic. I can't read 10 and fully understand them.

No.

And the speed at which I'm able to read them would take me the entire day, nor am I able to synthesize any data that I've, to any significant degree, that I would pick up from seeing other patients or that other people are seeing in dealing with their patients. So, I get it loud and clear. So tell us how it's going.

Unbelievably. I mean, this has been the most fun thing I have ever done in my life. It's so gratifying, you know, at this point in my career to just be playing in this field, much less being at the leading edge of it. So we're primarily marketing to concierge physicians. You know, so they have patients who are paying them, in many cases, a lot of money to keep them healthy, to spot disease processes years before they manifest in symptoms, and to be able to seek out the most advanced personalized treatment available when disease does develop. And so they've been very receptive to this. They see Now, you know, smart doctors see that this is the future. We've talked about precision medicine for a long time. Well, precision medicine really starts with knowledge of your genome. That's what makes it precision. And so being able to have every medical decision informed by a person's genome is not something that a human being can possibly be expected to do. Yeah, you know, you're a brilliant guy. I know. But to give you 3 billion base pairs, you know, 20,000 genes, and say, David, interpret all that for me, you know, my blood test and everything else in light of that information, you know, no human being can be expected to do that.

I feel for these clinicians these days. You know, you go to the conferences, you see these sequencing companies that will sequence your patients for you. And then everybody gets the results. The patients get the results and you get the results. And the printout, it's, you know, they spit out a monograph of 3 or 4 pages that this patient has these genetic variations. And as such, it's MTHFR, whatever it may be. And therefore, you know, you might want to talk to them about their B vitamins or something like that. But—

Yeah.

To think that there is a way to incorporate that with their existing medications that may be depleting B vitamins or other health issues like cognitive decline as a consequence of elevated homocysteine because of that MTHFR polymorphism, to take all that information and then present that in such a way that it's actionable. Again, um, so I reached out to you. I reached out to you because I heard about this. thing, this Bioscope thing, uh, in the context of other people who are leveraging AI as an ally in the practice of medicine. And that's what I wanted to learn about in connecting with you today. And it's, uh, I'm having a tough time containing my enthusiasm for what you're doing. So concierge doctors around the country are now offering this to their, uh, to their patients. The patients then sign on. upload information with the assistance of their doctors, but getting information from their other doctors, previous labs that they have done, they track those down. And once it's done and uploaded, one would assume it's obviously ironclad in terms of being a safe repository. And we'll talk about that in a moment. But then further uploads, do they happen seamlessly, automatically? Or do you have to just continue to massage the upload every time you have a lab value generated?

Well, it depends. So, we have a website where you can upload, physicians can upload anything for their patients. And, it's pretty wild how good the ingestion process has gotten. You can take scanned images, pictures of images, And the AI can faithfully extract, uh, the information. I mean, again, this wasn't possible a year ago.

Yeah. So the issue was, how do I get my data into this platform? Because my data may not be LabCorp or DSI or whoever has my data may not want to transmit it. But, you know, we've seen overall the past couple of years data become fully available to patients and in various forms. You're saying we could The individual could send you images that could be incorporated. AI can look at the image and say, you know, that's a glucose value. We're going to put it, standardize it and put it into the data.

Yeah, it's wild. Now, in addition, we have built interfaces to some popular EMRs.

Good.

But unfortunately, there are hundreds out there. So, We're kind of taking them one at a time, but some of the big ones like Elation, Athena, Epic, there are a few biggies that we've done or are in the process of doing. But we did make it so that if a doc can get something in a file form, it doesn't matter what the file format is, a DICOM image, a PDF, a JPEG of a picture. Yeah. picture of something, they can upload it and we can extract that information.

All right, let me push a little harder here.

Yeah,

rather than upload an MRI report, could your software actually evaluate the MR images?

Yes, yes. Our team is very proud. Their image analysis is one of their strengths. So one of the features we added recently was from a brain image being able to quantify the anatomic regions. So hippocampal volumes and so on. And then we are working with partners who are coming to us and saying, I've got a particular ML model. One that comes to mind is a company that is able to very accurately predict predict the due date of a baby from an ultrasound image. And so they and others are coming to us and saying, we'd like to build this into your platform for this particular scenario. In that case, a pregnant woman who would like to know, you know, when is my baby coming? So we do a lot of that image analysis ourselves, but then we're partnering with the many companies who are springing up every day who are doing more specialized image Well,

that said, you know, that technology is evolving very quickly. How does your system learn to interpret and then incorporate new technology?

Well, I mean, for one thing, we've built an extensible platform, part of good software engineering. It's what we did at Lifomic. So we made it so that we can take additional cloud services, plug them in and then interface seamlessly to those from our interface. And then our own team is constantly— we'd license huge sets of images that we can continue to train on for other sorts of image analysis, say quantifying liver fat from an MRI image or an ultrasound. There's just so much valuable information from those medical images.

Well, that's going to be, you know, that's certainly underrecognized in that particular, you know, liver fat these days. What is the biggest misconception, do you think, that patients have with respect to the AI interface in their healthcare moving forward?

Well, one misconception I think both patients and physicians have that I just want to mention real quick, you kind of alluded to it, is What we do is not some static test, right? So you mentioned that you can go out and get a genetic test, you know, pay a few hundred bucks, get a report, and, you know, kind of file it away someplace. That's not what we're doing. We're sequencing that genome. And then for the rest of time, the rest of the time the patient is in our system, everything is informed. By that genome. Every week, the NIH, there's a database called ClinVar of gene disease associations that's updated weekly. So that's the first notion I wanted to dispel, that this is not a static report. It's very much a dynamic system that is learning, evolving, making use of all the latest information. I think for patients, the biggest worry is that they've gone out and used ChatGPT or things like that, and they worry about hallucination. What if this tells my doctor to cut off my leg? And like I said, we've taken great pains, first of all, by having kind of a committee of AIs, and then using that corpus of academic knowledge that I mentioned through this API, and then also showing the chain of reasoning so that when our AI comes back and says, I recommend that you test for arsenic, you can see what were the logic steps that you went through to arrive at that conclusion. And I think the thing that's going to blow people away is what we're building next really is that artificially intelligent persona. We have even given it, her, a name. Her name is Lumina. Lumina. She has a self-model of who she is, what she knows, what her principles are. And the physician can engage in a dialogue with Lumina. Lumina is the AI persona intended for the physician. We're in the process of speech enabling Lumina and actually a whole set of AI personas so that what's going to happen, and very quickly, is that you're going to have, say, a 65-year-old lady as a patient who's kind of frail. She's suffering from diminishing bone mineral density and reduced muscle mass. And you're going to say, Margaret, this isn't good. We really need to work on a program for you that's going to make you stronger because that's going to help everything in your body. So what I'm going to do is assign you an AI persona. His name is Cairo. And I'm going to give Cairo a 6-month program for you. And I want you to have an Alexa or something by your bedside. And every morning you'll check in with Cairo and he's going to say, Margaret, we're going to get 30 grams of protein in your body at breakfast by hook or by crook. It's either bacon and eggs or it's a protein shake or, you know, whatever it is. And then at noon, we're going to do some lightweight training. And before you go to bed, remember that you've been prescribed some peptides that will help your body secrete more growth hormone, and that's going to make it easier for you to put on muscle mass. And so Margaret is going to be able to converse with that AI persona, ask questions, ask why, And if something needs to, it goes to the physician to be alerted to the fact, oh, Margaret strained her back. Her form doing those squats wasn't good. We need to work on that.

So

that's where all this is going.

That was going to be my next question. I was going to say, what does it look like 5 years from now? But as we've been talking, I think even 1 year from now, do you think there'll be avatars like that to interact with patients?

Yes.

The current interaction is via the physician being that intermediary?

Yeah. So that's the next thing. The next thing is what we've done is we've assigned different personas to different logins. So there can be a patient login. So if a physician chooses, as I said, we sell to physicians and then it's up to them. But if they want to, and most do want to, they can give a login to a patient. We know that that's a patient logging in.

Of course.

So what we do behind the scenes is we have a different set of AI prompting that we load. So the patient is seeing exactly the same interface, but it behaves totally differently because The AI has been conditioned, has said, you know, this is a patient. Don't give them a definitive diagnosis. Don't tell them about a cancer biopsy path report that their physician hasn't discussed with them yet. And so absolutely. So they get to use the same interface. They can ask questions. They can. Ask what supplements should I be on, you know, based upon, you know, my genetics and the medications I'm taking. You know, how do I go about getting stronger, you know, and reversing reductions in my bone mineral density?

And how quickly have physicians jumped on to utilize the technology?

It's been unbelievable. We exhibited at the A4M conference in Vegas. That's

where I learned about you.

Oh, okay. Yeah. And we were mobbed. I mean, it was just so cool. I've been in business for a long time now, and I've never seen anything like This, that the timing is right, yet there are kind of barriers to entry. The big guys like Microsoft and Google, they're out there trying to apply AI to the electronic medical record. Well, we're off with the genome, the microbiome. We're working on the epigenome and the proteome. So we're several steps ahead. Because as good as AI is, we can't just feed it. I can't just give ChatGPT my whole genome sequence of 100 gigs of information and say, what should I do? It'll choke. The context window for these AIs ranges from about 200,000 to 2 million tokens. And so it means that you have to have a team of people who understand the problem well enough to, like I said earlier, increase the signal-to-noise ratio. Take that mass, that 200 gigabytes of data from a whole genome sequence, and turn that into 500, 1,000 tokens that really represent the most salient information. That's the trick.

On your website, do you have a way of finding that concierge practice or whatever the practice may be via zip code so people can avail themselves of this?

Not yet, not yet, David. This has all happened so quickly that we're just trying to— sometimes I feel like the dog that caught the bus. So we're just trying to meet the demand of the physicians You know, who have signed up and rolling out to their patients. But stuff like that is coming. You know, we have a number of physicians who are now approaching us and saying, I want to become a concierge physician.

Just so I can have Bioscope.

Yes. And so they're saying, please just give me kind of a concierge practice in a box with the backend, the frontend, everything that I need. So we're busy putting that

Understood. That's very exciting. Well, this is incredibly important information, exciting information. And let's plan to talk in a year or two, see how far we've come. But congratulations. And also, let me say thank you because this is really, really going to move the needle aggressively in the right direction in terms of healthcare. So you're doing great stuff.

Well, thanks, David. I really appreciate it. That means a lot coming from you. But like I said, this has just been the dream of a lifetime to be able to participate.

Great. All the best. We'll talk soon.

All right. Thank you.

Well, we want to thank Dr. Brown. This has been an incredibly thoughtful and certainly a timely conversation. And what stands out to me is how clearly his education and a systems-based training show up in how he approaches health. It's not looked upon as a collection of seemingly disparate parts, but as a living adaptive whole. Bioscope, the company that we talked about, that it reflects that philosophy and it reflects it really quite beautifully. Rather than oversimplifying biology, I think it welcomes the diversity and complexity of biology because it uses artificial intelligence to respect this complexity and help us navigate it certainly more intelligently. And it's that distinction that really matters, especially as we think about things like prevention and all of this in the context of personalization and the long arc of chronic disease development that we've talked about a lot on this program. And as we've talked about today, the future of healthcare won't be determined by how much data we collect, but how well we're able to work with that data, how we can synthesize it, how we can analyze it. And certainly, uh, the use of artificial intelligence has become our most powerful tool in dealing with the incredible number of data points that are being generated for each and every patient. Uh, Dr. Brown's work represents A meaningful step towards a more integrated and ultimately more human-centered model of how we care for individuals, one where technology is enhancing our insight without replacing wisdom. So what an experience. This is where the future is going and is certainly where the future already is in many respects. So thank you for joining me today on this episode of The Empowering Neurologist and for helping shape the future in which a world of artificial intelligence becomes a Powerful ally, uh, in our quest to stay healthy and live a long and healthy, productive life. I'm Dr. David Perlmutter. Bye for now.