Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Neuroscience's Computation Revolution | $529m Precision Neuroscience President Dr. Craig Mermel
0 Plays2 seconds ago
In this episode, we speak with Dr. Craig Mermel, President and Chief Product Officer at Precision Neuroscience, a company building non-invasive brain-computer interfaces that translate neural signals to restore neurological function for patients.
Craig began his career as an MD-PhD at Harvard, where he studied at the intersection of biology and computation during the early genomics revolution. We talk about how that experience shaped his view that biology is increasingly becoming a data problem, and how that perspective carried into his work at Apple and Google. He breaks down what brain-computer interfaces actually are, the tradeoffs between different approaches, and why you don’t need perfect neural data to drive meaningful outcomes. Ultimately, this conversation explores how turning signals from genes to neurons into computable data is shaping the future of healthcare.
Recommended
Transcript
Introduction to Healthcare Theory Podcast and Dr. Craig Murmel
00:00:00
Speaker
Welcome to the Healthcare Theory Podcast. I'm your host, Nikhil Reddy, and every week we interview the entrepreneurs and thought leaders behind the future of healthcare care to see what's gone wrong with our system and how we can fix it.
Dr. Murmel's Career in Genetics and Engineering
00:00:15
Speaker
Today's guest was one of the first to witness how biology supposed to become computational. After getting a dual MD and PhD in genetics at Harvard University, Dr. Craig Murmel, now the Chief Product Officer of Precision Neuroscience, ended up working across the industry, both at MassGen, the Broad Institute of across Transnational Research,
00:00:35
Speaker
but even more in engineering at Apple, working on computational problems, and then at Google as a senior staff
Precision Neuroscience's Mission and Technology
00:00:41
Speaker
research scientist. So he's seen the research and the implementation side of, and today he's working on brain-computer interfaces, helping use neuroscience and signal acquisition technology to help alleviate paralysis and ALS in patients at Precision Neuroscience, which is $500 million dollars company building non-invasive technology.
00:01:00
Speaker
Hi Craig, thank you so much for coming on the podcast and welcome to The Healthcare care Theory. Thank you, it's great to be here. Of course, and before we get into precision neuroscience, I want to start with your background a little
Dr. Murmel's Educational Background and Career Shift
00:01:10
Speaker
bit.
00:01:10
Speaker
i mean first, tell me a little bit about that. It's definitely unique and I think when you got your PhD in biology, it was around that time with the Human Genome Project that we saw biology transition from something strictly in the wet lab to the dry lab more and more with digital digitalization of genomics. So when going through your PhD, what were the trends that you were seeing at the time?
00:01:30
Speaker
and how did that influence what you're pursuing yourself and and kind of the things that you're overall most interested in in this new field of biology at the time? Yeah, great. I'm happy to start there. And actually, before I even get into my program, just a little bit of background about about me. So um I've always been somebody who has had two interests in life. I loved math and you know related to math, computers and technology, but all ah really I knew my career I wanted to spend in biology and and healthcare. um So I was a math major and a biochemistry major in undergrad. And um you know when I went to go pursue my doctoral training, I joined the joint MD-PhD program at at
Evolution of Biology to Computational Methods
00:02:10
Speaker
Harvard. And I was particularly interested in trying to figure out like
00:02:14
Speaker
Where, where, you know, the question you asked me, where would somebody with these dual interests apply? And to be honest with you, it really wasn't clear in the early 2000s. um A lot of the biology labs, you know, biology then was very much a wet lab discipline. ah And, you know, that required a lot of, you know, a lot of your time was spent setting up experiments, running them,
00:02:33
Speaker
ah the results were very analog and not very quantitative. And usually because of the cost and the time that it took to get data, you were really in a data light regime. um And so it took me a little bit a while. I started my PhD in 2004, so this was like early mid-2000s to try to figure out where that application would really be.
00:02:52
Speaker
um I found it at the time, you know, across the the river in in Cambridge at MIT at the Broad Institute, because, you know, for context, we had just sequenced the human genome. um And obviously, that was a lot of data. But that was like, you know, and it took us 15 years and a few billion dollars to get the first genome. But it was very clear already then by the the early that the technology to get the data out and... It was just on ah on a purely you massive exponential trend. um So you know many of your listeners may not be aware of this, but the cost of sequencing you know a genome went from
00:03:31
Speaker
um you know millions, hundreds of millions of dollars to you know about $10,000 in the course of like the six years that happened to coincide with my PhD. um And so what that meant was, you know when I started my PhD, you could start to get data, but it was still pretty expensive, pretty manual.
00:03:50
Speaker
But by the end of that PhD, for the same amount of time and effort, you could get 10,000 times more data.
Moore's Law and Its Impact on Biology
00:03:55
Speaker
And that meant pretty profound, obviously, one, there was a ah need for the ability people like me to come in and think about you know how to do this in a way that wasn't just like looking at the data, but actually building algorithms and computational approaches to scale our analysis and at the same rate that the the data was was accumulating.
00:04:14
Speaker
um But the actual the actual changes of living through that was really profound. I'll give you just a concrete example. When I started my PhD, if I had proposed my thesis that I'm going to like discover ah one gene that might be involved in the development of cancer, that would have been like my PhD committee thesis would have said, that's ah that's pretty aggressive. You may want to be a little bit less ambitious so that you can graduate.
00:04:39
Speaker
By the end of my PhD, because of the explosion of the amount of data, and at the time there were also public consortiums like the Cancer Genome Atlas Project that allowed us to sequence hundreds or thousands of cancers of each type,
00:04:51
Speaker
we could literally in an afternoon take all that data run it through algorithms we developed and we would get a list of hundreds of genes on which were all of the known genes related to that cancer as well as like 30 new genes that had never previously been discovered and that was routine and we were doing that all the time and so that just that experience of living through that really was for me a profound taste of what happens when biology and computational sciences um survive. And from that point, I was kind of hooked. I was like, all right, this is a real thing.
00:05:20
Speaker
um And the lesson I really learned from that is that profound things happen when, you know, Moore's Law and biology can
Entrepreneurial Journey with Symbiotics and Apple Partnership
00:05:27
Speaker
intersect. And um part of my career since then has been to try to figure out where those where that convergence is happening and try to position myself to be ah at the leading edge of of of those convergences.
00:05:41
Speaker
Yeah, I just find that so interesting. I'm taking a class in genetics right now, and it's like I almost take it for granted. and I mean, back then, math and biology it wasn't really a clear intersection. we have And now we have an entire building at the University of Chicago dedicated to computational biology, computational neuroscience.
00:05:59
Speaker
So can't imagine how things how different things were back then, but you've had some work at the Broad Institute, which of course has been famous for blurring the lines between basic science and real-world transitional impact. And so working in that area, I know you started to call started a company called Symbiotics. I would be curious to hear, what did that look like? I mean, what was your...
00:06:18
Speaker
why did you want to go into entrepreneurship? I mean, getting an MD, PhD, you had a lot of options, but what were you hoping to build and what inspired you to do that? And the thesis behind taking a huge step into starting symbiotics? Yeah, it's a great question. um And I think it connects, you know, the the the work i just described to you in the the period i was in genomics, because symbionics was founded by myself and Ben Rappaport, who's one of my MD PhD classmates.
00:06:46
Speaker
um good friends and also the co-founder and my partner here at Precision Neuroscience. So, symbionics is really ah an important stepping stone. We got interested in symbionics really because while we were living through this transformation in genomics, um there was a a belief at the time that the that everyone understood something unique was happening, but there was also a belief that that was unique to the genome.
00:07:09
Speaker
And that, of course, the genome is special. It's the carrier of a biological information. So, of course, it's going to be computable. But Ben and I believed that there was something deeper. And we both had an interest outside of our medical training and academic work. We were both interested in running. And so we started experimenting. This was you know late 2008, 2009, 2010 with wearable.
00:07:31
Speaker
technology, just as users of that technology. And we saw the ability at the time to measure you know how many steps you were taking and then to measure physiologic variables. um And we saw that that trend was going to continue, that there were going to be devices that were incapable of measuring more things, more people were going to be using them.
00:07:48
Speaker
um But what we saw was lacking was the, no one was really doing anything to give you novel insights. You might able to say like, okay, I took 10,000 steps, but was that good? Was that bad? um So Symbionics was really born out of this idea that like we could potentially use the increasing amounts of data that would be coming from pervasive wearable technology to try to do meaningful, give people meaningful insights into health.
00:08:13
Speaker
um We did not intend to really, I've never intended to be an entrepreneur to start a company, um but the origin of that was simply this was work we were doing outside of our, you know, day jobs as students and
Insights from Product Development at Major Tech Companies
00:08:27
Speaker
physicians. And so we kind of decided to start this company as a sort of home for the IP we would develop and as a way to see if we could do something interesting and potentially scale um to ah to to have a real impact, to put a tool out in the world that people would use. That was really our goal. We wanted to like enable people to gain new insights into their health and fitness.
00:08:49
Speaker
We did not have a great understanding at the time we started this, what that would entail or how we would even accomplish that. um The good news was, you know and I've learned something important, that relates to her that timing is very important, right? You can have the right idea, you can have it too early, you can be too late.
00:09:04
Speaker
um Arguably, we were a little bit too early to symbionics, but the good news was we were, this was a side project, so we didn't move that quickly. So this was something that I did while we were finishing medical school and then starting residency. We we each had, you know, 80 to 100 hour day jobs. But by the time 2013 and 2014 was around, we had done something kind of interesting and unique. We were looking for how do we get out into the world?
00:09:30
Speaker
um and cut a long story short, we found a partner that, you know, by that time, many major companies were were interested in starting to get interested in wearables. And so we were able to to bring that technology to Apple. um And so in 2014, I left, you know, my job in Boston as a resident and I came to the Bay Area and, you know, was a full-time employee basically on the Apple Watch team. And we took some of the technology we had been working on and got a chance to help build that into, you you know, is now one of the largest, you know, consumer great wearable platform on the planet, which was which was itself really fun and exciting and began, you know, my career then and really understanding what it takes to go from a an idea into a ah product that could serve millions ah of people.
00:10:18
Speaker
Yeah, of course. And I think you see this now today. It's really interesting how healthcare is still personalized to that extent. And it's almost a little bit of like a mix of timing and luck and there's skill involved too.
00:10:31
Speaker
But having such a large role on how it sort of evolves and how it scales quickly is is really exciting. but and And now going... the Silicon Valley and Apple beyond that I mean of course you want to escape the cold weather but what did that look like why did it feel like working for these large skilled conglomerates and they had so many different problems they were solving but what were you trying to solve there i mean this intersection of math and biology again what did that look like and what it will it actually look like today with what your work was like there because I can imagine um an extremely different environment from working at symbiotics or being a researcher at Harvard too Yeah.
00:11:06
Speaker
I mean, i think it was, first of all, an amazing experience and I had a had a great time, um partly just as a fan of technology, you know, as a consumer of these things to get to be on the other side and see how,
00:11:22
Speaker
how they're made and to know some of the people that are working on it and to learn from those people was was a dream come true. in this And you know a lot of the time it was like, you know ah you're getting to work on the technology that on the other end you want to buy, but you get see it before. And and that gave me really a really deep appreciation for what it takes to to to build technology that meaningfully impacts people' people's lives. And so in a way, my experiences there really gave me that bug to, okay, you can do, you can, you know, and for for a lot of your listeners who might be interested in science and technology, there's sort of a choice. Do I stay in academia at the cutting edge?
00:11:59
Speaker
Do I work on tech? And the answer is both are great. But if your if your goal, really, it was very clear to me that if your goal is to get things out into the world, um that is a slightly different, it's a very different problem and it requires different types of teams and solutions. So, you know, the the thing you learn in a big company is how how do you get large groups of people, not just one brilliant brilliant scientist, one brilliant person to deliver something at Apple or Google. Nothing happens unless a lot of people have that idea and we figure out how to coordinate those people to
00:12:30
Speaker
take that idea and turn it into action.
AI in Medical Diagnostics at Google
00:12:32
Speaker
um And the hard part is often the science is the beginning of that. The hard work is really in figuring out the product, doing the hard engineering and system integration, and then making sure that that thing is going to assist survive.
00:12:47
Speaker
Quality you know is going to actually hold up in the in the rigors of the world real world. So you realize that as a scientist coming into these worlds, you realize that you know having the bre up the smart and good scientific ideas just the beginning and it's probably 5% of the way to actually having a scaled product.
00:13:05
Speaker
um I was lucky to get to do that, you know, as we worked at Apple to ship the Apple Watch and then a number of build on carry-on features. So we were interested, when the Apple Watch shipped, it was not a medical device. We had no interest in being a medical device. It was a fitness tracker.
00:13:18
Speaker
But very quickly, Apple realized that one of people's interests in where wearable technology really was health. And so Apple started to invest in um health applications. There are now multiple FDA clearances that sit on top of the Apple Watch that you know do things like detect if you're having an arrhythmia or could be used in in a a variety of different health applications. And so i I got to be part of that. I was at Apple. We got to be part of the team that was thinking about some of these early applications. It was really very early.
00:13:47
Speaker
um And Apple's come incredibly far since then. um When I went to Google in 2018, it was sort of writing then what was kind of funny, the hot area of AI was was computer vision, large. This was for large language models, but there was a similar effort in Google Research focused on how it could, you know, computer vision models be used to improve diagnostic specialties in medicine that involved images. So there are teams working on radiology and pathology and dermatology and they were looking for somebody to lead, um you know, the products for the the work around digital pathology, which was of course my background as ah as a scientist as a clinician, I'm a pathologist. And so it was a,
00:14:25
Speaker
Going to Google was an opportunity to combine my clinical expertise with with what I thought was the next really exciting wave of ah technology advances that were going to impact medicine. um You know, and at Google, we we had similar problems of how do we first show the science works, but then how are we going to get these technologies to have any type of impact in the world? and how do you bring together all the stakeholders, the device manufacturers, the the regulators, the payers, the clinicians,
00:14:53
Speaker
to get around new and transformative changes.
Brain-Computer Interfaces: Concept and Applications
00:14:57
Speaker
And that was itself you know an amazing experience. So, I mean, I learned a ton. um I was incredibly grateful for the time I spent at these large companies and I would not have been able to be in my current role at Precision had I not had the the sort of training ground um and support that you get in in much larger companies to get to learn from people who are experts in all of the different disciplines needed to take an idea from concept into the real world.
00:15:23
Speaker
Yeah, it's almost like those two entirely different worlds. I can imagine research, even in a place like math, when you're a super in niche field, when you're in your specialty, the people who you understand your research can be counted on a couple of hands.
00:15:36
Speaker
And now that you're at Google or Apple, you not only need to do deep research, but need to appeal to a much larger audience of people and different types of people with different incentives and stakeholders. So it's a more complex puzzle from what I can see.
00:15:49
Speaker
which some people just have the aptitude for. And I'd love to hear you've worked at these large companies and eventually gone to BCIs or otherwise known as brain computer interfaces. And it's a lot of people when they think of BCIs, they think of implants or sci-fi and things like that. And it's a little bit more than that.
00:16:07
Speaker
So I'd love to hear from you when you first heard of this technology, what were your first impressions about it? And what inspired you to go this route of entrepreneurship? If For sure. And I mean, I'll go back and, you know, the the first time I heard about brain-computer interface was actually when I met Ben Rappaport, who I mentioned was my co-founder, Ionix, and friend and Ben.
00:16:27
Speaker
Yeah. And Ben is a neurosurgeon and he did his PhD in electrical engineering. And i remember when I met him in 2004, asking him, what what are you going what are you going to be when you grow up? And he said, oh, very simple. i'm going to be a neurosurgeon and i'm going to work on brain-computer interfaces. And my reaction to that is, that's interesting. What's a brain-computer interface? So um the idea then was really very science fiction or very in the very early stages of science. And now here we are over 20 years later, and you know I'm working in a one of few companies that are trying to make brain-computer interfaces a reality.
00:17:02
Speaker
um Very simply for you know the listeners who aren't familiar, brain-computer interface is a medical device that connects human the human brain to to to to put the digital ecosystem, the computers.
00:17:14
Speaker
And you know the initial applications of this technology are basically allow those users to control computers using thought alone. That could be things like control a computer cursor, like click click buttons on the screen, type,
00:17:29
Speaker
um or to communicate, to speak, to you know go from thought instead of voice to text, go thought to text um in a way that can allow them to communicate um in ah in a way that's fundamentally new.
00:17:41
Speaker
And importantly, the medical applications and precision very much focused on the medical applications of technology are to restore that function for people who have lost lost it through a variety of different types of neurologic conditions. um conditions like spinal cord injury, stroke, or ALS.
00:17:59
Speaker
And to give you an intuition for why those conditions, these are all conditions where the brain, the thinking parts of the brain, the cortex are intact. So the user you know the the user can still have an intention to move their body or to move a cursor, but the somehow the connection between the brain and the muscles that actually affect that um action has been has been severed. So in a spinal cord injury, that's very clear. and There's like the brain, the brain is entirely intact, but the the the brain's ability to speak to the body has been shortcut. And one way we like to think of what brain all brain computer interfaces do is to act as sort of a bypass, a digital bypass
00:18:36
Speaker
that allows you to collect the um the the intentions at the point where they originate, but then communicate that out into the world to affect the user's intentions, desires, agent you know their a to restore their agents in a way that um overcomes their their injury or or disease. um You're thousand percent right. Doing that In all of these and all the places I've worked, one of the great things about working in products and innovation is that it really spans a host of disciplines. I can just say um you know working in the brain-computer interface industry, it's probably the most disciplines that we have to put under one roof because it' everything from fundamental material science and developing biocompatible interfaces with the brain all of the electronics that convert the brain's signals into digital bits that can be computed upon,
00:19:29
Speaker
We then turn that in that take those bits and turn them into you know intentions, and there's a huge amount of machine learning and artificial intelligence and software involved in that. um And that's just including the technical domains. Then we talk about how do we build tools and bring them into the operating room, into human brains? How are we going to support those devices once they're out of the field? How do we get those that technology regulated? how do we get it reimbursed? So we have
Interdisciplinary Development of BCIs
00:19:54
Speaker
um you know precision we're fortunate we're about you know we've got a lot of amazing people but they span everything from very basic material science and expertise to to people thinking about economics and reimbursement and so we we really need everybody and we also need to put all those people together into a cross-functional team that can work together and that's both extremely hard and extremely rewarding um to get to get to do that work and
00:20:21
Speaker
Yeah, and I think BCIs are generally categorized under neuroscience, but really, as you've mentioned, it's It's a lot more than that. It's not just that. It's signal acquisition, signal analysis, um the actual hardware and technology, and maybe some of these larger problems in BCIs are often tied to a lot of those ideas.
00:20:39
Speaker
So love to hear a little bit into what are some of the conditions or neurological conditions that you guys are alleviating today. Obviously, things like neurological disorders like depression are not there yet, but paralysis, ALS, spinal cord injury.
00:20:53
Speaker
I think a lot of people think of BCIs as chat GPT inside their brain or LLMs. And as you know, it's nothing like that at all. It's about the brains. The brains are full of electrical signals and you're trying to stimulate those when people's normal biology can't. So now you have these two clinical trials. Where do you think BCI will come into play in the near term or long term? And what does that look like in terms of how it'll shape the treatment landscape from your perspective, especially as new pharmacological treatments come up in the future?
00:21:22
Speaker
And I would focus this on two fronts. There are the initial indications that we say we have a high degree of confidence in. And by the way, the reason we have a high degree of confidence in these indications is because there's been 20 of of work leading up to the current state of the field that was done in academics and translational labs to to demonstrate that BCIs could do this function.
00:21:45
Speaker
um And then there's a bunch of things. And I think, you know if we zoom out, going back to the this what we discussed at the beginning, some of the most exciting and ah and impactful applications of genomics were not any of the things we imagined were going to be possible 20 years ago.
00:22:00
Speaker
At that time, we were interested in like finding the diagnosis for a specific condition or understanding gene genes that were involved in a particular condition. Nowadays, we can design custom vaccines for cancers or for new emerging infectious diseases. That was science fiction 20 years ago.
00:22:19
Speaker
right? So the thing to keep in mind is that what brain-computer interfaces are today are medical devices that are going to serve a very concrete set of initial indications. They're the ones I mentioned earlier. So spinal cord injury, neurodegenerative diseases like ALS, and you know stroke.
00:22:39
Speaker
What we're really excited by is the pipe the fact that every time we're we're in patients, and we'll get to this when we talk about precision, but many devices are now in the clinical phase of development. So these brain-computering phases for a long time were done in very, very small studies. We're now as an industry in a position where every day and year we're seeing the accelerating outtick of the number of patients who've had these devices put in their brain.
00:23:05
Speaker
And those those studies are both proving functions for the current indication, but they're also gonna teach us a lot about how the brain operates, how the brain is potentially affected in different conditions. And i don't I can't tell you with this high confidence what those next applications are, but I can tell you just based on my own experience, they' with very high to degree confidence,
00:23:28
Speaker
that there are gonna be new insights to come from this technology.
Joining Precision Neuroscience and Innovations in BCIs
00:23:31
Speaker
And hopefully, hopefully, and I believe this to be true, but can't say it with certainty that hopefully those insights will show ways in which we can use the underlying technology platform to address an even larger range of neurologic conditions. So um hope my hope is that in 20 years ago, you know, 20 years when we're talking again, we're gonna look back and say, wow, um you know, look at how the radio interfaces have transformed neuroscience, our understanding of the brain and the treatment of a whole host of conditions.
00:24:00
Speaker
And we'll say, yes, of course, it's obvious how the early indications of paralysis led to that. But um going forward, it's much harder to predict than anybody who tells you for certain they know where that what the next 20 years are going to be is is probably fooling themselves and a you Yeah, and it's like, who even knows what it might be? I mean, we might, maybe we find causal biomarkers for these diseases, and that's already hard enough. And there's so many things that we could do.
00:24:28
Speaker
But as you've noticed, as and i and I think it's part of the reason you're in BCI today, is that in the past, in the 2010s, since you got your PhD, there's been so much momentum here. And I think part of that, of course, is it's in the public eye with Neuralink, and there's a larger focus on computation and mental health. But we've had much better progress in sensor technology, compute, and so much more. but um And part of that's machine learning too. So I guess I'd love to, I think it'd be interesting to hear if with all these different things picking up and neuroscience still progressing as always, what was it like joining Precision Neuroscience and why did you end up going to that area?
00:25:02
Speaker
And what's really unique about that mission that maybe puts you guys in a good position to succeed? I know you have the seven-letter cortical interface, which is really interesting, but I guess it's better left in the hands of you. How would you love to explain what you guys are doing today and why that's so important to you?
00:25:20
Speaker
So, you know, Ben really believed there was a different way, one that could respect the the brain's, you know, tissue boundaries, but still capture high-revolution signal that could be used to drive, you know, really compelling function.
00:25:34
Speaker
And so he founded Precision to to start that. And, you know, what we've developed at Precision is a um novel you know neural interface. um We call it the layer seven cortical interface. That's an allusion to the fact that the cortex has sort of six cellular layers and our device kind of sitting on top of it acts like a synthetic seventh layer.
00:25:55
Speaker
um It is a very thin film micro-electrode array. So think of it as like for your users listening or It's like sort of um you know something about a fifth the thickness of a piece of scotch tape. it kind of um So it's extremely thin and flexible, but on that sort of thin film are um you know printed tiny each of which record the electrical activity of the brain in a tiny region um you know on the brain's surface.
00:26:28
Speaker
We think of that you know as basically like you know that array can kind of pick up the brain's electric activity kind of like ah pixel like a pixel can, like ah a camera can.
00:26:39
Speaker
um but wherere we're not just taking photos we're kind of creating um living video of the brain's electrical activity um what's you know cool as and we've demonstrated this in you know animal studies and now human studies is that that device can be placed on the brain where it doesn't because it doesn't damage the brain surface it can also be removed um without causing any damage um so you know our our interface is intrinsically you know safe and reversible um And also because you're not poking holes into the brain surface, it's also possible to cover larger areas of the brain than would normally be possible with traditional approaches. So it's also much you know more scalable, at least in terms of the amount of brain
00:27:17
Speaker
coverage We actually have hold the world record. We've put four of these underlying thousand channel units. So we've actually had over 4,000 electrode channel human brain at the same time. um Obviously the field is moving fast. And so, you know, these, these numbers are going to continue to change, but I think it just shows you the, that the underlying platform itself is is highly scalable.
00:27:37
Speaker
um We realized very early on in the company's founding that this advantage of being non damaging and reversible would give us a different path.
Navigating Real-World Constraints in BCI Development
00:27:45
Speaker
into clinical development than our competitors.
00:27:49
Speaker
And it really rests on the the safety and the reversibility of the electrode. So when you penetrate the brain, you're really, you're going to damage and then you're going also do damage when you remove it. So there's really no way to do short term brain computer interface implants. What that means is you have a very long road from the time you first develop your technology to when you can actually be in a position to test that in human patients.
00:28:10
Speaker
For context, Neuralink was founded in 2016. They did their first human implant in, I think, 2024. Other companies in the space, and that's like with a ton of resources and brilliant people, that it takes you know eight years. Other companies have taken 10, 12 years before they get their first device in the human brain. That's a very long time to go. before you get any feedback on how your technology is working and whether, you know um and so we've taken a very different approach to development at Precision. We went from you know company founding in 2021, 21, we did our first human clinical test in 2023, pretty much two years to the day of when we founded. We're now you know four and a half to five years in and we've done over 70 human implantations at you know of nearly a dozen US s academic medical centers.
00:28:58
Speaker
um We've been in more places of the brains. As we said, we put more electrodes and we've been able to do um a lot more work and a lot more you know to study this question of like how similar are is one brain to the next.
00:29:11
Speaker
and we've been able to do that faster and that were you know that directly results from the, like I said, the core architectural decision of trying to build a system that doesn't intrinsically um you know damage the brain and that um you know has has opened up this this this faster path. I mean, that's really important and been been an accelerant for us and why we've been able to move so quickly into this clinical stage of of practice, as opposed to other companies that have had to take a much longer time to to get to the same level of clinical validation.
00:29:43
Speaker
Yeah, it even almost ties back to your experience at Google and Apple, which, of course, the basic science matters, but there's so many more stakeholders that need this, that are in the way of this getting implemented. So um building BCIs isn't really about like a science project where you want to have the best data possible. Instead, you're building a medical intervention that faces real-world constraints on multiple levels, and you need to implement that at a large scale. So, I mean, again, i think it comes down to the trade-ups in BCI, as you mentioned before, um yeah if you don't have these like small needle-like electrodes that get every single neuron's data, and you're relying on larger surface-level cortical signals, which I think can take you very far, and i think it'd be interesting to hear what type of data you're actually getting from this. You can't get as deep as the single neurons in the basal ganglia, for example, but
00:30:30
Speaker
You have entire cortical layers, which is really exciting. So what is that tradeoff and sacrifices you're making opposed to an approach like Neuralink, which is more invasive and harder to implement clinically? um What does that look like in terms of what you guys could have done and what you guys are doing now and how that impacts the real world outcomes and interventions you guys are getting?
00:30:50
Speaker
I'll say this at the start, and we don't think, I think you said it perfectly, that these are trade-offs. um You know, you're going to make different systems for different applications are going to weigh different trade-offs. And so we don't think there's one interface that's going to,
00:31:05
Speaker
you know solve every problem in fundamental and translational neuroscience. We actually, and you know as much as we're competing with other companies, we're really glad that there's a health, that you know there are multiple approaches being developed because ultimately what we care about is seeing this technology out into the clinical world. And so having more technical abilities and capabilities is just going to mean more opportunities to help, um you know, people with with neurologic disease. So, um you know, what we, what I'll say is, you know, we we are measuring, you know, um activity on the cortex, which is the surface of the brain. So, you know, and while that is limiting, we're not going deeper into the deeper structures of the brain.
00:31:48
Speaker
as your listeners probably know, you know, lot of the, you our conscious understanding of the world and our intentions happens in the two millimeter sheet that surrounds the brain. So there's quite a bit that can be learned on the cortex. And one of the things we learned, so you know yes, we're on the surface, but um by making our our, we have a thousand channels and I didn't give you the dimensions, our electrode array is about a one and a half square centimeters. It's about the size of your thumbnail.
00:32:14
Speaker
So and we're getting a very high resolution picture of each of those patches of cortex. And I think what we've shown through our research and preclinical you know preclinical and clinical work is that there's a lot of information that can be learned about the brain's intentions at that signal, that you don't need to have single neuron level resolution to be able to decode things like intention to move, intention to speak.
00:32:35
Speaker
We're obviously working on this with our partners and we'll have quite a bit It's enough, it's easy for me to say it. We're actually working on scientific and public demonstrations, but we you know have been able to get, to develop the both the ability to put these devices in those book in those you know parts of the brain that control movement and speech. And we're now working on, you know we've been able to to decode that and and allow people to control computers you know and communicate using thought-based control.
00:33:02
Speaker
um And that I think alone is an important milestone for the field to know that You don't have to get single neuron resolution to be able to drive performance. That's an important technical milestone. Obviously, the most important thing is we have to build our technology into a product that can go into the human body for years at a time. Right now, we're doing these studies you know, in the operating room and in the days and weeks following neurosurgical procedures, but we're limited for regulatory reasons to up to 30 day studies. And our intent is, of course, if you're paralyzed, you don't want a device for 30 days and then have it removed. do You want this for months, years. And that is the, you know, that requires a whole host of other problems that are mostly engineering and, bio you know, ah
00:33:48
Speaker
bioengineering and clinical and regulatory science to demonstrate, to build that system, have it work and sustain, to survive in the human body for decades, and to then get that cleared for you. So we're working on that in parallel.
00:34:01
Speaker
And that's going to be the next major milestone for precision to do not just a temporary study in a human, but to be able to do this in ah in a chronic system. Other companies, including Neuralink, are in the stage of testing their chronic system.
00:34:14
Speaker
I think we just take a step back. You know, you mentioned this the field sounds like science fiction, and 20 years ago when I first heard of it, it pretty much was. We're now at the stage that you have multiple companies that are well-funded, all of people from all of the relevant skill sets are now advancing clinical
Future of BCIs: Machine Learning and Regulatory Challenges
00:34:31
Speaker
trials of these systems. like We are going to have devices cleared by the FDA for use in humans you know in this decade, for sure. like and um and That's extremely exciting.
00:34:42
Speaker
um and It's also a catalyst for you know this is no longer something that is just being driven by academic communities. There are actually like commercial and clinical and regular stakeholders that are advancing this technology forward. So the pace of progress, I think from here on out is only is only accelerating.
00:34:59
Speaker
um And that's exciting. And you know we're we're we're we're grateful and fortunate precision to be part of the field and trying to move that forward. But it's not just it's not just precision that that is is is making this possible.
00:35:13
Speaker
Yeah, of course. I think it's like it' almost as if someone asks, what is the best drug for Parkinson's or cancer? There is no best sh drug. depends on the patient, the symptom, your goal, and a million million other variables. So you can imagine, I think, for a variety of approaches in neuroscience or brain-computer interfaces, they're they're all possible, but...
00:35:31
Speaker
and necessary for the industry and you get a nice network effect with these different types of research going on but now that we're hitting the stage is more commercial payers and other parts of the industry the fda is getting more involved with this technology what are the main obstacles that you're looking forward to tackling in this area i think that of course you have the signal acquisition problem you're focusing on and made significant strides but then again everyone's brain works entirely differently so on the data analysis data analysis side i can imagine it's a pretty difficult with even with machine learning and LLMs. But what are the main obstacles you're looking to tackling for the next few years that and what needs to be tackled for BCI is to really reach reach adoption at large, especially at a company like precision neuroscience? I think you hit on in your question two of the things that I think are most exciting. For one, the the thing that we're working on now and
00:36:21
Speaker
is most exciting and why I think there's the biggest tailwinds is exactly what you said, is is in machine learning and the interpretation of the human brain data. We're collecting, you know, every device, every brain-computer interface device collects a lot more data than we are actually able and positioned to collect.
00:36:37
Speaker
extract and use. you know We collect you know data at you know billions of data points a minute, you actually, sorry, a second, you know and yet you know we decode output streams that can be measured in like bits per second. So the compression rate of the decoding is extreme, and as machine learning techniques, and frankly, just as we have more data, is going to allow us to do just matt amazing amazing things with this data stream. And that's probably the most technically and and clinically exciting aspect. And I think that we're really at an inflection point and it's driven by the availability of these devices and our permission and ability to get them safely into the human brain at the same time as the AI tools that are out there are just getting exponentially more powerful. So that's something that is extremely exciting and we're living through and anybody interested in this space
00:37:29
Speaker
um should be very, you know and wants to contribute, should just be very excited because you know the the whole field has done somewhere on the order of like 100 implants up until this point. And you know in the year the next, well, probably as an industry in this year, gonna do more than double that across all of the various approaches. And you know it's only gonna accelerate from there. So like I said, in the genomics,
00:37:50
Speaker
I think we're going to see in the next few years just an exponential increase in the amount of availability of data and what we're going to be able to do with ai is just very, very exciting. That is you know technical um and there's a lot of I'm very confident that that's going to be something that the field itself just makes more rapid progress on.
00:38:09
Speaker
And you know that'll be a flywheel because obviously the better the devices work, the more people will want to have these procedures and the better the AI is going to get. So that's exciting. You mentioned the other one, which is very ah a very big problem, which is you know having the technology is only part of it.
00:38:23
Speaker
In order for that to get out into the world, you need to train doctors and clinicians and systems to use and adopt. and you need the regulators to approve it, and you need ah reimbursement model that ah the pays for it, right? um And I think that of those three problems, you know, FDA has been amazingly engaged, and actually, you know, but a lot of the, you know, BCI researchers have gone to FDA, so like the the FDA is very knowledgeable and aware of
00:38:54
Speaker
this industry and while they they have an important job to do to make sure it's safe and effective, I think they're very clued into the technology and and you know we see them as partners and trying to make sure the the technology is is available.
00:39:07
Speaker
The one that I think needs to be addressed still, and there's a lot of active work streams and you know I think this is this is something I'm confident we'll be in, but it's not done yet, is is to engage payers and figure out a pathway for reimbursement.
00:39:21
Speaker
It is extremely important. Obviously precision is a commercial entity without reimbursement. It's hard to be viable. I would just frame it as for the, the users, you know, the idea that they're going to have this technology out there that doesn't help them if, unless there's a way for that to be covered by their insurance.
00:39:38
Speaker
Um, and you know, it's, it, this can't be a technology that's only available to a few people who can afford to pay out pocket for it. It needs to be equitable and accessible. I'm optimistic. that you know when people see what these devices are gonna do and the potential impact that we as society are going to you know make the choice that this technology needs to be covered in some form. But until that's done and that's clear, that remains a major risk that I think the industry will determine whether there's a viable, not like viable precision, whether it's a viable BCI industry or not.
00:40:13
Speaker
but and And I think that's something you know people who really care about the the last mile should be you know paying paying attention to.