Next Unicorns: Making surgeons bionic via computer vision & AI with Proprio’s Gabriel Jones | E1790

0:00The data is actually the most important piece here because you get very detailed information and measurements about the performance of the surgery from the plan and the diagnosis, from the imaging before the surgery was ever done, through the entire surgery and through the outcome. So this is the first time in history that anybody, any company has been able to invent and bring to market a technology that could actually connect all that longitudinal proprietary data that says, what was the intent? We know the plan, so we know your surgical intent. What were you trying accomplish here. This Week in Startups is brought to you by Superside.

0:38Design and creative are crucial for growth. Tech companies like Shopify, Amazon, and Meta have found the perfect solution. Superside. Get$2 ,000 off with Superside's Startup Accelerator Package at superside.com slash twist. Vanta. Compliance and security shouldn't be a deal breaker for startups to win new business. Vanta makes it easy for companies to get a SOC 2 report fast. Twist listeners can get$1 ,000 off for a limited time at vanta.com slash twist and LinkedIn marketing. To redeem a free$100 LinkedIn ad credit and launch your first campaign, go to linkedin.com slash next unicorn. All right, everybody, you've heard all about computer vision, cameras, AI, all of this stuff has been coming together to enable founders to create new innovative projects.

1:39One of the projects that we've been watching, robotics, surgery, is really compelling. And today, we're going to see how real-time computer vision is working in the OR. the operating room so with me today is gabriel jones he's from proprio am i pronouncing correctly proprio yep proprio p-r-o-p-r-i-o and you can go see their website propriovision.com tell me gabriel uh what is uh proprio and what are you building thanks jason yeah hoping to blow your mind here with some real-time product demos so we are just outside of the proprio headquarters operating room facility, Mach-O-R, if you want to think about it that way.

2:27I think it's on the door behind me, the precision lab, as we refer to it, where we do real-time surgical simulations. And when I say simulation, it means there are cadavers and samples like that being used in computer vision surgery live behind me right now as we sit here. So I wanted to give you a better version of the product demos that everybody else who comes on your show shows you. I'm not just going to share a Zoom screen. There's a body in the lab being operated on by our team right now. And this is not just a demo for the sake of a demo. We're developing applications, including already FDA cleared applications, for performing surgeries like spine surgery, in a totally different way compared to how it was done in the past.

3:09And happy to dive into more detail there. But the simple analogy that I think is really easy to communicate with your vast audience is up until very recently, all of surgery was operating in a map quest kind of a context. What does that mean? We expected surgeons to basically take an x-ray or a CT or an MRI, that type of imaging, effectively either print it out or have it on a screen, store this wealth of knowledge in their brain from 20, 30 plus years of experience, try to pull all that together computationally within the one human brain and then apply it to take care of the patient. Oftentimes, they're quite literally printing out an x-ray and taking it into the OR.

3:51You and I are close to the same age. So we remember when you went from Rand McNally to MapQuest. That was amazing, right? I have a custom map for the entire world. You know, I just had to print it out, type in, you know, the origin point and the destination, take it in the car with me. And it was amazing, right? Turn by turn. Except for if anything changed. If anything changed at all, whether it was a construction project that wasn't on MapQuest or an accident that occurred, or I wanted a sandwich. There was no real-time aspect to that information or that data or how it was presented to us. Now, fast forward, your good buddy Elon and others have taught us how to navigate in real time with GPS and Starlink and all that good stuff.

4:36So now we have real-time updating for driving across the city or the country, and thankfully plenty of charging stations. Well, we forgot to do that for surgeons, which is frankly unforgivable given that they take care of you and me and our children and our grandparents. And so now's the time, we'll probably talk about why now, all these technologies coming together, as you alluded to, can actually provide real-time GPS, real-time 3D volumetric intelligence for a surgeon to perform at their very best every single time, really in pursuit of perfection of medicine i think that's the exciting moment we're in right now and for people who are not watching just go to youtube and type in this week in startups and you'll go find this episode uh in our in under the videos tab but behind you literally uh right behind gabriel is a surgeon working on a cadaver yeah so um first of all i want to thank all organ donors out there.

5:38They make things like this possible. I'm an organ donor. It's on my driver's license. So with a lot of sort of humbleness and respect, we and other companies developing both drugs and medical devices and software for these kinds of applications are able to acquire. Don't ask me how I get them, Jason, but bodies show up here. You don't want to know the details. And fortunately, we're able to develop applications that can perform in cadavers. And then there's a whole FDA clearance process to get that into the operating room to work with live humans and solve problems for them. But what you're seeing is that testing and development occurring behind us.

6:16And for those listening on the podcast and not seeing, you have members of the Proprio development team in a mock operating room right behind me using the Paradigm system, which is the name of our core platform made by Proprio, to perform surgeries. And we're going to talk more about that. But here we have a spine surgery, very common. Think about like a lumbar fusion. You've probably had friends who have had a procedure like that. Yeah. Or on the very other end of the spectrum, something like a scoliosis correction procedure. There are 7 million Americans who have that affliction. And these are in the same sort of family of orthopedic surgeries that happen millions of times a year.

6:57And again, they're done with essentially that x-ray that we were talking about, that MapQuest. type of information just to be clear how this works you uh will take a 3d model of a person's body say their spine that would come from an mri i take it or an x-ray or some combination of those and build a model of my specific body as opposed to somebody else's because everybody's body would be unique and whatever surgery is going on is unique and then you will create a way for the surgeon to see this now does the surgeon wear goggles do they wear an apple headset how do they actually see this manifestation of say the spine and maybe we could do a little product demo here that we could sportscast people through yeah let's do it and i love this sportscasting analogy because it connects to a core principle which is we've instrumented the athlete and the astronaut right there's people lebron james has a whole team of people following him around uh i bet he can't take you know he can't go to the bathroom without somebody sampling that and saying you know how he's doing digestively yeah and it makes sense right he's probably spending millions of dollars a year as our sponsors to perform at the absolute peak that he can for as long as possible there is a literal 3d computer modeling um system for watching people shoot three-pointers and the reason the nba has gotten so much better collectively even the worst three-point shooters are better than the average 10 15 years ago because of that 3d modeling system that is in the gyms that shows people their arc their stance their footing you know the release etc so computer vision to help people hit three pointers and here we have computer vision uh to help people do perfect surgeries or more perfect surgeries that that's well said um if you didn't have a day job i'd be recruiting you jason so what are we seeing here i see like a some sort of an overhead uh is this a projector is this a computer screen what is this overhead device here that yeah you're seeing the paradigm robotic computer vision system which has an array completely proprietary array of sensors think of this as like the lidar type systems that are on teslas and such yep and basically we've designed and engineered a system that can dynamically reposition throughout the operating room a set of sensors think of them like the satellites flying above right and we're mapping the mountains of the body dynamically right and we can use other types of imaging you alluded to, MRI and CT.

9:27Basically, our core capability is this real-time 3D volumetric imaging. You're familiar with Lytro from back in the days and the ability to basically capture an entire scene dynamically live in 3D. Then you can do magical things with that once you have that core kind of VR video, if you want to think about it that way. But it's live, and that's our core innovation. We invented the system to capture that and to render it live and to give a surgeon or a user the ability to interact with it live so now you can there's this overhead sensor array looks like the size it looks like a pizza box basically um and that has a bunch of sensors it's flying over my body i'm laying across an operating room table and it is in real time looking at if you've opened me up you know my spine whatever it is and it's going to make a 3d model in real time of what uh the surgeon is seeing so it's an extra set of eyes for the surgeon essentially super high precision x-ray vision without the x-rays think about it that way all right if you're listening to this podcast you care about innovation and one sector that really needed a shake up was design i'm always talking about world-class design and how that is the difference between getting funding getting customers and maybe not getting funding and not having customers in the past what were your choices you could work with an old school agency expensive, slow, right?

10:47Or you can go to freelance marketplace is great. But let's just say it'd be variable quality. It's just a lot of work. Yeah. And so you have these two choices, you can go low, you can go high. But what if there was a better way? Well, let me tell you about super side, it is the new way to get designs done quickly. They call it has C A S creative as a service. It's a fully managed end to end service. And it's completely hassle free. When you subscribe to superside, you'll get an amazing dedicated design team built specifically for you. And you'll get access to a platform that makes it so easy to request designs and have them delivered quickly.

11:25Amazon, Meta, Salesforce, Shopify, all use superside for a reason. And so do a bunch of fast growing startups and superside. Just so you know, they only hire the top 1 % of designers from around the world. And they have a full range of capabilities, whether you're doing creative landing pages, motion design, or custom illustrations, Superside has an amazing offer for twist listeners, you're going to save$2 ,000 a month with Superside startup accelerator package, go to superside.com slash twist. That's superside.com slash twist for$2 ,000 off. So generational shift in medicine, let's just go back in history.

12:02So what we're looking at here is the last 25 to 30 years of progress in what's called surgical navigation, right? That's essentially what we're describing. GPS has been this map quest. So we're seeing x rays that were used to essentially pin an x ray to the body and then rely on the surgeon to take that x ray, do all that 3d calculation, and then perform a surgery, somewhat magically, you know, doing a lot of that in their brain. And what you're seeing is, is a spine surgery that's being navigated in that way, really simple, elegant solution 30 years ago, but we haven't really made a lot of progress since then.

12:39And so this is a product from a company like Medtronic or Striker or Johnson Johnson. And that was the whole universe of innovation. So what we're seeing here is actually that real-time light field rendering that Proprio does. Here you see real data. Actually, this is light field data from multiple cameras and sensors capturing depth information, RGB information, like from your cell phone, infrared, non-visible spectrum, it's all baked into the same algorithms that then present that information to the surgeon. Here you can see the system operating at the edge of these robots moving around in the operating room.

13:18It feels like something out of Star Trek because it's on the path to that. And here what we're showing is the ability to look into a very small incision. This would be a minimally invasive surgery. And what Jason's seeing is essentially a CAD rendering 3D model that's the blue of inside the body. And all we are seeing, all Proprio is seeing, is about an area the size of your thumbnail. I assume you have pretty normal and healthy thumbnails. So, about five millimeters square. And once we've got that, we've locked onto it. And we have telemetry for the anatomy to perform the surgery. That's wild.

13:57So, you have this 3D model that's been built already of that spine, that piece of the spine there then when they cut the person open they're laying on their stomach um this array of sensors is reading it in real time and then attaching that view to the 3d model view that's already been taken ahead of time and then by putting those two things together in real time the surgeon then has a more accurate view of the spine in this case to make whatever um whatever incisions or manipulations are necessary to help that person yeah accurate and live that's the big distinction so what's hard to do about this all this computational work to do it live back to the why now question right our good friends jensen and their team over at nvidia have helped to accelerate a lot of this work we're going to get into ai and and really practical ai applications i think in this conversation too but the nvidia gpus you're saying have made it easier to create these real-time models you have that massive gpu power forget about ai this is just making the renderings at high fidelity in real time being driven by gpus which were put into computers to play video games so video games creating better surgery is if you wanted to find the lineage of this if people weren't demanding to play call of duty at higher and higher resolutions and frame rates this technology would have not advanced to here this would be a very expensive rig to play call of duty but you could do it um okay so let's keep going here if i can show you this so what we're showing is the ability to you had mentioned ar vr yes absolutely once you've done the hard computational work up front and what we're seeing is is the spinal anatomy aligned to the real physical world with the digital and we have a surgeon with a headset on which is a headset developed by proprio to look inside the anatomy as he's placing what's called a pedicle screw so this is an implant used in a spinal fusion hundreds of thousands of times per year about 1.6 million of these are done per year in the u.s very very common but the problem is if he's not using proprio this surgeon can't see inside the anatomy and sort of predict the potential impact of what he's about to do and that's the true magical revealing kind of the seeing around corners component of what proprio does this surgeon is going to be able to avoid putting that screw in your spinal column got it which is a big deal so when they go to put this screw in he's going to be watching it in ar or vr is this going to be augmented and he's looking at your actual body and it's helping him align the screw or is it all done in vr it can be done always you know the a surgeon who's maybe 20 or 25 years into their career might not want to wear a headset no matter how good the apple headset you know becomes over time for example yeah our bet at proprio is that the data is actually the most important component and how you choose to consume it is up to you got it so some folks might want some surgeons old school surgeons they've got a feel for this they're going to look at the monitor and they're going to see it maybe some new surgeons coming up they might like to put on a headset eventually and see the model in real time and then see through it with ar so they're kind of it's a perfect augmented reality application when you think about it um obviously i'm biased but yes i agree with you got it so how close is all this to you know working in the field you literally have a cadaver uh thank you to the people who are donating uh we don't want to make light of that in any way it's pretty serious business here and thank god people do donate um you know their their human vessels after they're gone to help technology like this because this is going to greatly reduce errors um and i guess less lower costs eventually of different surgeries or make more surgeons in the world is it what do you think the ultimate impact of this is does it make people faster and better or does it make you know being a surgeon uh you know a career that maybe more people can uh more accessible to people ideally all of that in my background i worked with the gates foundation and bill gates and we were trying to solve very big problems over time you know 25 years out in healthcare globally.

18:30And so we kind of have that motivation as an organization to extend access to care. Today, we're constrained by the ability to make new surgeons that are excellent and skilled. And J.K.L., I don't know if you're a Scotch fan, but I hear you and Chamath like to drink some good wine. But similarly to Scotch or whiskey, it takes 25 years to make a good bottle. yeah but you can't if you and i decided to double our consumption tomorrow um the production process to make more great scotch whiskey is going to take 25 years to get that product to us it's a good analogy for thinking about my co-founder dr broud who's a world-renowned pediatric brain surgeon who's on the screen right now that we're seeing he's in his early 50s and it took probably 25 years of training from pre-med until now where he is world-class ah so time to world-class could drop precipitously dramatically yeah because training um is going to be more available and so i assume with this system you're going to be able to replay surgeries and i'm assuming that's how surgeons i think learn is they replay surgeries they ideally but you'd be surprised how little data is actually available for them to actually kind of beam back in to what dr browd was seeing in that surgery and see it right out of his eyes as many times as they want so that 10 000 hours is very difficult to acquire today so right now if i wanted to become a brain surgeon a literal brain surgeon yep it's not like there's some corpus of here are a bunch of surgeries where you can see deeply into the brain stem whatever spine wherever this work is occurring and so the only way to do it is to to stand to the right of somebody or be watching from one of those or you know rooms where they have the i guess uh stadium seating as it seems like it seems like a crazy way to try to learn how to do surgery yeah you would think that we just have videos so when if you're an auto mechanic or you my mom repaired her own dishwasher she was like literally in brooklyn i was talking to my mom she repaired a dishwasher by watching youtube videos made for the melee person it's like yeah just watch the video and i bought the part online and i did it it's kind of the equivalent here You're going to have training videos that are so good that you would have people time to amazing surgeon could be cut in half or cut by 80%.

20:54Yeah. And this is the why this technology is particularly powerful from a training perspective is light field gives us the ability to essentially create as many arbitrary camera positions as you want in the volume in the space. So that means I can see right out of Dr. Brown's eyes. i can see what another doctor is seeing live in that procedure and i can see it from anywhere in the world so from a training perspective yeah so let me just double click on that for a second if i was a surgeon if there was a rare surgery occurring brain surgery is rare i think um you could have and this doctor who's a legendary doctor was going to do this brain surgery one day you could have a hundred brain surgeons from around the world getting to see their perspective and how they handled this and everybody levels up from this very rare case and this very rare instance of getting to watch the pilot as it were navigate landing a plane in an emergency type situation that is wild to think about yep if you're a sass or services company that stores customer data in the cloud, then you need to be SOC 2 compliant.

22:05You knew that from a third party and you need that third party to close big deals. And if you want to get compliant easier and faster, you need to use Vanta. V-A-N-T-A. Vanta makes it so easy for you to get and renew your SOC 2. On average, Vanta customers are SOC 2 compliant in just two to four weeks. Compare that to three to five months without Vanta. And Vanta can save you hundreds of hours of manual work and up to 85 % of compliance costs. This is a total no brainer. And Vanta does more than just SOC 2 compliance. They also automate up to 90 % compliance for GDPR, HIPAA and more. You can't afford to lose out on major customers.

22:42We all know that. Listen, it's a hard year. Last year was hard. You can't lose those major customers because you don't have your compliance dialed in. Just work with Vanta. Get your compliance automated and tight and tight is right lock down those big deals. Here's the best part Vant is going to give you$1 ,000 off that's 10 hundies get$1 ,000 off at vanta.com slash twist that's vanta.com slash twist for$1 ,000 off your sock too. And let me go even further Jason. So we capture about 250 gig of data per hour of surgery per system running out there. So that's on a compressed basis. So the real number is quite bit bigger than that again thanks to our friends from nvidia and intel for helping crunch through all those data but the data is actually the most important piece here because you get very detailed information and measurements about the performance of the surgery from the plan and the diagnosis from the imaging before the surgery was ever done through the entire surgery and through the outcome so this is the first time in history that any body any company has been able to invent and bring to market a technology that could actually connect all that longitudinal proprietary data that says what was the intent you know we know the plan so we know your surgical intent what were you trying to accomplish here and i'll show you a very specific example right now actually so here's what we're looking at is scoliosis in an adolescent girl oh this is in the public sphere so there's no uh phi that we need to yep and this is what we're trying to accomplish in the surgery so folks at home who don't have the video on you're seeing a what's called a deformity uh in spine so this is a young girl whose spine is is curved quite dramatically uh it's impacting the rib cage and the shoulder blade and and just leading to a lot of discomfort um we don't know why the industry and academia does not know why the medical community does not know why the scoliosis of this variety impacts young girls at 10 times the frequency of young boys wow and and literally this poor child's uh or adolescent's spine is an s when it should be straight i guess when looking at it from this perspective from the back perspective obviously there's a is a there's a healthy curve to a spine but that's not this this is looking at it from the back it looks like it's an s as opposed to a straight spine but you've touched on something very important the s is a healthy shape for a spine from the side yes but it can it can be contorted and twisted in 3d which it makes for very difficult corrective surgery and so what we're looking at here is actually a scoliosis surgery in parts a couple snapshots that we've taken and we're showing a very open surgery with a curved spine that is unhealthily curved and the way the surgery is done today is literally printed out images with target angles on them that are either brought up on a screen or on a literal piece of paper so there's your map quest analogy and the surgeon is going to do their best to try to figure out and orient themselves in 3d relative to a bunch of pieces of moving anatomy and it's just an incredible math problem that's very hard for even the most brilliant surgeons to solve effectively so just so i can recap here for people watching you have the spine you've got 12 or so vertebrae it looks like if i'm using the proper terminology here that need to be realigned and they're instead of being curved looking like you know a tree that's leaning into the sun we got to get this tree to go straight to get it to go straight they print out each of these uh vertebrae as you said like printing out a map printing out your map quest as opposed to using gps in your dashboard um and they have to make their best estimate guess and how to align that spine that's an imperfect you know despite being noble and and you know people doing it accurately let's face it they're not winging it but there's a little bit of a freestyle going on here where they have to make i guess their best estimate on how to how to realign the spine correct that's right and their only other real tool is intra operative imaging so hitting the patient and the entire staff with a whole bunch of radiation right so which is taking like x-rays during this live yes and so as opposed to the better way which is what is removing the need for intraoperative radiation entirely so real-time light field-based navigation requires no radiation whatsoever not only that but i'll show you a little demo here so what i'm showing jason is a demo of a spine on a table here it's called a benchtop simulation and in this case we've placed a lot of screws into the spine which is representative of the surgery then they're using a titanium cobalt rod to force the spine into a different shape and what we're showing here is just live geometric calculation of each independent part of the anatomy moving and we can actually just calculate that geometry live and update it at a moment's notice for the surgeon and here we're backing out of that we're undoing what we just did in the procedure to show that it's all captured live and what's magical about this jason is this is the biggest predictor of the outcome of the surgery both clinically and economically this number predicts the outcome of the surgery amazing so now you've got this perfect precision gps as we're using the analogy here so the chances of making a mistake the chances of a successful surgery proper outcome is going to be dramatically increased and the number of surgeons who will eventually be able to do this surgery would not be limited we have a surgeon shortage correct that's right about 25 000 in the u.s uh physician shortage and it's growing as as you and i age and we need more surgeries the average american is going to need between seven and nine surgeries in their lifetime the longer we live the more we're going to have but we're not making surgeons fast enough and so this is really targeted at making every surgeon as effective as they can possibly be and just to put a little bit of a finer point on this the imax of the imax print of dune 371 gigabits that's a two and a half hour film about 150 gigs an hour you're doing 250 gigabits gigabytes an hour on a compressed basis yeah on a compressed basis so i mean this is more than an imax film that you're recording which you gotta think when you have all this data and then you start matching it to outcomes you the learning is going to go up dramatically as to um maybe what we could do better in the next surgery absolutely jason and let me just challenge you to go even further than that okay because my mind is trying to conceive of what happens when you actually have the surgery recorded at that fidelity level what what can you do with that So we get everything from the surgery.

29:38So go back to Elon's master plan, part one and two, right? Let's make a car, let's make it really fast, and then let's work our way down the stack so everybody has one. Okay, now we need a satellite and a charging station. We've got to put all those pieces together. So effectively, that's what we're doing here is we have invented the sort of tip of the spear surgical navigation platform, and it must be great. It must be the best. And that earns us the right to be in the operating room capturing all these data. And of course, we can drive deep learning models and we can train AIs with the best surgeons to perform better and better and better over time.

30:11That's very exciting. And that would be enough of a cause to build a company. It's very profitable enterprise, etc. But I don't think that's enough for proprio. And that's not why I'm here. I'm here to drive the company towards a future that is very data driven and where the entire industry has to react. And what I mean by that is the whole medical device industry is very focused on selling expensive capital equipment. So think about a robot that performs a surgery or an MRI machine that costs$2 million. Or they're very focused on selling you a screw that goes in somebody's back. And that screw costs$1 ,000,$1 ,200 a piece, and they make it for$2.

30:51It's the most lucrative razor razor blades model maybe in the world, right? To break that model and reduce the cost of care while increasing the quality of care, we have to invent things like Proprio, the paradigm, take it into the operating room and perform and improve all these surgeries. But where you can break the business model is say, well, now I know what it takes to perform the surgery really well. I know who does well and what circumstances. I know what products, both hardware and software, can be used in that surgery to improve it. So now you have more than one customer. You're not just going to a hospital whose finances might be challenging and saying, you got to write me a$2 million check.

31:33And maybe the most innovative thing is a capital as a service model. Now you're saying for everybody else who benefits from that surgery being performed, it's something approaching perfection. And I can see the gears turning. You're going, well, what about when do the insurers pay these guys? Right? when does the revenue cycle management company that goes and collects the bills for the hospital start wanting access to this data yeah when when does the medtronic the the implant maker go well these guys can make my implant more effective and tell me how effective it was yeah i mean that's amazing when you think about it and then how many of these surgeries do you need in order to feed it into an ai and say how do we do this better and it's probably somewhere between 50 and 100 surgeries i'm guessing the ai can start to have enough data to meaningfully give feedback there's nothing like that going on in the world right now nobody's feeding an ai model these surgeries and saying how can we make the surgery go better or does that exist some great great work going on with like taking endoscope data which is really you know video camera data of low fidelity and trying to teach models how to recognize a polyp and differentiate appreciate cancerous tissue and non-cancerous tissue.

32:48They did some wonderful work in those areas. And Intuitive Surgical has a lot of data from their systems, of course. No one has gone in and mapped everything that's happening in the most important areas of the body and the anatomy and the surgery end to end and captured everything. And I think that's where it's probably the most exciting area is, yeah, we can build models that can confirm things that we already suspected. but i think that the messy kind of belly of the beast if you want to go joseph joseph campbell with this is what don't we know that we don't know yep and what models can we build that will drive down the cost of care right we're creating 10 20 100 opportunities to monetize the data from a surgery i think that's where it's most exciting and that's that's what we're pretty fired up about here i guess then the question becomes robotic surgery and you know at some point um replacing humans in this process and that is i guess been scary or the holy grail depending on how you look at it eventually you know remote surgeries are correct me if i'm wrong already occurring so a surgeon can do a remote surgery across the country is with robotics is that currently the state of art and is that actually occurring or do they just get on planes and do it live yeah it so the first answer is yes it's possible latency issues have been the the sort of delay in that being adopted widespread and from a technical standpoint but you and i are human beings and and we you know we would prefer our physician to be on site he was also a human being in the room with us even if everything goes perfectly it's just it's much more comforting to have that hopefully world-class surgeon on site with you in the same room and let's say if it was a pediatric surgery right if it was a child involved and i'm the parent yeah of course not sure i'm comfortable having the entire thing done today unless you're a parent in a frontier market previously known as the third world but you know some market where they're the surgery is not even available or there's not a doctor they're capable of doing it you might say well this remote doctor and a local crew uh to be there in case of a problem maybe that's a trade-off that you would want to make if you were in a frontier market but i guess you could also that's the path that we're headed down for sure eventually it's you pop on a headset you've got hundreds of thousands of surgeries and the intelligence from that all beamable in maybe via starlink or something and and that intelligence is with you and on board everywhere you go uh today is you know there's too many pieces that need to work to make that happen.

35:24But if we wanted that future to be reality, we would go collect all the data from millions of surgeries to train the AI to be able to provide that guidance anytime, anywhere. And that's the path we're on. When you're selling to B2B buyers, you really want to get your pitch in front of the decision maker, the person who gets to sign the check, because these upper level execs, they're the ones who make the purchasing decision. Everybody can have an opinion on the team. Of course, it's 2023. But there's always somebody where the buck stops, and that buck stops on their desk and doesn't get into your bank account.

35:59These high level folks are hard to find. They're hard to target on social media platforms. But LinkedIn is the social network for business and they have 930 million members ready to do business with you. And that includes 180 million senior level decision makers. Plus don't tell anybody, There's also 10 million C-level executives there. That's a ton. Purchasing power. LinkedIn ads is built specifically for B2B marketers. No other platform in the world can offer these eyeballs and you can target them, obviously, by their location, the size of their organization, their vertical and their title. When you think about business, I want you to just think about LinkedIn.

36:34LinkedIn equals business. Business equals LinkedIn. It's that simple, folks. When you present them with an opportunity, they will, of course, be in the mindset to receive that because they're not posting pictures of their food for mentally on vacation make b2b marketing everything it can be and get a hundred dollar credit towards your next campaign by going to linkedin.com slash next unicorn to claim your credit that's linkedin.com slash next unicorn terms and conditions apply because linkedin is so generous so that this week in startups audience are you focused on one surgery one type of surgery now in terms of you know perfecting this and is that spinal surgeries is that where you can do the most good yeah so spine is very interesting it's a great starting point it's a very large business it's 30 billion dollars a year in the u.s and reimbursement alone so not counting the capital purchases or implants or any of that side of the equation it's just reimbursements from medicare medicaid insurers is 30 billion dollars a year and growing as the population ages as we were discussing so it's a good landing point the spine is also very complex and there are a lot of unsolved problems and so we chose that for various reasons also my co-founder is a brain and spine surgeon as we discussed earlier of summer now so we we knew those problems really really well and the market rewards you for solving those big problems so it's a good place to start you follow the spine up of course it goes to the cranium so brain surgeries cranial surgeries make sense similar set of problems uh believe it or not jason when you open the cranium and do a craniotomy the brain is actually under pressure and in a a bag called the the dura and so when you open that up the brain actually moves around quite a bit yeah so this real-time imaging real-time navigation knowing where the anatomy is and what it is in 3d space live is actually unsolved in the cranial space too and you start going through you know the stuff we do as entrepreneurs right you do that segmentation and you go okay well does knee surgery have the same problem it turns out in knee surgery they have to put a pin kind of like a qr code little marker in the lower leg and the upper leg above the knee and lock the knee into place why because they can't track it in 3d life so does that sound familiar to what we were just discussing in spine yeah well knee is 23 billion dollars a year in the u.s alone wow and the amount of time it will take to do these surgeries should go down dramatically what's the average time for spinal surgery and then where do you think this reduces it in the coming years in terms of time under the knife yeah the faster ones are you know an hour and a half to two hours that would be like a two level spinal fusion that happens hundreds of thousands of times per year in the u.s and then the other end would be a six hour scoliosis reconstruction like we were just going through um so six hours is is not uncommon now if you added a tumor to that that's even more complexity uh and in fact um i happen to have one for you right here uh look at that so this is a 3d reconstruction a printed anatomy and what you see what we're looking at here is is about 10 levels of a of a human spine reconstructed from a ct and mr scan fused together and then printed at super high fidelity so this is probably do people do that before a surgery print out somebody's 3d print out somebody's spine like they do and isn't it crazy that this would actually be your pre-operative plan wow you would print this out and very skilled surgeons we know will take in a literally i'm not kidding jason a ziploc bag because it's got to remain sterile into the operating room and hold it above the body so they can visualize the anatomy subsurface while they're working and making a lot of changes to the anatomy and they're looking back and forth between this and they'll have somebody holding it for them in the or oh wow isn't that crazy and what we're talking about here what you're holding up for the people listening is essentially a 3d printed model of somebody's spine with a big sponge of uh pink material which i assume is the tumor that's attached to it that's got to be cut out without causing damage to the person and what you can't see um but i can see here because i have the benefit of it right in front of me is that it's actually inside the spinal column too got it so you got to get in there between each of those um yeah and take out the wow that's wild so back to the original question is it we know that the fidelity is going to be increased so uh the chances of success go up but is it does it meaningfully change the amount of time because it does seem like the longer you're under in a surgery am i correct that the increased danger goes up yeah that's exactly right so there's a number of different dangers um we we anticipate shortening surgeries from 30 to 80 minutes on average depending upon the complexity with our system and you can just imagine let me give you a simple comparison if you're going to roll in a ct scanner which is like a giant donut in fact there's one in the room behind me you can kind of see over my shoulder there that's a million dollar scanner and if you roll it into the operating room it's about 20 to 25 minutes to use it once and that'll give a surgeon a sense of where they are in progress in the surgery but that's 25 minutes of delay at 200 an hour minimum so quickly you're spending five six eight thousand dollars just to get a snapshot of where the anatomy is and that is radiation and cost exactly and time And so all of those things together, you could save 25 % and then less radiation and less cost.

42:17Exactly. And you'd be able to achieve something like this much faster and safer. So this is what I'm showing Jason here is the after of the before tumors that I showed him before. And here they've removed several levels of the vertebra. Actually, four vertebra are gone. And what they've brought in is a substrate to strengthen and support the entire spine. And you see a healthy curvature here, right? that's called lordosis and kyphosis think of it as convexity and concavity and the level of structural support that has been brought in is just mind-blowing it's amazing but when you actually have that tumor and they have to take out the vertebrae they basically build an architecture around your spine to make up for the fact that the vertebrae have been removed that they're not being replaced with something yep and wow this person needs to stand upright right and gravity is unforgiving so physics doesn't quit yeah you have to be able to go walk out into the world and they will if this surgery is done done well the other number to take away from this is what percentage of the time are they able to actually achieve that correction of the spine within their targeted range and it's about 35 of the time huh and so you are now uh fda approved to do this uh and so when does this go from being science fiction to reality well as you can see behind me it's happening now we're happening on live patients yep fda clearance was achieved in april of this year which is a major milestone um we will likely be in live surgeries uh again at the university of washington doing first in human which is what's called kind of in the industry within the next few weeks jason wow that's amazing how many years did it take you to get to here in terms of research and investment i know you can relate uh seven years in the making overnight success seven years in the making so seven years to get this you know uh two patients and the venture community tends to have a 10-year window and so this is one of the great conundrums of our time you could do this inside of a university uh which is a 30 or 40 year process of grants it's basically your entire career and then if you look at the venture community we have a 10-year window and so i know that my friend matt acco is an investor from dcvc absolutely um how does the venture community embrace a technology like this knowing it's going to take you seven eight years to start bringing the cash register and that's when the end of the venture engagement and they need to start liquidating and giving returns to their lps so tell us about the the conundrum of academia versus venture and and where you stand on that spectrum because it seems like this is a really hard needle to thread yeah exactly and i went back and watched your interview with matt in 2019 and oh yeah by the way you were good then i'm not just blowing smoke you've been working on this man you're really good at this well you get your 10 000 hours in like a surgeon you can become a surgeon as a world's greatest moderator that's right.

45:30That's right. But it's true for venture as well. And I think what our friends at DCBC have done well is they are great at deep tech investment. And if I recall, when Matt laid out the framework for you, he talked about, we're not investing in companies that are making a case that there's a new physics that is to be invented. This is not paying a bunch of PhDs to go do it, you know, deep research to see if it could ever work. Right. And so there is a sweet spot where, we do real-time light field rendering. And once we had gotten a demonstration of that working well, it was a beautiful demonstration and it could actually be run on a laptop and with an Oculus DK2.

46:10And so you need to see that kind of a proof point. And that's the time at which investors will come in. And we need our friends who are angel investors and our micro VC investors and all those folks, high net worth individuals to lean into these kinds of technologies and bridge the gap to uh an investor like dcbc or lux all of our good friends across the valley who support these kind of things yeah and it's uh really needs to be more of a 15-year window i think uh which is what steve jervison's future ventures they just told everybody we're going to do 15 years they told helpies tell me about the fda approval process you know we there's a lot of criticism of our government here in the united states and we're dysfunctional uh yada yada takes too long is when you look at the fda you know having to live under approvals etc does it feel like we are high functioning in terms of our approval process and getting through this or does it feel like it needs to be improved i'm trying to be generous here because i know you have to work with them yeah i appreciate that i was going to make a comment about that um first of all, it's a highly regulated environment for a reason.

47:19This is, you know, a lot of stake. It is a lot of stake. Is the process perfect? No. But you have to look at the regulating agency as a partner. Right. And it's a process where you build a relationship, you're earning trust, you're building hardware and software development practices, which are documented, that they are that an agency come in and look at those and say, Why did you make this decision to this change to of code, right? And I think this is foreshadowing for the conversation about AI that we'll get further into here. And you have to be have it be entirely explainable and auditable end to end.

47:56Now that's difficult to move fast and break things in that kind of an environment. But it is very intentionally highly regulated. So this would fall under what's called an FDA 510k process, which is really a smart highway that was built. And that's a reference to the sort of the code and the regulations that says, if there's a product that has some substantial equivalence, right? A camera system that does this and a camera system that does that plus more, you can start with a 510k, which will accelerate your process to market. There's certain stringent proof points you have to hit. And if it's accurate and if it's faster or whatever those criteria are, fda if you meet those bars fda will give you a basically a slight fast track to market got it so if it recognizes your building and you're standing on the shoulders of existing technology they can fast track you a little bit and not make it you start from zero uh and i guess in your case the sensor arrays and being able to uh build off of that which already exists um what would help you the smart strategy there which is what we've pursued is not try to boil the entire ocean day one that's going to be confusing for a regulatory body you know what is the key innovation here and what what are you enabling today and where are the proof points and where's the 10-year patient outcome study so that's not a great strategy for dealing with fda that makes them nervous understandably yeah so what you want to do is go stepwise through build a platform get that cleared then come layer different software packages and upgrades and as you said expand the surgical adjacencies go after knees and cranial and those kind of applications if that's where we're going um and those are those are essentially additions to the regulatory strategy and that's much less confusing for fda to get to this point in time seven eight years about a hundred million dollars to get a product like this over the finish line yeah it's a good guess um we went to very educated folks in the space early on and i'm a very competitive person i don't know about you but I got them to kind of outline what it's taken to bring DaVinci and other systems to market and it was about 10 years 100 million dollars was the estimate for for deep tech execution and innovation in the space I'm happy to report that we're at least three years and tens of millions of dollars ahead of that schedule fantastic which which is exciting and it affords us the opportunity to engage with the investment community to build a lot more on top of the platform we've already built and launched well listen uh it's just great that you're out there doing this kind of work um and it's great that all of this activity from smartphones and video game consoles and pc-based video games and nvidia storage i mean you couldn't do this without having a massive amount of storage you couldn't do it without having these sensors dramatically drop down in price although you can spend a little bit more money on sensors that sensor array has got to be 10 20 000 worth of sensors i would take it when you're that's not a bad estimate jason yeah it's well i mean if you just think self-driving cars have 40 000 30 40 000 worth of sensors if they're using lidar this feels like it's a little more it's directional so you don't need to have it in four different directions you'd have this in one direction basically so it would be slightly less but we have an advantage in that it's on a robotic arm and so we know where the robot is so it gives us a computational uh head start we don't need to compute the entire world right we're very focused on the area that matters most to the surgeon and that accelerates things quite a bit well i know you're hiring and so everybody should go to the website now uh which is pro p r i o vision.com and uh you go to the careers page what are you looking for in terms of needs in terms of you know high-tech folks who can come and help you reduce suffering and increase the availability of these high-end surgeries to people around the planet?

51:59What do you need? Yeah, thanks for shining a light on that, Jason. To do something as ambitious as this, right, it's not one type of software engineer or hardware engineer you need, right? We need computer vision, AI, robotics, augmented reality, medical device, software and hardware engineering. We have all those things here in Seattle, and I guess we're growing on all fronts. so if you're interested in solving meaningful problems that are unsolved today clinically um both in spine surgery and beyond we would love to talk to you yeah go to the career page i see you have unity to engineer so you're using unity as the 3d modeling environment uh um it's it's useful for a lot you alluded to it you nailed it so the video game development environment we not only like video game players to come work here but we like uh you know software engineers who've worked in that environment.

52:50And part of that is thinking ahead to applications that other companies and teams will develop. Right? I don't need to reinvent all of the wheels or every single app that's going to live on our ecosystem or have an API to plug in to our data. If you are skilled in developing a unity environment, you have a product that you think could be amplified. Yeah, by working with proprio with our singular data set, we would like to hear from you too. Amazing. Well, listen, continued success. Thanks so much for sharing this incredible progress. And And so coming to an OR near you soon. Thanks so much, Gabriel, for joining us.

53:24And we'll see you all next time on This Week in Startups. Bye-bye.