#42 - Anatomy of a startup - Q&A with Bill Hunter, CEO of Canary Medical (55 mins)

Brought to you by Tiger Buford.

Welcome back to Inside Orthopedics. This is Tiger, your orthopedic industry insider and retained recruiter for early-stage orthopedic companies. This is episode number 42 in a series Startup Series titled Canary Medical, a Startup Bringing the First Smart Implants into Orthopedics. I first wrote about the coming of smart implants on OrthoStreams back in 2012. In this pod, I was thrilled to talk with Dr. Bill Hunter, CEO and founder of Canary Medical, who's actually doing the work. Bill and his small team are creating orthopedic implants that talk and will predict problems before the patient or healthcare provider even knows there's a problem. This is a great insight into the future of orthopedics. Enjoy the conversation. All right, so I'm live with Bill Hunter at Canary Medical. Welcome, Bill.

Thanks, Tiger. Thanks for having me.

Yeah, yeah, thanks for your time. So this one is near and dear to my heart, as I guess we met online. I wrote an article about 10 years ago about this. I thought smart implants were going to be the future of orthopedics, and You're the first one that's done it. So I can't wait to have a chat with you. So, well, first of all, tell the audience who you are and what your background is.

Yeah, so I'm completely unqualified in orthopedics. I'm a doc by training, but I've spent most of my life in cardiovascular. So my My first company was a company called Angiotek that was involved in drug-coated stents. So we developed the TAXUS stent and paclitaxel balloons and all that kind of stuff. And so it was a big cardiology interventional radiology business with a few thousand employees, but not in orthopedics. And then my second stint was at a pharmaceutical company called Cardiome that made antiarrhythmic drugs. So my background is really, really in cardiovascular. But when we were at Angiotek, we developed a wound closure device called Quill, which was a barbed suture. And it ended up that that product was used primarily by plastic surgeons and orthopedic surgeons, and during— particularly once doing total joint replacement. And that's how I met Fred Kushner, orthopedic surgeon, and that's how my— career kind of touched into orthopedics and how I ended up where I am.

Interesting. What year was that?

I probably met Fred, I'm gonna say, you know, 2005, '06, somewhere around there.

Wow.

And he was, he was one of the first ones to be using Quill to close after total knees. And he did a lot of our, not just our design work, but he did a lot of our clinical validation work, and we just kind of kept in touch from then on.

All right, and so when was the foundation or the very beginning of Canary?

Yeah, so it would have been about the time you were writing your article, 2012 or so. You know, kind of just one of those weird things. The premise behind Angiotek was that medical devices fail because they get inflamed and infected, and there were pharmaceutical compounds that were anti-infectives and anti-inflammatories, so why not put them together? So the business was really an attempt to cross over an outside technology into medtech. That's how I ended up in medtech. And Canary was really, as I'm fond of saying, old dog, same trick. It was the concept that Hey, you know, we've got all this sensor technology, we've got all this— all these connected devices in the world. Why in the world are not, you know, major medical devices connected to the internet? Why do they not self-report? Why do they not self-diagnose? It just seems, you know, so inherently obvious. And, you know, so the concept was to bring, once again, technology that was— you wouldn't traditionally think of as medtech technology and bring it into medtech as a—

So you founded the company in 2012 or thereabouts?

Yeah, yeah, about that time. I started writing the IP. I still had a day job at a pharmaceutical company at the time, but yeah, so I started writing the IP and then a lot of the engineers that I'd worked with and folks I'd worked with at Angiotek who, although Obviously were, you know, had been working on drug coatings and other stuff, but, you know, had worked on some pretty sophisticated medical devices. So, so they joined me and, you know, that's just slowly gained momentum from that point forward.

And you started by yourself and you—

Yep, by myself, by Fred Kushner, who's a surgeon at HSS, by Jeff Gross, who was my lead engineer at Angiotek, and David McMasters, who had been an IP attorney that I'd worked with for almost 30 years now. So, you know, just kind of got the band back together and everybody, you know, worked for equity, not for salary, and that's how we got going.

Got it. Well, that's a typical story. That's— so everybody had a day job somewhere.

Everybody had a day job, yeah, yep.

I love it, I love it.

And so the patents took probably 2 or 3 years to issue and Yeah, so we wrote a fair bit of IP, and PersonaIQ and the knee, which obviously we'll probably spend most of our time talking about, was only part of it. We basically took the premise and said, okay, if you're gonna put a sensor on a heart valve or a coronary stent or a hip or a knee, what sensors would you use and what would you want them to do? So it was a much more broad approach. It wasn't orthopedic-focused. It was really focused on, you know, kind of what I would call the sensor and, you know, medical device overlap. And truth be told, we ended up in orthopedics not just because of Fred, but that was a big part of it, but because the technology 5, 6, 7 years ago was still pretty big, and we needed devices that were large enough to actually hold the components that we needed to use.

Yeah, there's a lot of— I used to design total knees and there's a lot of space in the keel. And yeah, it's perfect. It's actually perfect. And I'm sure over time the batteries and things will get smaller, but it's a great starting point.

I'll give you an example. So we started building this stuff, I don't know, probably 7, 8 years ago when we really started to try and make them. But now we have an implant that we're making for brain aneurysms, it's actually a millimeter cubed in size. So we're able to get a power source, sensors, and transmission technology into that small a package. But that was not the case 6, 7 years ago.

Yeah, yeah. Wow, that's really cool. Well, yeah, it's just technology is your friend with minimizing the size of the device. Interesting. So how did you guys— so you thought about putting sensors in Total knees. So how did you— did you do animal work? Did you— how did you do the early proof-of-concept stuff?

Yeah, so, you know, we met with a few orthopedics companies who were, perhaps not surprisingly, weren't overly enthusiastic. They looked at it and said, well, you know, 2 things. One, you know, I'm not so sure that the industry will support adding cost to the implant, which I'm I'm sure you've heard a million times in your career. And secondly, you know, go do it. Like, show me how this works. Like, it's a great intellectual exercise, but, you know, show me how it works. So we spent the next couple of years just proving that we could do it. And believe it or not, we didn't do so much animal studies for the obvious reasons. You know, there's no animal model of a total knee. But there's a bunch of biological surrogates that you can use that that perform like human tissue. And so we would, you know, we had to prove that we could build the thing. We had to prove that we could power the sensors. We had to prove that we could transmit from inside the body out. In fact, you know, transmit several meters away from the body so that we could actually capture that signal and communicate to the internet. So the first couple of years was really all about power and transmission and proving that, you know, we could come up with something that worked. And then, you know, it became more about figuring out what the doctors and the corporates wanted. You know, how long did you want us to monitor? How many times? What did you want us to do? What did you think would be commercial? And, you know, people said, we need you to monitor for 20 years, and we need you to monitor a lot at the beginning when the patient's recovering, and then we need to follow them a little bit less over time. All those decisions about how often you monitor and all that type of thing, has an impact on how much power you need, how big your battery is, and all those kinds of things. So a lot of back and forth for a few years, just trying to refine it. And then, and only then, we actually bypassed animals. We went right from there into cadaver studies.

Yeah, that makes sense. You need a lot of thick tissue to go through.

Yes.

Interesting. It probably— your technology development was probably faster than the big orthopedic companies could figure figure out what they needed. They're not really good at challenging their own legacy products and thinking out of the box. So that's really interesting.

Well, it's funny. I've had kind of 2 experiences in my career. One was when I first started telling people they should put drugs onto their coronary stents. And I literally had an executive pull me outside of the building and point to the name on the building and said, do you see the word pharmaceuticals anywhere there? And adding drugs and pharmacokinetics and toxicology and biodistribution, that's just not something that most medtech cardiovascular companies had any interest in. Obviously, ultimately, the technology evolved. But the same was kind of true when we got started. Really, this device is— its closest comparator is a pacemaker. You know, in orthopedics in particular, you know, you don't have a lot of active implants. So even though you've got some really skilled engineers and you've, you know, you've got people who've been through the business a long time, they hadn't spent a lot of their time building that particular type of implant.

Right. Interesting. Yeah, it's— that's really interesting. And when you say active, you mean electrical components?

Yeah, yeah. If our first indication had been in cardiology, this would have been a no-brainer. There's, you know, Right, right. 4 companies in the world could build it in half an hour. But in orthopedics, you know, quite different.

Yeah, there's a— I guess there's some bone stimulators that are electrical, but that's about it. That's interesting. So, and how was the company funded early on, Bill? After, you know, you guys didn't take a salary, and then at some point you needed real money. Yep.

Well, our first company had been good to us, so we were able to self-fund for quite a while, and that allowed us to stay quiet and do what we wanted to do. So for a number of years, we just funded the business ourselves. It was even harder on my founders because they not only had to give their expertise, but they provided funding too.

Right.

You know, we self-funded and we didn't take what I would call professional venture money until 2019.

Okay, got it. And that's like just Series A?

Yeah, we did one Series A in 2 tranches. You know, we probably did somewhere in the neighborhood of, you know, $8-ish million in seed funding. Again, you know, kind of friends and family internal. Then we did $40 million in Series A in 2 tranches over 2 years. And fortunately for us, we've had a lot of business development success. So, you know, right now we have $65-odd million in cash on hand, so, you know, more than we have raised.

Wow. Building—

I'm sure we'll talk about this too, but building the support structure for the data and doing the whole thing, that was probably close to $100 million worth of work.

Wow.

A lot of people really underestimate what's involved. Great, you've got the data, but capturing it, processing it, user interfaces, storage, AI, machine learning. It's expensive to turn that into a usable product.

Let's get into the product because I want the audience to understand what's the hardware first, and then we'll talk about the software side.

Yeah, it's pretty straightforward conceptually. You need a power source no matter what you do. In total joint, the device is big enough that you can use a pacemaker battery, which is great because the density of power delivery from those batteries has been getting better, you know, record speed. Then, you know, you populate it with sensors. In orthopedics, we use 3D gyroscopes, 3D accelerometers, step counters. Obviously, orthopedics is all about movement. So it's really about collecting movement in 3-space. So that's the sensor arrays we use. We use something quite different in vascular. And then you need a transmission technology, and there's several to choose from. But in our case, we use the mixed radio system that's used by pacemakers because it had a 20-year FDA history. And we wanted to keep the components things that FDA had seen before since this was Really a first-in-class device. Then there's a bit of memory there, so we can store data. If you happen to be away from your home for a few weeks, that's no problem. The device will keep recording your data even if it's not connected to the internet. And we can also keep track of— your knee will always know which doctor put it in, where it was put in, which parts were there. So that, you know, if we needed to find a particular part, we could actually find the 20 patients who had that part, you know, for safety reasons. So that's the basic componentry of it. That's what— that's the hardware that goes inside. The battery is so powerful and the power draw is so low that, you know, we think we'll be able to pull, you know, in time, 20 years' worth of data off of that implant.

Excellent. And so, and how big is this? Describe it since we don't have pictures here.

Yeah, yeah. So right now it fits into kind of a standard tibial extension. So it's bigger than a stubby, but it's not as long as the longer extensions. It's probably the size of a AAA battery.

Okay.

For a visual. Yep.

Got it, got it. And it fits in the tibial stem, and so a manufacturer would have to basically just make a cavity inside the tibial keel and just seal it.

Yeah, so the typical Persona tibial extension is a solid piece of metal. So Zimmer provides us with hollowed-out version of that, and then we put the electronics in it and build it and send them back, you know, the finished stem. So, from a product point of view, if their docs know how to put in Persona, they'll know how to put in Persona IQ. There's a couple of minor things about how you put it in because you can't hit it with a hammer.

Right.

It is, you know, it's the same stem that you would use if it was solid metal.

Got it. Interesting. Okay, and so you send it back to Zimmer and then it goes into a patient in, you know, somewhere Philadelphia. And then the day of surgery, how does the device know the patient and the components?

Yep, it was a great question. So, you know, the good thing about knee replacements is it's an elective surgery, so we can, you know, prep the patient in advance, if you will. So the— we have a little base station, it's about the size of a modem, and it needs to be in the patient's bedroom and it needs to be connected to the internet. And it's literally as simple as if the green light is on, it's connected. If it isn't on, we will know that and we'll send somebody out to connect it for you. But once that's done, that's the entire patient compliance. So it sits in the bedroom once a day, you know, at night. It downloads all the data that the knee collected during the course of the day, and that's pushed automatically to the cloud. So We go to the patient when the patient's going through their preoperative assessment. That's when we provide them with the base station. We make sure that they go home and they get it all set up and it's running so that when they come home from the hospital, it's already up and running in their house and they're not sedated and groggy and trying to figure out how to hook up their knee. So that all gets done in advance. And then with respect to the device itself, to preserve the battery, it gets shipped in deep sleep. And then there's a base station in the operating room, and that wakes up the device, tests it, makes sure that it's working, makes sure that it does all those things before it's implanted. And just as importantly, you know, does those things we talked about in terms of keeping track of all the different components of the knee that were used and any other important things that might be needed for the patient record so that that gets into the system as well.

And everyone's got a unique identifier?

That's correct.

Yeah, that's really interesting.

Wow.

And so, and the data that's captured is lots of motion, but explain how detailed the motion is and what—

Yeah, so, you know, we've kind of had this chicken and egg phenomenon, which is that we can't put these in until we're approved. So, you know, all the stuff that we've done is not from implanted devices. However, we, you know, we've been making these things for a while, so we've been able to strap them onto people and have them walk around and validate our whole system and, you know, be able to do the things we want to do. So, the easiest way to describe it is that you've got these 3D, so X, Y, and Z gyroscope, which is angular velocity. Then you've got 3D accelerometers, which is acceleration. And so you're getting all this 3-dimensional data, points in space. If you think about the human eye, we see at about 25 frames per second. These devices collect movement at 800 frames per second.

Wow.

The system that we used was actually developed initially for drones as the navigation system. And as you can appreciate, those drones exquisitely know where they are in 3-space, right?

Yeah.

So, it's very precise 3-dimensional information. Anyways, the reason I point that out is that I think the technology, in our mind, had some benefits that I don't think we even really conceived of when we started working. Because of the richness of the data and the fact that we have 2 different sources of it, so we have kind of 6 channels of information, we don't just see macroscopic movement. So we don't just see the leg moving, the 25 frames per second that our eye sees. We also see the movement of the implant itself. So every time your heel hits the ground, it sends a shockwave through the entire skeleton, and that's gonna be transmitted up through the tibia, and it's also gonna be transmitted up through the implant, right?

Mm-hmm.

And so we actually see the implant move within the bone. So we can tell whether that implant is firmly in place or whether it's moving around. Think of hitting a tuning fork and then putting your hand around it. If the tuning fork's just vibrating, it has one signature pattern. If you dampen it with your hand, you get a different signature pattern. So we're able to get—

Yeah.

Exactly. So we see macromotion for sure in very— Finite detail, but we believe that we're going to be in time capable of seeing very micro-type motions and patterns of behavior of the implant itself, which I think have the potential to be really important going forward in detecting other things.

So once you figure out what attributes are seen with a loosened tibial plate, then you can You can catch that early.

Yeah, so, you know, the analogy I always use, probably because of my background, but I always use the analogy of an ECG, right? And so if you kind of think of an ECG tracing your mind, something most of us are familiar with, it actually has a whole bunch of different elements to it. So first and foremost, it tells you what's going on. It tells you the anatomy of the heart, right? If you have a problem with a P wave, you know that the issue is in the patient's atrium. Well, the same is true with a gait tracing, right? We all walk slightly differently, but we all do the same things, right? Our heel hits the ground, we roll on the ball of our foot, we push off, we swing our legs. So if you see an abnormality in the tracing, you know where in the gait cycle that abnormality exists, right? So you have anatomy. The second thing you have is rhythm, right? So in ECG, you can tell the difference between AFib or any other rhythm abnormalities by looking at the pattern. The same is true here. A person who limps, a person who has a fixed contracture, they walk differently, and their tracing is different than a person who's walking normally. So you don't just get quantitative gait information, you get qualitative gait information. So it's not just that Mrs. Jones took 500 steps, but what was the quality of those steps? Was she limping, or were they long athletic strides, or were they something in between? And then the aspirational part, which is gonna take some time, it's gonna take some work on our part, is the analogy I use for heart attacks. So once upon a time when I practiced medicine, if you look at an ECG, You can tell that the patient's having a heart attack. Well, ECG doesn't measure blood flow through the coronary arteries. It doesn't have anything to do with it. But it causes such characteristic changes in the ECG, and when you've seen them enough over a longer period of time, over enough patients, they ultimately, of course, became diagnostic of heart attacks. So when you ordered the ECG, you can say not only that the patient's having a heart attack, but which artery's involved. Well, we think the same is probably going to be true here. You're going to see both macroscopic and microscopic changes if a patient has a contracture or if a patient has micromotion or instability or potentially even if a patient is developing an infection because they will walk differently, the implant will move differently, and we may see patterns that are diagnostic or at least suggestive of that type of abnormality.

Wow, that is really cool. There's going to be benefits That nobody can even predict yet. That's amazing because you've got somebody looking at the patient in incredible detail 24 hours a day. That is really interesting.

I'll give you a quick background. It's the wrong medium for this, but we have a video that we show, and we show a person walking on the treadmill with the device, and you can see the tracing of what It looks like you. You look at their tracing. Then we put a knee brace on them that restricts their range of motion, right? And they walk on the treadmill again. And just looking, eyeballing it with your eye, you can see that their knee tracing is different, right? And so what you can do is you can say, okay, well, not only does this person, you know, is not walking normally, but because we have 3-dimensional information, we can say, hey, it's You know, the restricted extension of 15 degrees, you can actually measure it. You can see it. Now you go one step further and you do the machine learning type aspects on it, and you have patterns of people limping and you have patterns of people walking normally. Now after a relatively small number of samples, you can show a tracing to the computer, and without knowing anything else, the computer can tell you whether that's a limping tracing or a walking tracing. It learns, right?

Yeah.

And so that's where, you know, a lot of work to be done here. I don't want to— you know, this is all aspirational stuff. We've got to do our homework here. But the potential here to do different things is pretty exciting. And, you know, once we are hopefully approved and once we start to collect data from inside the body, you know, our hope is that the richness of that data will be will be really insightful.

But no, you make a good point because AI is really good at pattern recognition. That's one thing AI is good at. And so once it says, okay, this person's limping, this person's got a loose implant, this person's got an infection, once it learns the data, the motion data, then it can preventively You know, scream out, this patient's about to have a problem.

That's really cool. And so, you know, this once again, I can't prove this. We haven't done it yet, right? But at least in theory, take something like infection, right? The hope is that once you have 25, 50 patients with infection. You'll then have this, you know, we have tracings every day for the first year from the patient. We have tracings 3 times a day every day for a year. So we have all kinds of data coming in. So once we have 50 patients who had an infection, you know, is there, are there changes that are characteristic of infection, just like that ECG? Now, do we see an elevated ST segment? You know, do we see a change? that says, hey, this person is, you know, has an infection. And then working backwards, again, to use the cardiology analogy, if I have a patient who has chest pain, what do I do? Well, I do a stress test, and we try to elicit that response because we want to diagnose the risk of myocardial infarction before we have myocardial infarction. And that, of course, is the holy grail. So do we understand the pattern of what an infection looks like? And then can we work backwards from that and come up with early indicators that allow us to say, okay, you know, this is a patient that's recovering normally, you know, fantastic, we, you know, we'll follow them this way, but this patient over here is showing signs that I don't like, maybe this is someone I should come in and work up, you know, further for something that, you know, could be developing.

Yeah, I always thought that infections would come from, A temperature sensor, but it makes sense you could back into it with motion. That's really interesting, especially if a patient's motion changes. That is really interesting.

Well, you know, it's funny. It's funny. So the first part of my career was spent trying to modify the biological reaction to an implanted body, right? That's what resynosis is. And so looked at an awful lot of biological sensors. And, you know, we went through that too, but the problem with temperature and pH, for example, which would be the obvious ones you'd use, is that that sensor would have to be exactly at the spot where the infection was, right?

Yeah.

And so you'd have to have it on the femoral component. You'd have to have it on the poly. You'd have to have it on the tibial component. But if instead you can see signatures, much like an electrical pattern can tell you about a vascular injury, if you can see that, you know, that covers the whole implant and that gives you a chance to pick it up regardless of where that infection may be developing.

Yeah, that's really interesting. Yeah, it's such a local phenomenon that—

Right.

You can hack into it with motion signatures. That is really interesting.

Wow. Well, the other thing is, you know, back to my previous life, the problem with surface sensing is that all of our implants, all implants ultimately become encapsulated, and once they do that, they don't really detect— you start detecting the microenvironment, not the macroenvironment. It's been the problem with diabetes sensing for a long time, right?

Interesting. That makes sense. They're sort of blocked off from Seeing bone, real bone.

Yes.

Let's talk about— man, this is— I can see the future.

You did, in fact, 10 years ago when you wrote the article.

I mean, you're doing it differently than I envisioned. Who's the guy that said the future is already here, but it's not evenly distributed?

Exactly.

Yeah. What's the business model? So you obviously have a partnership with Zemer.

Yep, yep.

But are you a hardware company or a software company or both?

Yeah, that's a great question too. And the answer is, it really depends. So the best way to describe it is to kind of look at 2 extremes. So we're working on building a device that can detect if your aneurysm is going to rupture.

Right.

Now, let me take abdominal aneurysm as an example, right? You have an 8% cumulative risk that your aneurysm, if you put in a stent graft, that your aneurysm will rupture in any given year. So by the time you're out 3 years, your risk is, you know, 24%, and it just keeps going on in that amount. In any event, that's something you'd want to know about for the rest of your life, right?

Yeah.

If you were monitoring that, you'd want to know if there was pressure building up in your aneurysm every day of every month of, you know, the rest of your life because, you know, that's life-altering type information. In that situation, the device isn't as important as the data. And in fact, you're less worried about making money on the device because the data is where the value is. And so under those circumstances, you You're really more of a data company, and I'm not suggesting you give away the vascular graft, but how much you get paid on the graft probably doesn't matter as much. At the other end of the spectrum, think about making a smart screw, which is something else we're working on. And let's think about just a simple femoral neck fracture. Well, under those circumstances, monitoring to see the callus formation or monitoring to see if the screw is broken or bent or backing out or You know, where they get normal healing. That's valuable, but it's only valuable for a few months. And so you better make your money on the device because the data lifespan is relatively short.

Yeah.

And so under those circumstances, you're really more of a device company than you are a data company. And the knee is kind of somewhere in between, right? The first year is gonna be really important, and, you know, that data in the first year is really gonna be where a lot of the insights are gained. And after that, you're probably looking more at annual management, right? Yeah, once a year, twice a year kind of thing. So, you know, it falls somewhere in between. So the answer to your question is it really depends on what clinical problem you're trying to solve.

Okay, so, well, can you share what the cost is? I mean, I'm sure the cost is going to go way down, but what's the cost of the hardware that goes into Yeah, what I will say about that is we're building a pacemaker, right?

So our cost is never gonna be like $100. This is a pacemaker, right? We have a pacemaker battery, we have telemetry in there. The sensors and the boards are all getting cheaper and cheaper, but batteries are not free and the transmission technology. So we'd We're analogous to what it would cost to build a pacemaker.

Right, right. And then obviously volume's your friend. So when you start making hundreds and thousands and, you know, it's—

It matters a lot. Like, the volume— volumes matter a lot. The good news for us is that, you know, as we start thinking about making hips and shoulders, you know, a lot of the componentry will overlap. And so we can order bigger and bigger quantities. and help reduce costs that way as well.

Right, right. Okay, so we got a patient with a knee with CHIRP. I guess the product name is CHIRP.

Yeah, that's our generic name.

So they got a total knee with a CHIRP inside, and it's monitoring 3 times a day for the first year. And then where does the data go? The data goes from the So, the bedside thing through the Wi-Fi to the cloud, and then where is that— who manages the cloud?

So, that goes to our closed cloud system. And perhaps not surprisingly, from a regulatory perspective, our product is not just the implant. Our product is the implant, the transmission technology, the the data processing and the user interfaces that the patient and the doctor look at. So it goes to the Canary Cloud system, although the user interfaces that people will be using will be the Zimmer Biomet, the MyMobility interfaces that are already out there.

Oh yeah, that's right. They got the app. Yeah, that's great.

Yep. Yeah, no, they have a wonderful user experience there that— so they already have You know, the patient management tools, the rehab tools, all that kind of stuff. So this will just fit into what they already have.

Perfect. Yeah, it's a perfect partner for you. So is there— so there's data coming in. So let's just say a patient has an issue where the, you know, the base plate's loose or there's an early infection. So how— I mean, I know we're talking about the future, but Who is alerted? How are they alerted? What happens next?

Yeah, so you know this is you know this is this is all in a perfect world. So you know what where you'd like to get to, you know at some future date, right? Is I kind of look at it as. A multi-step process. So the most often quoted statistic, although I'm not sure how accurate it is, but the most often quoted statistic is that 80% of people are happy with their knees, right? 20% of people are dissatisfied for a collection of different reasons.

That's ballpark, yep.

Yeah, and so I think kind of your first level is, can we identify those 80% of people who are doing well? One, because You know, they probably don't need as much physio, they probably don't need as much follow-up. And obviously, can we identify the 20% who seem to be lagging behind? You know, we measure range— we measure all kinds of other things too, right? So we measure step count, cadence, walking distance traveled, walking speed, all those things. So you're getting a real measure of what the patient's activity is, and then we're also measuring the range of motion and the gait and all the other things I talked about. Is there a sequence of events that says, hey, this is the pattern of a patient who needs to be called in for more follow-up? That's kind of the first bifurcation. Can we tell normal from abnormal? Then the next level of that is, all right, among those 20%, can we identify those where early intervention would be useful? If I've got a patient who is having problems with contracture, well, if I don't figure that out until 3 months, you might be taking that patient to the operating room for manipulation under anesthesia or even some sort of scar tissue release. But if I knew that at 21 days, that might be a patient that you could send to physio or you could send home care out to manipulate the joint. Break up that scar tissue before it becomes too tenacious. So to me, the second tier is identifying those recovery periods earlier.

There are those—

find the 20%, that's valuable, and then find the 20% as early as possible when you can still intervene. And then of course the holy grail is, can you start to look at it and say, okay, this is a patient that is heading for a particularly significant problem like an infection, and can I get to that patient where they could be treated with local antibiotics and exchanging the poly instead of, you know, them not showing up until you've got to do a, you know, a total replacement.

Right. So that's—

and, you know, that's going to take a lot of patience, that's going to take a lot of knowledge. From a regulatory perspective, you know, you're going to have to develop that almost as you would a diagnostic test. Those are all separate regulatory submissions to the FDA, all evaluated based on the data that you collect. But that's kind of the path that you would like to progress on the technology. I think in the beginning it's going to be fairly straightforward, just looking at how well people are moving around, seeing if they're active or inactive, and then hopefully we can just continue to build on that as we get more information.

Interesting. Wow, that is— yeah, this is great. So what's the status? I know you've got a regulatory submission in the US. First of all, was it just— is it a Class II, or was it—

Yep, the whole system is a Class II. It was a de novo 510 without clinicals.

Okay.

And we were granted breakthrough status, so we're under the breakthrough program, which has been I think has been really, really helpful for us.

And it's cooking at the FDA?

It's cooking at the FDA. I make it a policy of not handicapping the FDA, but we're well down the pathway. The beauty of the breakthrough program is that it is an interactive review, so we've been You know, getting questions back and forth. It really— they have helped us a lot because, you know, it was a first-in-class product and, you know, I think they wanted to make sure that it was developed in a way that they were comfortable. So I'm hoping we're pretty close to the goal.

Got it. Got it. And so let's talk about future. So you mentioned a couple, you know, products, Total Shoulders, Total Hips, Probably. And then other markets. So while you were talking, I was thinking, you know, countries where they— where there's a national healthcare system where they have a value on treating a patient for life may be really interesting.

Yep. Yeah, I practice medicine in a socialized system. I'm Canadian. And I've had conversations with the NHS in the UK, with senior folks even at NHS. There is, you know, I think the holy grail, and, you know, if I can shout out my partner here and give them credit, I think Brian Hanson and Yvonne and Rachel, I think they've been really fairly visionary, but the holy grail here is Hey, can we then learn from the implant? And does that implant inform not just best practices or how we rehab our patients, which is really important, but does it start to inform how we put the implants in?

Right.

Do we learn something from PersonaIQ that tells Rosa how to put the implant in differently in this particular patient who has these particular characteristics. And so, you know, we've built a wearable so that we can look at their gait preoperatively and get an idea of what that's going to end up. And will you be able to do those things to really reduce the patients that are unhappy with their knee, but potentially even change the practice and standard of care going forward? And, you know, that's— when you talk to the nationalized systems, that's what they're really excited about. They look at that and they say, okay, If we've got access to all the data of all the patients, that sounds like something that could be a national initiative.

Wow, that really does sound interesting. Think about the robot aspect. So if, for instance, patients do better if they have a little bit of tibial tilt, tibial baseplate tilt or something like that, then the system learns and says, do more of that. And that can be programmed into the robot. That's really interesting.

Yeah, I mean, the— what, you know, more is more, right? I mean, the more you learn, the better your chance of doing this. And, you know, or so my data scientists tell me. But, you know, you could imagine that, you know, right now, I don't know, what is the normal number of steps that a 65-year-old male with a BMI of 25— who knows, right? But with time, you may know that, and you may figure out how you stratify your patients, and you may find that different patients— to your point, I'm not an orthopedic surgeon, so I probably won't have the right description, but do we use different angles for different body types or different gait patterns? That would be really exciting stuff to know.

Interesting. Wow. Well, this is great. Just let me ask you a broad question. Where is Canary going in the future? I mean, where—

Yeah, you asked me earlier about orthopedics and the business model. And, you know, orthopedics is a tough business model. You know, you're working in a capitated system. You're working in a DRG that, you know, a bundle system. And if, you know, you take money out one place, it's got to come from somewhere else. And the business model is tough. And, you know, we— you and I had to talk for, you know, half an hour about how you would do it in terms of understanding where you might be able to extract value from the data. There are other indications where the value is obvious. Like I said, I'm an old stent guy. I believe that if you knew if your stent was open or closed and you could do that, I'd I think you wouldn't use any other stent. So we're spending a lot of time on the vascular side. That's a much higher bar. You know, what we're doing in orthopedics is not life-sustaining or supporting, right? In cardiovascular, that might not be the case, but yeah.

Interesting. So you're looking at other medicines in cardio's An obvious one.

Cardiovascular is big. You know, there's even some devices like heart valves, there's enough room, and if you know the pressure on one side of the valve or the other side of the valve, you know, the neat thing about vasculature is that it's mathematics. If I know pressure and flow, you know, you can calculate cardiac output and ejection fraction and systemic vascular resistance. It's a little bit like having an ICU inside of you for the rest of your life. So, you know, that I think is probably a fruitful place to work. And we expanded our relationship with Zimmer, and so we're going to work on hips and shoulders with them. So hopefully we can fill out their product portfolio.

And have they made an investment? Is it a strategic— don't tell me anything that's not public, but—

Well, I mean, I think it's very strategic. They are not equity holders in the company, and we are independent organizations. But yeah, very strategic. I mean, Zimmer, we're the— you know, as one of my mentors always said, you know, when you're a mouse dancing with an elephant, never forget which one of you is the mouse. I mean, you know, they're the elephant. That's a good one. You know, they have all the hardware. They have, you know, an amazing commercial sales force. But they also had been fairly proactive, right? They already had the relationship with Apple. They already had Rosa. They already had MyMobility. And so we are just really fitting into, you know, hopefully one of the last pieces of their puzzle. And, you know, I think their vision in many ways is much broader than ours because they're, you know, they're the ones who are hoping to be able to Combine preoperative, intraoperative, and postoperative data to get new insights in patient management.

Well, you definitely chose the right partner. I mean, they're really— I mean, Zimmer tries to do spine and everything, but they're really— they are a total joint company, and that's what they do really well.

Yeah, I mean, we've—

They're the biggest.

Yeah, they are, and we were really lucky. I mean, And, you know, if we were going to have one partner, they were the obvious choice. But even to the extent that, you know, Persona is their flagship franchise, and, you know, a company that was less ambitious might have put us on, you know, a smaller product line and taken a wait-and-see approach. I mean, they put us on their number one platform. And so You know, if this is successful, they are going to be able to make it so. And, you know, that's something that we couldn't have ever hoped to do on our own.

Yeah, they have a huge marketing muscle. So, yeah, so if you get clearance, let's just say you get clearance by end of the year, so could you see clinicals next year, 2022?

Yeah, so we're kind of backwards. So the way that I think it'll end up working is, you know, we'll get approved. So, you know, and I'm certain that we will do some post-marketing studies, formal post-marketing studies for sure. But, you know, I think what you will see is that the first group, the first cohort of patients, first, you know, 3,000, 4,000, 5,000 patients, I think we'll try and do in a fairly well-controlled manner so that we can collect all those insights that we're hoping for. Because I think for this technology to really realize its vision, we're gonna wanna be able to do some of those things we're hypothesizing about. So I think you'll see kind of a, you know, less about the commercial footprint than about the learning input by going to institutions that can really help us figure out what we need to figure out. And then once we have that, we'll look at each other and have a better understanding of what we have. Is this going to be a niche product? Is this going to be a great product? But we'll look at that data and that will probably dictate what the next wave Got it.

I understand. So it'll probably be pretty quiet for a while while you guys are—

Well, I think the first time a human being walks out of the operating room with a talking knee is going to be a pretty significant event, I think. But I don't think you should gauge the story by how many get sold in year 1. I think year 1 is going to be about learning what we need to learn and And it might be, you know, a couple years out before, you know, we, we know enough to, to really, you know, have more of an impact.

Got it, got it. Okay, well, that's great. Well, thanks for your time, Bill. This is awesome. I know the audience is going to chew this up. We may have to do another, another little chat maybe in a year or so. So how can listeners— what's the best way for listeners to find out Yeah, you know, I think 2 ways.

You know, first and foremost, you know, Zimmer has invested an awful lot in this program and has, you know, is starting to be a little bit more— well, you know, tell the world a little bit more about it now that we're close. So, you know, Zimmer has a lot of information on that. There's a lot of work that's going on. AOS, both Zimmer and ourselves will be there with—

Okay.

Basically showing the things that you and I talked about, how it works, what that looks like, how the data is collected and all that kind of stuff. So I think AOS is— AOS, hoping that it goes off as planned, but AOS will be probably the big coming out.

Okay, yeah, so if you're in San Diego, listeners, in the next week, stop by the Zemmour or Canary booth. Probably right next to each other.

Yeah, we're close.

All right, and then the website. I know you guys are pretty stealthy. There's not a lot on the website, but there are a lot of news articles.

Yeah, the website, as you can appreciate is being— the website and the information we provide is also under regulatory control.

Yeah.

So once the product's approved, I think people will find our website far more informative. But right now, we, you know, we're getting feedback from the agency as to, you know, what we can say about the product. So that won't really be of assistance until post-approval.

Got it. All right, well, we'll look for the press release. Well, wow, thanks for your time, Bill. This is terrific. It's great to see— it's great to see the first company through the chute. This is definitely the future, and we'll talk again in a few months. Awesome.

Thanks, Tiger. Thanks for your time.

Yeah, thank you. Have a great weekend. Bye-bye.

Thanks.

Bye.