This transcription is provided by artificial intelligence. We believe in technology but understand that even the smartest robots can sometimes get speech recognition wrong.

Today, we have another interview in action from the conferences that just happened down here in Miami and Orlando. My name is bill Russell. I'm a former CIO for a 16 hospital system and creator of this week health instead of channels, dedicated to keeping health it staff current and engaged. We want to thank our show sponsors who are investing in developing the next generation of health leaders, Gordian dynamics, Quill health tau site nuance, Canaan, medical, and current health.

Check them out at this week. health.com/today. Here we go. All right, here we are from 5 20 22, and we're here with, , Kennan medical hit. And Tim, what's your last name? Dawson. Tim. Larry Sitka. Okay. Your chief's, , your secure or strategy strategy officer and I'm a CTO CTO. There we go. , that was a slow start on my part.

So I apologize that I'm not, I'm supposed to be professional here. I'm looking forward to this conversation because every time I think of Canon medical, I think imaging and I think of cameras and I think of imaging and that is a part of your past. You guys, I mean, we're talking here and I'm learning a ton about where you're, where you're at and where you're going.

Larry, give us a little bit more on that. So Canon is, if you look at Canon as a visionary company, being able to bring all these different products, ancillary products together as one common platform that actively runs for all imaging services to be able to not only capture, but also persist the information.

So data persistence is really what DNAs are focused on. But then also the normalization side, right? Being able to take the data itself, the metadata, normalize it, make it accurate so that these AI algorithms that are going to use it can actively use it going forward. So Canon is really evolved from, I'll say, a modality company or just a product based company into an enterprise class solution.

So enterprise application, not just around imaging. So chief technology officer, what does that look like? Maybe what is, what, what are we talking to CEOs about this one? All right.

So you have to handle the whole gamut from obviously today. Cybersecurity is a huge piece, right? So we have to build and deploy our solutions in a way that can be secure and they can provide interoperability with other areas of the overall enterprise.

Another big piece is business intelligence, right? Getting the information out of the system and being able to. , bridge the traditional imaging, the information that we have out of the DICOM field, along with the HL seven and marry that data together, to be able to get real actionable intelligence and really improve the quality that you can get out of our, our base and our, our history of image analysis, and be able to build more on top of that with, , with access to other data.

I think of it as a data persistence platform, right? We have a slew of HL, seven data coming in. We have a slew of DICOM data coming in. We normalize it, make it readable, and then make it available for searching across the context between the two, you know, the two different protocols, which is quite unusual.

Usually we'll get a measurement. or you get a measurement of DICOM imaging. The other thing it allows us to do is not only just focus on just icon, but any content. So a PDF document, a word document, an Excel file, you know, a JPEG image, an MPEG video, et cetera, can be stored labs, analyze links

together. It almost sounds like a data management platform. Am I, am I missing it? Or

so that is so having come from the BNA construct with my prior company, that I, so one of the biggest differences that I saw, and one of the reasons that I came to the organization was primarily focused around data orchestration. So storing this stuff was pretty easy to do, and that's what I really focused on.

If you can orchestrate the data to make it, what I call it. Persistence and perception. So data persistence versus data perception, Canon is all about fidelity. It's all about perception. It's all about taking this pile of pixels and making it look to the best quality you can possibly make it from very simplistic and easy to use to diagnostic quality, to advanced visualization.

High-end 3d render. So they have that big, if you can put the two together, you have a platform. I don't know, I guess I would call it kick ass.

The, , so talk to me about security. I mean, a lot of I've been talking to CEOs over the last, , last 24 hours. We've done 14 interviews, about 10 of those have been CEOs of health systems and they will tell you that, , well, number one issue they have is labor, , either on the clinical side or on the side.

The second thing they will tell you is, , the biggest challenge they have cyber security. So we have those two challenges. , talk about, , give us an idea of the things you're talking about here, how they would dress either of those two challenges that, that the CIS are facing.

So we have, we take an approach that we, so we, we all read the NIST documents.

We answer probably three security documents a week. , different applications from coral applications to, you know, sensitive applications. And we took Tim and I took a step back about a year and a half ago, and we, we interpreted what we read, just using the basic common principles that already exist inside, but it's just a matter of implementation and configuration.

And being secure about what you do zero trust, right? Right. So we have, or we support our form of a zero trust architecture where not only do you protect from the outside, but we protect from the inside using TLS, using encryption on the disc encryption, in flight encryption, inside the application. So the only time it's unencrypted is actually in memory at the end of the patient.

I mean, the important thing really is there, there are two big pieces of cyber security that, that. Confidentiality integrity and availability free. Right. But the, a lot of times people focus on that confidentiality. They don't want to patient data being stolen and sold out on the black market, but availability is really the critical piece.

Right. You have to make sure that's what we saw with

these ransomware attacks. Exactly. Ransomware is the, is the number one thing that I'm planning around these days? Obviously we don't want the. To be stolen, but, but the critical thing is this, can we treat a patient that's coming in for a stroke? You know, if the system is down, that's not a good thing that, that doesn't yield to good outcomes.

Right? So at the end of the day, the critical piece is making sure that we use those, those items to lock the system down, to really protect and prevent these type of denial of service attacks or the type of ransomware thing that would, , that would really prevent your ability to treat patients, you know, bill one, one thing that.

If you could ask the CEOs that you speak with, ask them, do they have a PACS? They're going to say yes, absolutely. I'm going to say, do you ever have to pay to get your data out of that PACS? Have you ever paid for a migration? How's that really different? And you know what? It's probably more expensive to do a data migration than a ransom, to be honest with you.

Yeah. So I think we're already being held prostitution. And again, the constructor on what ONC is trying to bring forward, right. With unlocking the data when we have more time and can tell you horror stories about almost going to jail in a foreign country, because I was trying to access patient data.

It's that bad.

Yeah. Yeah. That does take us in a different direction. Yeah. The 21st century cures and all the things we're trying to do there, , Hit. So it sounds like you're a consulting organization. Are you you're? I mean, I know your enterprise enterprise software enterprise applications, but are you, are you, , coming in alongside these, , these health systems and helping them to really evaluate this and look at their security architecture, look at their, , their solutions around imaging and those kinds of things.

It really is part. You know, our existing customer base in the last, you know, three, four years going forward, it's a partnership, right? Because if you've done one EDI solution, you've done one, you know, because the people are different, right. That make up the solution. So every solution that we bring forward is, is worked on over at least the one year, almost time to bring these to market, you know, to interact with.

You know, do meeting after meeting, after meeting to do drawings, to do swim lane discussions, to make sure that we get an understanding of how that organization works. And if they don't know, we say, well, here's how some other organizations and we create what we call a storyboard, right? It says, you know, here's our journey of the solution going forward.

And we carry that, that storyboard, , interacting with the customer so that everything is back to. And then we turn it over to the implementation team. Post-sale.

So years ago we've made a shift from selling software to selling solutions, right? And the big difference is you have to have the services to go with it.

And these aren't just deployment services, helping you stand it up and then walking out the door, it really is configuring the workflows, configuring the automation. You don't want. You don't want to have things with the, you know, a lot of manual workarounds. You want to build as much automation as possible.

And so that's really where the services come into play. And then yes, obviously you can build a piece of software that's as secure as possible, but if you, , if you install it and then don't enable that security, , then there's, there's, there's no actual securities like building a house with really thick doors and you never lock it.

It's the same exact thing.

So what are you hearing at the conference? I mean, you guys are, this is a pretty good spot here. Huh?

I was actually quite impressed as walking in. , so. So I've been a member of time for a long time, , coming together with, with the other, with chimes, splitting out of hymns and coming together, , with health care too great vibe.

What I like to see is the incubation side of it, right? I'm a startup guy. So I, I love to see the fact that they're, they're nurturing these small startup organizations and that's where you get the great ideas. The technology and we need to do something different because the same old BS is just not working.

Now that's microseconds. Right. But back then when I left Ziegler and I went to fidelity, they gave me a C. And they gave me paper records. Right. And I walked out that's healthcare to them. So it's in the same state, right? It's following the same state. The problem is we, for some reason, we don't want to change, you know, change is hard.

The other thing is if health is really hard, it is really hard. Imagine if your bank account was kind of. That's healthcare serious. Yeah. You know, if you've ever, I'm simply amazed and x-ray gets read, you know, can you imagine a reading? , I get if it's a BR a brain study, but what if it comes across as a Brian's then these, these common things.

I had a colleague of mine because I sit on the him SIM committee for we're focused right now, just on body part and on, , anatomical region. And we wanted to go after. Several fields. We still look at and we, we could, because I said, have you ever just looked at those two fields? I brought them up and see the number of actual data parameters that you do.

A friend of mine did it as a search. There was 10,000 body parts and just six months there aren't 10,000 body parts. Right? So it's, there's so much garbage. You know, the first step in driving AI, you need accurate information to drive a clean data. We've heard that over and over again. And these interviews that's once that's our first step, I keep telling Tim or Tim, because I beat them up all the time.

I say, we have to do right data normalization. First thing you have to do when you get past it. And if we don't do that, you can have the best AI algorithm. It won't matter. So, yeah,

it's. And it is a true challenge that we hear AI everywhere. Right. And, and I'm now pushing people to define what they mean.

And what are you doing? What are you actually doing with the data? What does it look like? It's so sound like fancy algorithms. Others sound like Trulia.

So in my head, it's just, I'm a developer by trade. So it's, it's just additional information, you know? And then it's, it focuses on. And then various heuristic search algorithms.

You take those statistics, you feed them back into other applications. That's the machine learning side of it, but you can get algorithm drift. So we, , we also have the capability of a platform that allows you to. Those algorithms so that they don't drift. So in other words, as you use your application and it, it drifts out of line based off of a cohort.

So say an algorithm was used for just, and it's all it's patient population was, was European, you know, descent. It will, it will focus on just that it doesn't take that into account unless it's told. So when it does that it drifts. So if you come back and you have somebody, a patient. You get skewed results.

You have to make sure that you have the right data set that you built the algorithm on top of yes. In order to, in order to make the wise, just go way off from the mines. You have to be careful for that algorithm. Drift. And I think coming back to your question of what is AI, right? I mean, you're right.

That term is, is used to mean a lot of different things, right? And, and that's pretty common in our industry is we'll take a term and we'll use it to mean, and you know, everybody will, will put their stuff on it. It could be, , I've heard people talk about some of our algorithms that have been around the early advanced visualization days that we've literally just taken hand-coded algorithms that took years to code and get right.

We drop those in, that's not really AI, that's not machine learning, but it doesn't matter. It gets used as. If it can automatically diagnose something and, or automatically segment out and identify where, , you know, where stenosis is or identify a tumor. That's what really matters at the end of the day that you automate the workflows and that you streamline care.

That's what really makes the promise of AI is that machines are gonna be able to look at patterns or images or what they're looking for past. Right. And they're going to be able to look for patterns. I don't necessarily have to tell it what to look for. If it looks at enough of the patterns, it's going to say, I see this, I see this, I see this, right.

This is what I see. And then eventually we can tell it that's cancer. That's not cancer. That's whatever it is. And then they can go, oh, okay. And then it can look at the next batch of images and go, we think this is cancer. And it it's a platform that gets smarter as it goes. Right, right.

You have to take care of that.

That's absolutely right. But you have to take care about your input datasets and to make sure that you've got good. Ground-truth right. And, and that you're designing the training proper. I mean, we've all heard the story about the people that were taking visible light images of, of, of skin tumor. Right.

And they took, you know, hundreds of thousands of photos and, and every one of them had, had had a mold or a tumor or something like that on the skin with a ruler next to it, measuring it. Yeah. And then what they found out at the end of the training, that algorithm was that they had built a really wonderful ruler to detect.

And so, I mean, you have to make sure that you're doing the right thing. Right. And so, so it's not just as simple as throwing the images in. You have to have people that understand how to craft the data sets and how to build good ground-truth algorithms, how to, how to build in a test cohort, right? That, that isn't part of your training so that you can validate the results coming back out of the deal.

And you can rebase on it as, as Lori points. So all of those things are, are, are important. And you have to have that baseline of training and know what you're doing, , to, to make sure that you end up with a truly valuable clinical laboratory, the other, the other component that, so what, what makes Canon really desirable to me is that you have a three-legged stool, right?

You have imaging and imaging services, you have health it, but then you have this new up and coming piece. I call it the biology side of the business, where you have a genetic baselines. You have, , , what do you call those, , where they take the tube of blood liquid biopsy, the biopsy and being able to detect that way that's healthcare.

Right? Look at my genetics. Take my blood, spin all the good stuff out. What's leftover. Tell me what if it's bad, if it's, , you're, you're finding it at the molecular. You know, I, I get the pleasure of traveling all over the world. And this is exactly what a, , a Bangkok hospital does. One of the most amazing places I've ever been.

I go there to learn what they've done, just because they focus on healthcare. They don't focus on a payment model. They don't focus on cost, right. They focus on how do I stop the patient from becoming ill? That's the perspective that we need to take. Where is that Bangkok? Okay, it's called bomb and graft hospital, their health level seven hospital.

They're working on health globally, phenomenal team. And just, , I've been all over the world. I've met so many locations and so many places, and that team is probably the best I've ever worked with in my total career. So they're just phenomenal the way they lead it, you know, the way they're so focused on the wellness of the.

That's what I mean really fascinating. Yep. Well, I will have to look that up. I'll be honest. A couple of things I'm gonna have to, I'm gonna have to do a little research on drift and rebase lining and stuff. I that's, I mean, that's an interesting kind of, I could see how that could happen. , and I'm definitely gonna have to look up that hospital.

You know, the other construct is, is right now. There's a lot of. Negative impact of AI because it gets in the physician's way. You know? So I think thinking about the workflow needs to change in terms of how it's it's delivered right today, we deliver, we deliver the EMR over here, or we deliver the viewing solution impacts.

Here. We deliver image Sharon exchange and, and reporting here. And now you want to put a fourth window up to do AI reporting. It just won't work. Right. We got to keep their eyes on the pixels and let them interact with the AI solution while it's feeding outcomes. The other construct is we've focused on just one finding or two findings.

The contract, the concept of AI will never be acceptable as a single finding, just like blood work wasn't so blood work developed panels, right? Blood panels. You, when you get a blood panel, 50 tests right? In a blood panel, blood workout. And you pay for that. That's where AI will become a business profitable model so that your chest x-ray, instead of doing one or two findings, you get 150 findings right on your chest, whether it's a simple chest x-ray or a CT gentlemen.

Thank you. Thank you. Appreciate your time. Another great interview. I want to thank everybody who spent time with us at the conferences. It is phenomenal that you shared your wisdom and your experience with the community, and it is greatly appreciated. We also want to thank our channel sponsors who are investing in our mission to develop the next generation of health leaders, Gordian dynamics, Quill health tau site nuance, Canon medical, and current health.

Check them out at this week. health.com/today. Thanks for watching. That's all for now.