76: How New Technologies Transform Healthcare with Scott Brown of RevealixPublished May 4, 2021
Run time: 01:00:24
Every 20 seconds, someone loses a limb to diabetes. Scott Brown, Chief Technology Officer at Revealix, joins the show to share how his company is taking a digital approach to preserving limb health using artificial intelligence. Scott chats about how AI is transforming the healthcare space, the importance of testing for failure, the privacy concerns around AI, and the ease and difficulty of working with new technologies.
In this episode, you will learn:
- How Y2K spurred innovation
- How to prove safety and efficacy in healthcare technology
- The importance of testing for failure
- How thermal imaging and artificial intelligence are working collectively to prevent diabetic limb complications
- How AI begins with gathering diverse sets of data
- Why security and privacy need to be thought of upfront and not after the fact when designing products
- What AI, ML, AR, and VR are and how they work together to build novelty technologies
- The role of UX in adopting new technologies
This episode is brought to you by The Jed Mahonis Group, where we make sense of mobile app development with our non-technical approach to building custom mobile software solutions. Learn more at https://jmg.mn.
Recorded April 27, 2020 | Edited by Jordan Daoust | Produced by Jenny Karkowski
Scott Brown on LinkedIn
JMG Careers Page
Tim Bornholdt 0:00
Welcome to Constant Variables, a podcast where we take a non technical look at what it takes to build and grow digital products. I'm Tim Bornholdt. Let's get nerdy.
A quick note before we get into this week's episode. We at The Jed Mahonis Group have a lot of fun projects coming in the door, and as a result, we're looking to expand our team. We're looking to hire an Android developer. So we at JMG place an emphasis on hiring for fit as opposed to skills. Skills are something that can be taught and fostered through mentorship and experience. Fit, on the other hand, it's a lot harder to define. But we've outlined some of the traits we're looking for on our careers page. It's over at jmg.mn/careers. So whether you have 1 year of experience or 20 years of experience, it doesn't really matter. If you're interested in talking and or working with us, please reach out at firstname.lastname@example.org. We'll put that email address and a link to our careers page in the show notes as well.
Today we are chatting with Scott Brown, Chief Technology Officer at Revealix, a smart mobile imaging and remote patient monitoring solution, which uses thermal imaging and artificial intelligence to prevent diabetic limb complications. Scott has spent nearly 25 years in tech leadership positions at companies like Digital River, Medtronic and Zipnosis. He joins the show to chat about how AI is transforming the healthcare space, how medical companies make sure their tech doesn't have catastrophic failures, the privacy concerns around artificial intelligence, and a whole lot more. So without further ado, here is my interview with Scott Brown.
Scott, welcome to the show.
Scott Brown 1:55
Glad to be here.
Tim Bornholdt 1:56
Great. I'm glad you're here too. I I'd love for you to take a chance to introduce yourself to the audience here.
Scott Brown 2:01
Sure. So I've been in technology development for a little over 25 years. Began my career at Accenture, working on the Y2K issue, implementing ERP systems at large payers and government agencies. Then was part of the dot com boom working for Digital River and working in ecommerce in the very early days of ecommerce. Then spent 10 years at Medtronic building out their connected care system. At the time we called it Patient Management, which allowed doctors to remotely monitor their patients using a system called Carelink. So it was the first class three medical device on the market to be able to remotely monitor patients. From their went to XRS and helped transform that company from a hardware based company to a mobile based company. And then had my own healthcare startup for about three years, launched a pharmacogenomics company out of the Mayo Clinic. Went to Zipnosis and ran product engineering and security. CTO at player's health. And then now at Revealix as their CTO and working on commercializing the first platform to use thermal imaging and artificial intelligence to prevent diabetic limb complications.
Tim Bornholdt 3:29
That's so cool. When you said Y2K, I have to ask. So for a lot of people that listen to this show, some of them might not even remember the Y2K crisis. But in like in quick terms, what was it? What do you think of that whole, like, Y2K crisis? Like was it super overblown? Were we right to be having everyone freak out about that? What's your takeaway from your work on dealing with that?
Scott Brown 3:57
Yeah, you know, in the end, it was way overblown. So what the Y2K issue was, was that to save on computing power, the year was two digits. And so when it went from 99 to 2000, computers that only had two digits would think it was 1900. And they predicted that there were a lot of issues, there'd be issues with billing and all sorts of things, you know, pretty much predicting the apocalypse because of the limitations of their computers. So, a lot of the companies at the time, you know, this was the infancy of the internet. A lot of, you know, Fortune 500 companies were using paper based systems and so on. So they didn't have a lot of back office IT. So Y2K drove a lot of that innovation.
Tim Bornholdt 4:52
So companies like PeopleSoft and SAP and so on came up with ERP systems, which is enterprise resource planning systems, to do accounting and human resources and kind of a lot of the back office stuff. And so they would spend, you know, 10s of millions of dollars replacing their manual paper based systems with computers and, you know, software packages that would, you know, automate and manage a lot of these things. So it's funny to think back, you know, 20-some years ago, how far we've come. But you know, at the time, this is a huge problem. And, you know, the predictions are pretty apocalyptic, and so on, but in the end, pretty overblown. I remember New Year's Eve 1999, you know, everybody was kind of on standby at work. And we're, you know, watching the news and seeing, Okay, well, you know, how's it going on the other side of the world? And is it midnight, you know, around the world, watching the news. No, you know, things are still functioning. Yep, planes are still flying, things are still working, doesn't look like it's gonna be the end of the world and, you know, kind of passed and things work, you know, a few glitches here and there. But, you know, there were a lot of precautions taken in there, a lot of money spent to avoid issues.
And it's one of those things where you'd think time would be a construct that we've solved in computer science or just in the world. But it's like, if there's one thing I've learned as a developer over my 25 years, it's like, anytime I have to manipulate anything dealing with time, I stand on somebody else's shoulders, grab a framework, you throw it in there, and you let somebody else handle the time zones that are like off by 15 minutes or something, like just all those different regions. Time is crazy, and it's just so funny thinking that nobody had that foresight of developing beforehand that maybe we would need to have more than two years, or two digits to represent years. But I think we've gotten a little more sophisticated, and like you said, better at it over the last 20 years, I would imagine.
Scott Brown 7:06
Yeah, you know, and I think the temporary solution becomes the permanent solution, you know. At the time, you know, they did it for a reason. It was because they had limitations in computing. And so how do we save, you know, space? And so it's like, Well, we'll just use two digits for the year and someone else is going to, you know, solve this problem. We'll kick the can down the road. And so when 2000 came around, we're like, I guess this is when we need to solve the problem. Otherwise, there's going to be issues. So hopefully, when it becomes the year 10,000, we'll have solved it by then for the next time this comes out.
Tim Bornholdt 7:48
Scott Brown 7:49
Tim Bornholdt 7:50
Listening to you talk about your journey through your career so far, a lot of it is in the healthcare sector. What drew you to working in this space?
Scott Brown 7:59
Yeah, so I really developed a passion for healthcare when I worked at Medtronic. I've always loved technology. My degree's in aerospace engineering. So I love solving problems. I love technology. I love building and creating. And when I got to Medtronic, which, if you don't know, is a mission driven company. And they always talk about, you know, the importance of the patient, and putting the patient first. And, you know, it's a very patient centric company driven by the mission that Urbach and the founder wrote many years ago. And every Christmas or every holiday season, they do this employee program, and they'll have one patient come in from each business unit to talk about being restored to health, you know, whether their life was saved, or, they, you know, overcame an issue or, you know, whatever it was, and by the end of that program, there wasn't a dry eye in that entire auditorium. It's amazing. And every single person in the company can relate, you know, had a hand in helping at least one of those patients. And so to see that your work is going into a product or a service that's restoring people to health or saving lives and so on, I find very motivating. So here I have the opportunity to work in technology. I get to solve really challenging problems and do really cool stuff. And then to know that it's going towards helping people is even more motivating. When I graduated from school, I had the opportunity to work in defense, and I passed up that opportunity because I didn't want my efforts going into those types of programs and systems and so on. And knowing that I'm contributing to restoring people to health, you know, now preventing diabetic limb complications, I find that very satisfying, and very motivating.
Tim Bornholdt 10:03
And I haven't done a whole lot of work like, you know, dealing in systems that have like HIPAA concerns or even healthcare just as software that would deal with like pacemakers or things like that that are mission critical that you can't screw it up. Like, if you screw it up, that's on the flip side of having people that you can see you save their lives. On the flip side, if you make a mistake, you could seriously hurt other people. And I think a good corollary is like people working in NASA, where it's like, you know, we flew a rover to Mars, and obviously, there's nobody on there. But if we're talking about building systems to support humans traveling in outer space, they have to be like, you know, very safe, and very tested. Do you think about that often when you're doing your work of how not only how impactful it is, but like the the flip side of it, of what happens if something goes wrong?
Scott Brown 10:58
Oh, absolutely, absolutely. And I think that's one of the great things about working at Medtronic is that, you know, seeing those processes, and having a really good process in place to make sure that you're building things that are safe and effective. And so a lot of people don't understand the role of the FDA, but the role the FDA is tasked with is two things, safety and efficacy. So safety is you're not harming people. And efficacy is it does what it says it does.
And so the patient management system was a class three medical device. And so there are three classes of medical devices. So like a band aid would be class one, and then an implantable device would be class three, the most regulated. And so the FDA before they'll give it their stamp of approval has to make sure that it's safe and effective. So you have to prove that. And so the safety is, you know, we're going to implant this device, and it's going to be safe for the patient to use this device. So a pacemaker, it paces the patient and it, you know, addresses bradycardia, you know, a slow heart rate, the effectiveness piece, is that, right?
We make these claims, we say it does something, and it does that. So at Medtronic, you have requirements, or, you know, if you're regulated by the FDA, you have requirements, and then you test to those requirements. We say it does this, then we test it, and does it do those things? And then a lot of times you need to do a clinical study, and to prove that it does that. So you know, you'll want to recruit patients, you'll do a study, you'll prove that it works in a limited group. And then once you have that data, the FDA will review it. They'll give it their stamp of approval, and then you'll be able to release it into the market. And I think growing up in that, you know, environment, I know, for me personally, it really challenges me to think about like, Well, what are we doing? What are the claims that we're making? How is this going to be used? How are we going to test this? How we're going to prove that it does that?
And I think, you know, the FDA gets a gets a bad rap for being overbearing, and so on, but they've got their function. And, you know, consuming products and using products that the FDA approves, I know the rigor that companies have had to go through to get those products to market. And so I feel good using them. It's like the vaccine, you know, kind of seeing the process that that these manufacturers have taken to get to accelerate the use. Now, they still had to go through those hoops. They still had to do it. But they took a more compressed approach, but they still went through those to make sure that it was safe and effective.
Tim Bornholdt 13:56
Yeah, it's really nice too having a system like that in place and having those tests. It's like, we're not all going to be building pacemakers or things that require that level of scrutiny. But it's like knowing that there are processes and kind of fail safes in place that make sure that whatever you're building is actually vetted appropriately and tested. It's like, it comes back to having just some faith in our institutions. Because yeah, for as much gaf as the FDA might get from people, and you know, for good reason, no system is perfect. But, you know, at least there is a system, and then it's kind of up to, you know, entrepreneurs and innovators and different companies to really push on those boundaries and find ways to improve on the systems instead of just saying, This is all crap. Let's get rid of it.
Scott Brown 14:47
Right. Right. Yeah. And I mean, you really have to look at how is that product going to be used. So let's say you build a Mars Rover, right, and you send it out to Mars, and it takes months to get there, and then it gets there. And then it's going to live the rest of its life out on Mars. Well, what happens if a component fails? You know, what happens if, you know, dust gets in it? And you know, and the arm jams, you know, how do you plan for that? And in a lot of those systems, you have to have fault tolerance. You have to have redundancy, you know. You can't physically touch it once you send it on its way. It's on its own, and it stands alone. Once we implant that device in a patient, you know, you can ex-plant it, but you really can't touch the device. Maybe you can do software updates, and so on. But you need to think abou the failure modes. So there's FMEA, testing, failure modes, effects analysis, you know. What if this component fails? What is their fault tolerance? Is there another system that can take over? You know, how do we handle these different faults and errors and the different things that can go wrong? And, you know, it's very different than just a website. If somebody is using it and it crashes, you know, there's not much downside. But if that device is implanted in a patient, or you've got the Mars rover and the component fails, well, then what? Like, how you handle that? How do you recover from it? How do you move on? How do you continue to make sure that it's safe and effective, and it's able to perform its function?
So there are a lot of different things like in the medical device world that they test for. I remember, I wasn't involved in this, but one of the tests was, they had developed a nuclear powered implantable device. And if I remember the story right, I believe it was a leader in Russia, and they wanted to implant a device that would last essentially the rest of his life. And so the batteries that Medtronic was developing would last about seven years, and they wanted to develop something that would, you know, last a lot longer than that. They developed this nuclear powered device. And if I remember correctly, it had to survive being hit with like a bullet. And they'd use, like, a 357 Magnum or something like that to test it to make sure that the uranium or the radioactive material that they use in there didn't then leak out and didn't, you know, create a nuclear contamination or disaster. And so things like that, and you know, you've got these particles that are floating around, and what if it flips a bit, then what happens? How do you address that? How do you make sure that the data that's in there is the data that's supposed to be in there? How do you, you know, account for these things, and make sure that there's integrity, and there's fault tolerance and checks and rechecks, and, you know, things like that that you need to account for? So it becomes very complicated becomes there are different levels, you need to test to that risk, and then make sure it's, you know, again, safe and effective before you release it to the market.
And then once it's in the market, then you're monitoring the performance. Is it doing what we expect? You know, we predicted so many faults, we predicted this type of performance, is it performing to those levels? If not, how do we take action? Like, what do we need to do? And so having all those plans in place before you release it to market, monitoring the performance, making sure that there's a place for patients and providers to report issues and so on. So they take it very seriously?
Tim Bornholdt 18:45
Yeah, I would have loved to bet on the gun range at the day that they were firing bullets at a device. That had to have been a fun day at the office. Well, maybe not fun, but at least an interesting day.
Scott Brown 18:58
Tim Bornholdt 18:59
Let's talk about some of the cool technologies you're doing at Revealix. So you created the first commercially available mobile platform that prevents diabetic limb complications from imaging software. What does that mean? And how does that work?
Scott Brown 19:12
Sure, sure. So there's been a lot of research that's been done on looking at the plantar view of the foot, so the bottoms of the feet, doing thermal imaging on there to understand blood flow, and so on, to prevent diabetic limb complications. So there are tons of, you know, lots of research papers. So if you go out and you look at medical journals, and Elsevier and other places like that, a lot of this has been done within the lab. So if you wanted to do this, there's lots of equipment that you needed. You need to be able to isolate the foot and use these thermal imaging scanners and so on.
And so Revealix is the first time company to be able to commercialize this on a mobile device. And so what we do is we, our target customer, or our target patient is a diabetic patient. And diabetic patients if diabetes isn't controlled, they get something called peripheral neuropathy. And so that's kind of a loss of feeling in their feet. So the nerves will die. And so they can't feel. They don't have a sensation in the bottoms of their feet. So they would essentially, you could walk, and you and I would feel pain, but since they've lost those nerves, they can't feel pain. And so they could get a wound and just kind of walk a hole into their foot. So they take off their sock, they've got a bloody hole in their foot, because they weren't able to feel pain. So once they have a wound, it becomes very difficult to treat. Because if you can't feel it, and you don't feel pain, it's hard to, you know, to treat that because you and I feel pain. So it's like, Well, that hurts, I'm not going to do it. But if you don't feel pain, you keep walking, it becomes worse. And it becomes, you know, that is like an $85 billion issue within our healthcare system. You know, because that ultimately leads to infections and moons and amputations and so on, if left untreated.
So what Revealix does is used thermal imaging to look at the feet and see the heat patterns within the feet. And so the heat patterns will show blood flow, cool patterns will show lack of blood flow. We're able to then take that and use artificial intelligence to then prevent diabetic limb complications. So if the patient can't feel their feet, we can look essentially under the skin of the feet, see the thermal imaging, and if we start to see a hotspot, a provider can intervene and get the patient off their foot and prevent that wound. Because if it becomes a wound, like I said, it becomes very difficult to heal and to treat and can lead to further complication. So here we can intervene before it becomes an issue to prevent the discomfort and the challenges with the patient and save a lot of costs within health care.
Tim Bornholdt 22:25
With the actual like imaging that you use, are they, you might have said this, but I think I might have missed it, was it MRIs? Or x rays? I can remember what you said you used for actually like finding those, where the complications may lie.
Scott Brown 22:39
Yeah, so it would be just a thermal scam. So there are a number of companies that make, like, these thermal guns, and so on. We use a device called a FLIR. It plugs into an iOS device. And there are two cameras on it. One's for thermal imaging, one's a visual camera. We're able to through our app, snap a picture of the plantar view of the feet, so the bottoms of the feet. From there we're able to isolate the feet, draw boundaries around the feet, you know, identify that their feet. You know, feet or not feet, you know, it's kind of like the old hot dog or not a hot dog. So our artificial intelligence identifies that they are feet. We're able to tell the right versus left foot. We're able to tell the boundaries of it, the forefoot, the mid foot, the hind foot, understand where those thermal patterns are. We black out the background, so we completely isolate the feet. Part of that we do for HIPAA, part of it is just so that you focus on the feet. So if there's the patient's face, or there's anything in the background, all that's blacked out. So it's just the feet so you can really focus just on that. And then we run an analysis looking at the temperature deltas within the right foot and the left foot, and then we compare the two feet and we look at the temperature deltas between the two feet. So if they have naturally hot feet, and you're looking at kind of the gradients on each individual foot that will give you certain information for a diagnosis. But then if you compare the two feet, the points between the feet, that can really amplify hot spots and cold spots and really help a provider see where there might be an issue.
And so that's, you know, that's the big advantage of Revealix. It's no touch during a pandemic. You know, you don't have to worry about sterilization. It scales really well. It's in clinic. At home, you can easily snap a picture within a few seconds, our analysis pops up within the device so you can see the temperature gradients within the foot and between the feet, generate a report for the provider and for the patient.
Tim Bornholdt 25:08
So cool. I know, like one of the other companies that we work with uses FLIR technology as well. And I don't want this to be a commercial for FLIR necessarily, but it's like, they know what they're doing. And there's some really cool, like, applications that you can do with this stuff, like, specifically with artificial intelligence. Well, I guess, first of all, maybe I can ask you this. So based off of what I've experienced with AI, you've got to have like good training data in order to kind of make a good expert opinion, you know, or at least a good artificial opinion of what's going on. How do you go about gathering the kind of information that you need in order to, you know, be able to accurately predict if somebody is going to get a hole in their foot or not?
Scott Brown 25:53
Yeah, that's an excellent question. It all begins with the data. And so right now, what we do is, we have a hierarchy of models and at the highest level, it's, Are these feet or not? And that just identifying feet is, you know, more challenging than you would think. So, if you look at the feet of Americans, we typically wear shoes. Our feet are typically compressed. You can typically tell the difference between men's feet and women's feet, just because of the types of shoes that we use. In other parts of the world, they wear different shoes than we do and so their feet look different. So if you're looking at like Eastern Asia, they were sandals and walk around barefoot a lot. Their feet will be flatter, you'll have space between the toes, they won't be as compressed, and so on. And so what we've done is looked at feet, you know, essentially all around the globe, and you know, trying to make sure that we are, you know, doing a good job with unbiased data in looking at different feet, men, women, different skin colors, different regions around the world, so that we can identify different feet, different shapes, because some feet are very curvy, some feet are very boxy. You know, and so making sure that we've got a broad set of images to train our system on.
And so from there, being able to identify feet, okay, the right foot, the left foot, here are the boundaries, you know, here are the different sections of the feet, so that we can look, you know, being able to compare the heels and the mid foot, and so on. And so we've spent a lot of time, you know, working to gather these images and train the system and refine the models. And, you know, constantly training the system. So when it comes to AI, that's probably the biggest challenge is making sure you've got a good data set to work with.
Tim Bornholdt 28:00
And I would imagine, I think you said it really well, like, you have to go and get more than just a sample size of what's in your backyard, if you're really trying to impact the world and make something that's not, you know, accidentally racist or any other kind of, you know, biased in one way or the other. It's like, we want to take care of humans, and, and humans come in all different shapes and sizes. So I think that's really cool. Did you like, so collecting this data, were you like going around the world and getting people's feet? This is endlessly fascinating to me of like, getting this training data, but I think it's interesting to hear from the user's perspective, as well, like, how do you make sure that you get a reasonable sample size of all different kinds of feed?
Scott Brown 28:46
Yeah, I mean, that's the challenge. And I don't want to get too far into our methods, but we spent a lot of time and a lot of resources on, you know, and different approaches of being able to collect, you know, a vast array of feet, you know, the bottoms of feet, and making sure that we've got a good understanding of the feet, and what a foot looks like and different feet and, whether it's really curvy or bigger, you know, boxy and the toes are spread out or the toes are compressed. And then you get into amputations too well, what if somebody is missing a toe? Or what if their forefoot has been amputated? You know, what if there's only one foot in the picture, you know. Things like that and walking through all those scenarios and make sure we have enough data to be able to create representative models so that you know that the system performs and you know, kind of gets into the fault tolerance too. Well, what if we can't tell, then then what do we do? How do we handle those scenarios?
You know, but as we use the system more and more and we get more images of feet, we'll continue to improve the models, we'll continue to improve, you know, the analysis and so on. But, you know, ultimately we want to have a really good experience for the patient and the provider, make it easy to use. It's clinically valuable. You've got a great experience, it flows, it works, you know, it's safe, and it's effective. So we've spent a lot of time building our current models, and we plan to continue to improve them as we go forward.
Tim Bornholdt 30:34
I appreciate you sharing that. Like I said, I don't want to you to divulge any trade secrets or anything, but it's just, I think at a high level, it's again, there's all these buzzwords that get thrown out all the time of AI and machine learning and VR and just, I think it's interesting to illustrate if you actually want to use AI for something, how important it is to get a diverse set of data and to really think through. I mean, when you said feet, you know, I instantly looked at my own feet and thought, you know, 10 toes, got some heels, you know, I have a screw in my foot. Like, that's something that's a outlier. But like you said, having missing toes and all that. It's just really interesting. So I appreciate you going there and explaining that.
Another piece of this whole pie as well with AI with is privacy and being in healthcare, I mean, HIPAA is a big thing, right? It's not any small beans. And just not even being in healthcare, also AI in general, there's a lot of privacy concerns around how it's employed, and how it's deployed out into the world. What are your thoughts around just AI and privacy, all of that stuff in general?
Scott Brown 31:49
Yeah, you know, it's something that we take very seriously. And, you know, when people think, HIPAA, you know, they think, Oh, the data is encrypted at rest, and in transit. And it's a lot more than that. It's a lot more than then, just, you know, making sure that things are things are encrypted. You have to make sure that people who shouldn't have access to the data don't have access. You have to make sure that things are constantly being reviewed. You need to make sure that, you've got, you know, intrusion prevention, intrusion detection, you've got all these different things in place. And, you know, even with the way that we capture the visual images, making it so that we isolate just the feet. So inadvertently, there isn't in someone's face, or, you know, some identifier in the background. So if that report, you know, if that image gets shared, or, you know, whatever reason, it's just one less piece of PHI, to be shared. So, I think it's something that, you know, we take very seriously.
Throughout my experience, you know, HIPAA, it's something that needs to be thought of upfront. It's very hard to design security and privacy, you know, after the fact. And so, when companies think about security and privacy, or you're developing a new system, that's one of the things that needs to be designed in upfront, and thinking through that, How is it going to be used? How are we going to share this? Who needs access to it? Are there integrations and things like that? And so from the beginning of Revealix, you know, we had done that, and we've taken very intentional steps to make sure that we have thought about privacy and security, how things are, you know, how data is going to be shared, who has access to it and things like that.
And even when I was at Zipnosis, I led them through a high trust certification, which kind of goes beyond HIPAA. So you can't be certified in HIPAA, but you can be certified in high trust. And so we were audited, and we got our high trust certification. And that was a tremendous amount of work. But I think it was a differentiator for us in the marketplace, just showing our customers and our partners how seriously we took security.
Tim Bornholdt 34:14
It's interesting, because you have obviously, within the bounds of AI that you're working with, you have HIPAA as at least a tool and a model that you can operate within when you're coming up with how you're going to deal with privacy. But, you know, AI at large, you know, advertising companies don't have a HIPAA, you know, that they need to worry about. And there's all kinds of things you could do with this technology that could be like privacy offensive I guess. I don't know what a better phrase would be but you know you with any technology, you can use it for good or for nefarious reasons. And I like it the point that you made about having to think through this stuff upfront because if you don't, you kind of walk backwards into an industry and into a reality where you, you know, can look at what you've got, and maybe sell it because you're, you know, low on funds, or you need to make a quick buck. So you can make a, you know, privacy intrusive decision. It is something like if you're going to go down this path of somehow using AI to improve your customer's life, you don't want to inadvertently ruin their life, or, you know, that might be dramatic, but you don't want to build something that then can be turned around and used in a way that you, you know, maybe not wouldn't have foreseen.
Scott Brown 35:36
Right. Yeah, and I think, you know, the thing about HIPAA. So there are two rules, there's the security, and then there's the Privacy Rule. And, you know, HIPAA gets dinged a lot for, you know, being heavy process and so on. But, you know, the one thing is is that it's really forced that discussion within healthcare, you know. Who has access to the data? Who needs to see it? And protecting people's privacy, and making sure that the data is secure. And, you know, some of the breaches that have happened have been outside of healthcare, just ridiculous breaches that have happened that wouldn't have happened had they taken, you know, if had they made security a priority, had they thought about privacy and so on. Having to be HIPAA compliant, you need to think through these things, and you need to, you know, kind of check in and review and you've got your plan, and, you know, you've got your your project to be able to continue to improve your privacy and your security. And so I think a lot of other industries could improve their privacy and security practices, and hopefully, it doesn't take a breach, but that typically drives change within different industries. So, I do think that HIPAA has, you know, forced the discussion in more industries and really forced the discussion, and made it a priority. So I'm all for the companies that retain my data to take security and privacy seriously.
Tim Bornholdt 37:21
Yeah, I could not agree more. So we've been talking about AI a lot. Another acronym that is kind of lumped in with AI sometimes is ML, with machine learning. And then there's also AR and VR with augmented reality and virtual reality. I mean, there's all kinds of acronyms that get thrown around in our industry. And I think having an audience that's non technical by nature, I think it would be interesting to hear what your thoughts are on what those, the four identified of like, with AI and machine learning and with AR and VR, just what are those technologies? And how can they possibly work together to build, you know, novel technologies?
Scott Brown 37:59
Yeah, so first of all, you know, artificial intelligence, you know, I think there are companies out there who have an algorithm, so you program the system, and the system does what you tell it to do, you know, calling what that algorithm artificial intelligence, and in my opinion, that's an algorithm, that's not artificial intelligence. For artificial intelligence, you create models, like you train the system, you're not programming it, but you train it. So at Revealix, we've fed 1000s of images to train it. These are feet, this is a right foot, this is a left foot, being able to train it on what a foot looks like, so that it can detect that. The model is there to identify the feet and identify different patterns and so on.
Machine learning, you know, to me is kind of that continuous improvement, kind of continuous learning, in developing those models, you know. So artificial intelligence, you build the models, you use the models, and then machine learning would be able to kind of have that feedback loop to continuously, you know, build and improve those models.
Augmented reality would be, you know, we kind of do that within Revealix, because what we'll do is we'll overlay the thermal patterns on the visual of the foot. I always think of it kind of being more on a mobile device, but what we'll do is we'll take the visual image, take those heat patterns, the clinically significant heat patterns, and then overlay it on the foot, so that you can clearly see where there could be issues, you know, again, clinically significant patterns. And so augmented reality using kind of a visual, and then overlaying it with something that's augmented, something that's not real. So you've got the real world visual image and then you've got an overlay of something that's been augmented or you know, kind of a graphical representation.
Virtual reality, you create a completely virtual world so there's no live or real component to it. It's all been created. And so I think, you know, there are a lot of opportunities for both. So within, you know, Revealix, being able to create that augmented reality, you've got the foot and being able to see the clinically significant thermal images within that foot. You know, and I could see, you know, virtual reality being used as, like, you know, when you're doing a virtual visit, you know, and being in kind of a virtual world with the doctor and being able to interact. So it seems more like an office visit, or kind of a traditional, you know, visit with your physician.
And so, you know, I think I've always taken the approach of making sure you understand the problem, and that you're applying the right solution to that problem. And you know, and I always say, a problem well defined is a problem half solved. And so just because you hear these cool new things, these cool new buzzwords doesn't mean that it has an application for what you're trying to solve. And, you know, making sure that you truly understand the problem and truly understand the need, and then evaluating solutions to be able to solve that because, you know, in some cases, it could be AR, some cases, it can be VR. You know, one of the things that we explored at Zipnosis was NLP, or natural language processing. And so with natural language processing, being able to take a spoken word, and translate it to text, and then once it's in text, then you need natural language understanding. So now that you have the text, now you can take that and you can do something with it. It's the meaning behind those words that become valuable. So you natural language process the voice to text, and then it's the natural language understanding to put meaning behind it, to then be able to drive a diagnosis or a soap note or a,, chief complaint or things like that. And so we had developed this really cool proof of concept where a patient and a provider could have a dialogue and through that dialogue, you could generate clinical content that could help with the diagnosis and the soap note or the, you know, make it more clinically efficient. And it became, you know, beneficial if you're having a multilingual conversation, or, you know, you're a really busy provider trying to do a lot and, you know, or you're trying to get to a diagnosis where maybe the chief complaints, or maybe the symptoms were, you know, outside of the norm. So, you know, I think there are a lot of applications within healthcare for these technologies, but it's making sure that you understand the technology, and then you have a well defined problem. And then you're finding the right solution for that problem.
Tim Bornholdt 42:45
This might sound like a facetious question, but I genuinely mean it, how easy is it to use these technologies? And the reason I ask is because something like natural language processing, well, maybe not Siri, because Siri is not the best at this. But there's plenty of examples out there, like commercial grade that are consumer grade that do a pretty good job of doing it. Like Apple and Google have these baked into the operating system. And I think consumers have this expectation or this, you know, perception that, Oh, you know, it's easy for Google and Apple to throw it in there. So it must be easy for any Joe developer to come in and implement these technologies. Is that the case with like, you know, specifically, we can talk about AI, but any of the things we've just talked about? Are these technologies that you see as they're becoming easier as a developer to incorporate or is it still, like, you need to really dust off your differential equations and remember how to matrix multiply and things like that?
Scott Brown 43:47
I mean, it depends on the technology. But I think, you know, Google and AWS have come up with new amazing tools to work ith AI and machine learning and analytics and things like that. It's just, you know, the capabilities on their platform are, you know, stunning compared to 10 years ago. So, you know, building these models, and having to implement AI, I mean, that was some pretty heavy lifting, you know, 10 plus years ago. And so now, within, you know, AWS, you've got Sage maker, and you've got a lot of these tools, and TensorFlow and all sorts, you know. Facebook has their own with the tech tron and, and, you know, pie torch and whatnot, and those didn't exist too long ago. And so there are lots of tools that are coming out to be able to bring these capabilities to the mass market, you know, so there's still some things that are harder to implement, but I guess it depends on what you're doing. You look at, you know, Apple, and what they're doing with AR kit and being able to incorporate augmented reality into your app. I mean, that's crazy powerful, the capabilities that you have now through Apple. So, yeah, I'm always impressed with what these companies are coming out and the tools that they're giving developers and what people do with them, you know. I know a lot of this stuff, you know, at Revealix we're looking at implementing, and we're looking at doing and we're keeping a close eye on different trends.
And even when we took the approach for AI within our platform, it was very, we took a very different approach than what a lot of the technical papers said and a very different way than they had done it in the lab. And that was because, you know, through our analysis, these tools that were coming out in the models and the capabilities that were coming out, we got better results than what they had done in the lab, and the tools that they'd use in the lab and the approaches and the models that the had built. And so, you know, I think that's, you know, one of the challenges of technologies, you know, being able to keep your finger on the pulse, and, you know, keeping up to date with the cool new things, and, you know, is it beneficial. So, had we not gotten the results, you know, we probably would have done it the way that they'd done it in the papers. But, you know, that's why we did the analysis. And that's why we did the testing. And, you know, so like I said, I think we've got a phenomenal solution. And we continue to keep our finger on the pulse and track trends within the industry.
Tim Bornholdt 46:48
Yeah, as long as I've been doing software, it's like, it seems like I will try to do something, like, there'll be some task that I want to try to do, and it's next to impossible to do. And then I look back again, you know, five years later, and all of a sudden, it's like, it's already an AWS plugin, or say something. It is crazy how quickly like even listening to this podcast, if you listen to this podcast in five years from now, I'm sure the technologies for all the things that we've been talking about will be even easier to implement. And I don't mean to say that it's going to be just as simple as, you know, touch one button and go, but I think it's still at the end of the day, most of it, what it comes down to is you've got this toolbox of technologies that you can pull from, and those tools just keep getting better and better and better. And then it's really up to you and your imagination, and whatever your business goals are to, you know, plug in the right solution for the right time. And, you know, technology will keep evolving. So don't be afraid to re-review what you're doing every so often to make sure that what you are doing is still considered to be the top of the line.
Scott Brown 47:56
No, you're absolutely right. I mean, you even look, if you want to see, like a physical representation that you look within med tech, you know, pacemakers used to be, you know, almost the size of your fist. And so now they've got the pacemakers that are, you know, like, they'd be about the size of like a dime, and so they get screwed into your, they go into your heart. And so rather than having a lead in this device, and, they essentially, you know, like I said, maybe a few grains a corn, and then you kind of screw it into the heart, and then it's there for, you know, like, seven years, or whatever the battery life is, but that's just mind boggling to me, like the power that's going into these devices and the batteries and so on, you know. They've got, I think it's like the the size of a, it used to be the size of a pack of gum, it's called the Reveal. And it was a monitoring device that would get implanted. And so now that I think it's just a little bit bigger than a grain of rice, pretty small, fit under the skin. And, you know, it's mind boggling the capabilities that we've got with, you know, AR and VR and an AI and all that.
So, yeah, it's, you know, I think it's the companies have to continue to look forward and have to carve out that time to be able to look at these trends. You know, I hear a lot about, you know, blockchain, you know, where does that fit in? Do you know enough about it to be able to implement it? Are you keeping track of these trends and so on, because you can very quickly get behind the technology curve and become irrelevant without kind of keeping up and deprecating some of these old technologies. And, you know, I think a lot of large companies are going through technology transformation, because of that, right? So they need to go through this big revolution, when you don't, you're not kind of constantly caring and feeding for the system and maintaining and keeping up with it. So, yeah, so I definitely think that it's advantageous for organizations to, you know, constantly evaluate and keep up with trends. So because like I said, you know, the cost of implementing some of these things, you know, being an AWS customer they make, and Microsoft and Google all have phenomenal capabilities out there, you know, serverless, and things like that have their place and give you a lot of flexibility.
Tim Bornholdt 50:39
Yeah, it's really, sometimes overwhelming as a developer to stay up to date with all this stuff. But that's why it's, like you said, periodically going back and just looking at what you've got, and what can be improved and what you can clean up, it's super important.
The last question I wanted to ask you was, you alluded to Zipnosis a few times, and being a telemedicine platform. I mean, we're in a global pandemic, that if you were, if you couldn't have picked a better place to deploy something like that it would be during the middle of a global pandemic, when no one wants to go into a doctor's office. But I'm curious to hear, you know, it took a while for patients and providers to kind of embrace all of this stuff. It seemed like a lot of people were just kind of forced into it. So I'm curious to hear your thoughts, specifically around like the user interface and the user experience of these technologies in kind of getting users to actually want to use these virtual care platforms.
Scott Brown 51:38
Yeah, I mean, I think so Zipnosis, you know, started a long time ago. So tele health's been around a long time, since I think, since the 50s. And, you know, it's kind of been this concept that's been out there. And they've proven it for a long time, you know, it was a phone visit. And then they did some video, and, you know, there wasn't really broad adoption, and Zipnosis was able to figure out really asynchronous care, where, you know, you don't have this two way, the synchronous interaction, like you normally do in a doctor's office, but being able to capture your symptoms, your chief complaints, that asynchronously goes to a provider, who gets notified, they review it, make a diagnosis, provide treatment, and so on.
Now, you know, and it was really natural, for, like, video providers, you know, Amwell, and Teladoc and MD live and so on to gain adoption, because it was really natural for people to go from, you know, a face to face visit in the doctor's office to a video visit with their doctor. You know, the challenge with that is it's essentially the same amount of time, with overhead doing the video visit. And so like, as far as the adoption, you know, there's this paradigm that, you know, I want care, I go into my doctor's office. You know, kind of the telehealth and so on. It was nice maybe if I'm in a bind, I'll use it. But the pandemic really forced that and I think this pandemic forced a lot of things like working from home. I know offices where everybody had to be in the office, but then they were forced to work from home. And now they realize that they can function working from home and zoom meetings, and so on. And I think the pandemic really forced this as well. So now the expectation is that to provide care, well, I expect there's going to be a virtual visitor, or I expect there to be virtual care as part of this. And so even with, you know, Revealix, you know, being able to build in, you know, virtual care component to our platform. Well, how do I manage my patients when they're not here? How do I do a virtual visit? And things like that. So I think what this pandemic has done is really accelerated that and, you know, forced, challenged a lot of these paradigms. So I think, you know, people, it's changed the way that we live, it's changed the way that we work and the way that we interact and so on. And so I think that you know, going forward, if you want to provide care, if you want to be a serious player within healthcare, then you need to think about the virtual components. You need to think about, you know, kind of doing things just as effective but more efficiently. You need to think about, how do we, you know, interact with our patients if they're not in the clinic and so on. So it's changed a lot, you know, and not all for the worse. So I think it's really going to, you know, healthcare will be, part of it will be better for this right. Even look at the way that we approved vaccines, right. We have our first mRNA vaccines and approval process and virtual care and so on. And I think that the society is going to benefit from these things, even though there's been a huge downside to the pandemic, there will be some benefits.
Tim Bornholdt 55:28
Yeah, it's looking back at most of the major technological leaps we've made in the last, you know, just say 150 years, but probably in throughout human existence, there's always some existential threat to the way that we live. And it comes up to some humans with some ingenuity to pop in and come up with a way to do things that we couldn't normally do before. I mean, thinking of like, the whole space race thing during the Cold War, and all of that, like, that's the first thing I thought of is like, Well, why would we ever need to go to the moon? But then it became a thing where all of a sudden, like, we had to prove that we could get to the moon just to, you know, say we could and to prove our, whatever, dominance or whatever you want to say. But it's something that, I think the corollary there is to how this pandemic made it so we couldn't do things like, you know, not a lot of people were doing video chats as much as they do now. And just look at the last year of every single conference, every single, like tech conference I've been to, it has been on a different platform, because there's 1000 new entrants into the way that we do virtual congregations. And yeah, like you said, you don't want to look at a global pandemic and see all the suffering that's been going around and, you know, be happy that that's happening. But you can at least, I think, we also want to try to find the positives in situations. And I think, being able to just see how you can take a bad situation and apply some technology to it and accelerate areas where previously, it was just kind of laughed at.
I kind of in the 90s, I got to go to Universal Studios. And they had this, it might have been Disney World, one of the two. But they used to have this exhibit where you could like video chat with somebody virtually and you'd stand in a room. And then like, you know, your grandma was standing across the room, and you could see her on a screen and chat in real time. And it's like, Oh my god, that's so cool. But then you kind of think about it, you're like, I don't really, the technology wasn't very great. It took a long time for it to adopt and all that stuff. And now like that's our primary mode of communication with our parents. Like with my kids, it's like, they'll be like, Hey, I want to talk to grandma. And just hit one button. And there she is in HD like, right there able to interact. And a lot of those innovations are directly related to times where we as a species are going through a necessary and painful transition. But yeah, on the outside of it, we end up with some actually pretty useful technology that, if not, you know, could fully replace what we were doing before it at least acts as a compliment. And you can then find ways to take the things that were really good about in person healthcare, for example, and combine them about the things that could be great with like you were saying asynchronous healthcare, like, How great would that be to just like, type in what you're feeling. And then a doctor will get back to you in a few hours. You don't have to sit and wait for an appointment or anything like that. There's certainly a lot of benefits to, you know, crappy situations.
Scott Brown 58:39
Right. Yeah. Like Ben Franklin said, In the face of adversity, comes opportunity.
Tim Bornholdt 58:44
Yeah, that's a great quote. And that's a great way to end on this. Scott, I really appreciate you taking the time to chat with me today. Where can my listeners get in touch with you and learn more about what you're doing at Revealix?
Scott Brown 58:55
Yeah, Revealix.com. We've got a website that kind of covers what we're doing, on LinkedIn as well. And, yeah, we'd love to connect with people and talk more about it. We're doing really cool stuff. And love talking technology and healthcare. So thanks for having me.
Tim Bornholdt 59:14
Yeah, it was a pleasure to have you. Thanks for joining us today.
Scott Brown 59:16
Yeah, my pleasure.
Tim Bornholdt 59:19
Thanks to Scott Brown for joining me on the podcast today. You can learn more about the work Revealix is doing at Revealix.com.
Show notes for this episode can be found at constantvariables.co. You can get in touch with us by emailing Hello@constantvariables.co. I'm @TimBornholdt on Twitter and the show is @CV_podcast. Today's episode was produced by Jenny Karkowski and edited by the youthful Jordan Daoust. If you have a minute quick before you leave, we'd love it if you left us a review on the apple podcast app. It doesn't take much time at all and it seriously does help new people find our show. Just head to constantvariables.co/review and we'll link you right there.
This episode was brought to you by The Jed Mahonis Group. If you're looking for a technical team who can help make sense of mobile software development, give us a shout at JMG.mn.