What Fascinates Dr. Anthony Chang of AIMed?

Bobby: On this episode, I talk to Dr. Anthony Chang. Dr. Chang has one of the most impressive professional ranges I've ever come across. I know him as the founder of AI Med, which you can find online@aimed.io, but he's also a veteran pediatric cardiologist working out at the Children's Hospital of Orange County, better known as CHOC.

He's also the Chief Innovation officer there. He has a couple of other professional roles, but it would take me another podcast to record them all. Just trust me. He's a super impressive guy. I honestly don't know how he juggles all of the things that he does, but he does a masterful job of it. As an example, on the day I met him, he had just flown in the day before from an AI med conference in China. Then the next morning he saw 18 patients. And then managed to carve out a bit of time to be a guest on my show. Simply impressive time management. The intersection of healthcare and AI is a super exciting space that I'm personally passionate about and something that Loka is doing much more work in.

It's gonna have a deep impact on the world. Have a listen, and I hope in some way more of you will get inspired to take a deeper look in the field and get involved. As a fun bonus. Stick around until the end and find out Dr. Chang's favorite James Bond movie. Enjoy it.

Alright, so today I'm down in Southern California in Orange County, specifically in the Children's Hospital of Orange County in the Medical Intelligence and Innovation Institute. With me, I have a very special guest that I'm super excited to share with you guys, and I'll allow him to introduce himself. Please go ahead.

Anthony: Thank you. My name is Anthony Chang. I'm a pediatric cardiologist and with a special interest in artificial intelligence.

Bobby: Terrific. Welcome to the show. Really excited to have you. I thought it might be interesting to start with, as I do with a lot of my guests, closer to the beginning. And so for you, I was just wondering, obviously you made this decision to become a doctor, but what made you choose pediatric cardiology as a specialty?

Anthony: I had thought about being an architect through my early high school years, and then because I was deemed a good student in biology, I was asked to go to the National Institute of Health Art Seminar series on Saturday mornings.

I begrudgingly went because I had already decided to be an architect and one of my good friends in my class, his name is Errol. You're out there somewhere, probably hopefully a doctor as well. He actually wanted to be a doctor, so I was going to hand my invitation to him. But my teacher, fortunately, my sake, insisted that I go.

It was a heart seminar series focused on coronary artery disease, which was semi-interesting to me. And then the third Saturday someone got up and spoke about congenital heart disease, and that one hour changed my life. The interesting part of that story is I actually had an opportunity to go back to NIH and give grant rounds some 25 years later.

And I say and I remember saying, usually the speaker gets up and says, I'm really glad to be here, or This means a lot to me to be here. I'm very flattered, but it was so emotional, I could barely say the first few words of my talk because I realized that 25 years ago I had been a student in one of those chairs in the auditorium.

Bobby: And there was just a really beautiful poetic bit of that story that you were telling me just at this lunch before we got to record this, where you were mentioning how even though you decided not to become an architect, there was a bit of a tie in with architecture to cardiology.

Anthony: I had always loved architecture and all the three-dimensional space and aspect of architecture and. My art history professor in college reminded me that maybe I was destined to be a pediatric cardiologist because it's really mother nature's architecture that I'm dealing with, especially the defects as in congenital heart disease and trying to fix it, right? She really had even more wisdom than I realized at the time.

Bobby: I thought that was beautiful. We're a Silicon Valley-based show and technology and innovation is always near and dear to my heart. But I think that what's interesting about you and why I'm always fascinating to talk to you is how that happens in so many different disciplines and even for a simple thing, when you and I were talking about in the field of cardiology, when you first got exposure to it as a teenager.

Just for people who don't know much about that field. Yeah. What were the over, from the 1970s and 1980s and 1990s, what were the major things that people would now take for granted in cardiology that were science fiction back then?

Anthony: Yeah. Back in the ‘70s, congenital heart disease was mainly diagnosed by angiography or just putting contrast in the heart and then injecting contrast and seeing where the contrast would go and, Now we take an echocardiogram or ultrasound of the heart for granted because it's so commonplace. Back then it was, echocardiogram had just started literally with very crude images and there was no such thing as an ultrasound image of the heart itself yet. Yeah. It was all so the linear demonstration of the heart anatomy.

But I was just fascinated by how many possible ways a heart can go wrong in mother nature's blueprint. And then how many ways you can try to correct that mistake by mother nature, which is not common. It's only 1% or less of children with congenital heart disease. And, but that 1% is a very special 1% to me.

Obviously children are not guilty of their heart disease. Adults can be, but not children. So that was my life passion and. I feel that I'm still passionate to this day. Some 50 years later.

Bobby: So could you tell us a little bit about how, this is a beautiful office space and really interesting work is going on here. Could you, for the folks that don't know, could you maybe give a little bit of a background of how you got involved with the Medical Intelligence and Innovation Institute, how that came about?

Anthony: Sure. So it's MI3 for short. It's a special tribute to MI5 and MI6 in London, Uhhuh because they deal with intelligence.

So my very generous donor, Michelle Lund, who is the granddaughter, Walt Disney. So it's very appropriate that we're dealing with innovation and intelligence since her grandfather was such a pioneer in those areas. So we do two things. One, we promote innovation in pediatric medicine, and the second thing we do is foster.

Application of artificial intelligence and medicine in general, not just pediatric medicine. Try to tackle both, and proud to say that three and a half years later, we're making big strides in both

Bobby: areas. It's super exciting work and deeply impactful to the patients for sure. One of the things that you and I have talked about is how you have a passion towards seeing that more and more doctors have a more holistic approach to their education and learning. And so try to get out of the silos of medical school and go to business school or the school of Computer Science and learn something. And that's something that you've been championing for a little while now, but I'm just curious, as you were coming up the ranks, was that something that just came naturally to you?

Did you just have an inherent interest in something outside of medical school from the get go? Or did that come later?

Anthony: Yeah, I think it was both. I think I had. Inherent curiosity about other things other than medicine. Not to say that pediatric cardiologist is easy, but at some point intellectually gets a little bit stagnant.

You wanna know more beyond what you read and what you experience. I'm hearing other people, but very early on I realized that a lot of my senior mentors are very wise in medicine, but oftentimes lack understanding in other key areas like business or computer science or public health and. I wanted to be more knowledgeable in those areas, and those are the areas I happen to pursue later in my career.

Bobby: Yeah. So walk us through what made you, you're an established senior leader in the field of pediatric cardiology. You've done it or have been doing it at that stage in multiple centers across the country. What was the genesis behind deciding you wanted to go back to school and try something completely new and follow a master's in biomedical data science and artificial intelligence.

Anthony: I've always been interested in statistics, partly because that was an area that my colleagues would not want to teach or lecture on in meetings and courses. So I took that on because I felt like while nobody wants to do something, maybe I ought to give it a chance. So out of necessity, I ended up teaching it and really liking it.

That went on for a couple of decades and then I got some education in public health which is, as there's a heavy emphasis on statistics in public health and when Watson, a supercomputer beat the human contestants on the show Jeopardy on February 14th, 2011. I remember that date. Because that was the night that I downloaded the application for the biomedical data science and artificial intelligence program at STA at Stanford, because my brother had already informed me that was a good program. He knew I was interested in data science, the epiphany that night was very big for me and was Valentine's Day. So it was a special alignment of the planets for me in terms of being a cardiologist, and it was Valentine's Day.

Here's Watson using aspects of artificial intelligence and being two very impressive human champions in the show. Jeopardy, with all its language nuances. And I always read about artificial intelligence. And I was someone that knew about Alan Turing before the movie, The Imitation Game, and it's here and it's real and it's gonna stay this time.

Because as we had two winters of artificial intelligence in the seventies and eighties. Yeah. I thought for me to have a deeper understanding of, all of this, I needed to go back to school and learn from some of the best people I know.

Bobby: So did you just leave the practice, take a sabbatical for a year or two years?

How did that, I'm just trying to figure, I'm just trying to understand like how that even happened.

Anthony: There's something good about getting up at four in the morning. You can get some extra things done that you can't normally. So I learned when I was pursuing my business degree that the one way to survive both clinical medicine full-time and education is to find time that you normally don't find, which is early in the morning.

I remember in those years I would get up at four in the morning, study for two hours and go to hospital and do rounds. And I learned from back then that it gives you the time that you need to study and do your assignments. And I decided to do the same. And I would get up at four in the morning, do my class assignments, and then find time over weekends to do all the things I needed to do and.

It was a long four year journey, but it was worth every minute of learning and into a realm that doctors normally don't go into, which is, data science and artificial intelligence.

Bobby: Yeah. So were you, this is amazing that you're able to do that. Did Stanford have an online presence for this or were you having to fly out to Palo Alto?

Anthony: It was online program, but to be realistic, you really need to be on campus. Once or twice a month do projects with your classmates or to take exams. So I would fly up there once or twice a month. I'm still old school, so I wanted to meet my professors. Even though there are a few classes where I could have done everything online, I would go up there and meet the professors, which Sure.

My professors thought it was interesting that here's a student flying up just to meet the professor. But I really enjoy meeting the professors as well as meeting my classmates and doing projects and was actually sad the day I graduated. That was all over.

Bobby: Yeah, it just sounds like an amazing experience all around. Tell us a little bit about how AI Med would come into fruition.

Anthony: After I graduated and then having this four year long epiphany about. How amazing data science and artificial intelligence would be in medicine if we were to incorporate this new resource into healthcare medicine.

And I'm not just talking about deep learning and medical image interpretation, but I'm also talking about just using algorithms to relieve the administrative burden in the hospital. I'm talking about getting doctors continuous CMOE  tools using artificial intelligence as they see patients. All these sort of day-to-day tasks can be really made much better with smart use of artificial intelligence.

So it was a very insightful four years, but it was also frustrating because I realized, That very little of what I was learning was actually in practice in medicine. So hopefully in the next decade or two we will be incorporating all of it. Yeah, I think that's

Bobby: The kind of genesis of my excitement about trying to record this podcast is I feel that there are a lot of people that would be super excited to get the exposure to this kind of application of artificial intelligence and machine learning in the field of medicine.

What exactly is the charter of AI Med?

**Anthony:**We're not formally a society yet but my vision would be that we create a domain in medicine and healthcare that would allow doctors, nurses, pharmacists, anyone that's in healthcare to be able to converse in this new language of data science in artificial intelligence and to find applications as if.

They're wearing a different lens now doing the professional jobs that they do. And I like to include patients in that too, because lots of my very sa data savvy parents have taught me how to do data visualization and how to get information from wearable technology. So I, it's basically everybody that's interested in this space can be part of the space.

And very exciting because I feel the numbers growing pretty quickly. We did our first meeting in this area about four or five years ago, we had a couple hundred now and have grown to about 600. And this year, we'll be doing three meetings in different international venues. We'll probably reach well over a thousand attendees for those three meetings.

We're also doing breakfast briefings around the world, about 15 to 20 some venues. And then we have our monthly meetings here at headquarters, so I'll try to promote the knowledge for artificial intelligence in healthcare medicine, but also hopefully inspire young people or not so young people who are interested to learn more and to be really a contributing force in this area.

Bobby: Let's talk about a couple of these examples and living, maybe use cases of the instantiation of artificial intelligence in healthcare and medicine. So let's just say from that moment when Watson beat the human contestants in jeopardy to 2011 to present, what would be like one or two that come to mind as far as things that you've seen in real time applications of solutions that involve artificial intelligence and in the field of medicine?

Anthony: I think there've been half a dozen very clear-cut solutions that are really improving both the knowledge as well as decreasing the workload of the clinicians. One of the best examples is the use of speech learning for computer vision, for medical image interpretation and combining that with natural language processing.

And some of the services are now available to help the radiologist interpret medical images, be it a chest x-ray or MRI or a CT, and hopefully someday soon have an echocardiogram. So clearly addresses what's creating a lot of burden for slinicians in terms of having an inordinate amount of images, most of them normal.

And how do we tease out the signal and the noise of normal to accelerate interpretation and with a higher degree of accuracy? So we often see papers now comparing clinicians, radiologists, or dermatologists or pathologists to computer vision and deep learning. But I always say, That's fine for entertainment purposes, but we ought to be really thinking long term about how to use the synergy between those two to improve our image interpretation.

So that's one area. I think another area that's exciting is the use of deep learning to discover knowledge in a huge database. And I think one good example is the MIMIC-III database, which deals with ICU data. From a single institution, but it's good enough to be able to have clinicians and data scientists work together to query questions and get answers right out of the database by programming for knowledge discovery.

So by looking at the data. So I think those are two pretty clear cut examples, and hopefully there'll be dozens more in the next decade or so.

Bobby: What do you think are in the nearer term, say in the next three to five, based on what you're seeing among your colleagues, what do you see as most likely to drop next in the next three to five, in terms of solutions that aren't quite there today, but probably will be soon?

Anthony: I think the maturing area is the two areas I've talked about. So I think in less than five years, medical image interpretation will be cheap and accurate. And democratized. So that you could be a primary care doctor in Kenya and have access to world class pediatric neuroradiology interpretation because it's just based on a very good image.

That's just amazing software. It's just amazing. It's so amazing that you think gee 10 years ago, no one would be thinking that this is even possible, just like no one was envisioning autonomous driving vehicle until now, it's commonplace in Silicon Valley. It takes a little getting used to.

But I think one amazing and exciting area of medicine. I always tell my son, interns that have the privilege of mentoring in the summer. Actually it's circular mentorship because they teach me a lot of cool words and concepts too. That’s the most exciting thing in 25 years in medicine that's coming up.

I'm just glad that I'm gonna be around for part of it. And they should be very privileged that they're gonna see all of it. So I think that's one area is the democratization of medical image interpretation of any kind. Could be a rash, could be a CT of a brain tumor, it could be an angiogram of a congenital heart defect.

I think within five years it'll be very commonplace to get world class interpretation. I think another area that would be more and more commonplace is ICU is pulling your data for. Smart decision real time in the ICU setting. We're doing some of that here at our hospital is to use real time early warning system for preventing a sudden cardiac arrest, but be far more exciting.

If we have 250 cardiac cuss in this country, I'll pulling their data to really give it more relevance and accuracy so that I see within five years as well. I think another exciting area is finally really putting meat into. The precision medicine arena, and instead of just talking these vague terms, really having a lot of data in different data to really realize the dream of precision medicine that President Obama had, which is really giving precise care, individualized care to all of our patients.

Instead of right now, giving the same medication at the same dose based on body surface area to these patients with. A myriad of diseases and we really ought to be making it a lot more sophisticated. So I see those three areas really developing very quickly in the next, within the next five years.

**Bobby:**So when you were  talking about image classification and you said echocardiograms will hopefully be there, what are the challenges today for why that isn't?

Anthony: I think echocardiograms being moving images are a little bit behind the static image modalities, but. As one of my deep learning friends told me, it's just more images. It's not really more complicated. So it's just the sheer volume of the image information. And in congenital heart disease, there are so many ways that a heart can go wrong during development and more than a few hundred different cardiac diseases.

So I think it's just gonna be a bigger challenge, but it'll be a surmountable challenge.

Bobby: Yeah. And also with probably the increase of Moore's law and processing power increasing as well, I think will help process the moving images as efficiently as we can do static ones today, I would think.

Anthony: Yes. And I think what we really need is the human resources to teach the computers and doctors willing to spend time with the data scientists. And I think, as I always say, Most of the AI in medicine projects really comes down to human beings working well with one another. So as ironic as that sounds, until machines can teach themselves or teach humans even more, they do teach humans now.

But but I think right now, humans teaching the machines are still worth a lot right now in the overall scheme of things.

Bobby: And so these are all, artificial intelligence and machine learning are very sort of data driven applications and involve that world. Have you spent any time or had any exposure or colleagues around the use of robotics and surgery and where that's going?

Anthony: Yes. I think, as robotic surgery was just like a lot of the other advanced technology was hyped and didn't really deliver the dividend that we thought it would. But I still think that they have a major role in the future just. Perhaps we were, as Bill Gates says over expecting in the short term, but we'll end up under appreciating the long term dividend.

I think that's one of the aspects of robotic surgery is that we got the short term down and it didn't deliver as much as we thought it would, but I think in the future it would be amazing in terms of being in symbiosis with human operators, not independently as a mechanical robot.

Bobby: That's probably still decades away from being reality.

Anthony: I think it's gonna be less actually. I think. I think the more positions, and that's why I really love spending time with young people. Cause you don't have to, as they say, to innovate. You have to unlearn what you know. It's true.It is. If they don't have to unlearn what they know, it's actually easier to convince them that. Some of these technologies are actually quite advanced and just waiting to be adopted. So I think as young doctors come forward and take on these challenges, I think they will be more accommodating of these advanced technologies as they, as we have been in the past.

Bobby: That makes sense. The younger generation will have more exposure and hopefully proliferate it in the market more. So you're coming off a whirlwind series of international trips and you were in London, you're most recently in China. Can you talk a little bit about, like two facets there, common things you're seeing among these regions and then different ones?

Anthony: Problem wise, I should say, I think if you were referring to artificial intelligence and medicine, healthcare, It's fascinating to me because I'm actually writing a paper on this. When you look at the three, just say the lack of argument that the two dominant AI superpowers are the US and China.

And if you add in UK and Canada to figure out the artificial intelligence in medicine situation, you have to understand first the healthcare landscape in that country. What's the prevalence of chronic diseases? What's the access like? What are the resources? And then you layer on top of that, the artificial intelligence landscape in that country.

Yeah. And then you come up with a sort of a tertiary domain of what AI and healthcare is like, and it's just amazingly different country to country. For instance, in China, they have access to data. They don't have access to doctors. So in a, in an interesting way, they actually have a more pressing need for data science and AI projects to meet the demand from a large population that's growing fast.

That makes sense. Whereas in this country, we're known for our deep learning through all the Silicon Valley titans that you're in the middle of. And we're known for our research and development and academic publications. But we're not very good in applying these directly into healthcare medicine other than medical imaging which we're known for.

So I think different countries have a different spin on all of this. The UK, for instance, is very good at academic accomplishments and artificial intelligence, but they don't necessarily have a large homogeneous population and access to the data really. Put those methodologies to use as much as they could right now.

So every country has its own set of strengths and weaknesses as well as problems that need to be solved. But it's just fascinating going around the world and hearing from everybody the different sort of situations in these places.

Bobby: And similarities too.

Anthony: Yes. Doctors all wanna do better for their patients. So universal that I think perhaps is the reason why I think. The AI and healthcare and medicine domain will continue to thrive as a universal one rather than in other businesses where there's true competition. I think doctors inherently want to have enough compassion and passion to help their patients, hopefully around the world that will enable us to work collaboratively in this new exciting domain.

Bobby: I think the more minds we can get at the table thinking about this, the better the outcomes, I think for the patients, for the market overall. So this has been great. I think we have a tremendous amount in common, but I didn't realize that we also had in common kinda a, a fanship around James Bond and Bond movies. So I thought a fun way to finish might be to just compare trivia and see if we have some common interests there but what is your favorite Bond movie?

Anthony: It will have to be Goldfinger.

Bobby: Did you have a favorite scene or element of that movie?

Anthony: Not from that movie, but there's so many scenes.

I think I just missed the Sean Connery suave persona as James Bond, and as I said, MI3 was a tribute to MI5 and 6, and it's intelligence in a different context, but it's still intelligence to improve overall good in the world. I have a special fondness for James Bond and all he was trying to do, and a man of principle at the end of the day, still really impactful as well.

Bobby: And so you're basically the James Bond in the world of AI and pediatric cardiology.

Anthony: My fantasy! But I do think, as we collect doctors with passion and education,in this area, it is nice to have a group identity as secret agents in medicine to improve intelligence. I often say it's much more about the meetings that we do, it's about creating an ecosystem and a sanctuary, I should say, for people to, to enjoy the space and help with the movement to change medicine. And make this a transformation rather than just an intellectual interest.

I think we like to go from evidence-based medicine that we've had in the last two or three decades, which as you can probably figure out, it's getting kind of data into what I call intelligence based medicine. Which is based on data science and increased knowledge. And, in 1990, it took 50 years for medical knowledge to double in 2020 as estimated to be less than 60.

There's not a clinician on the planet that will be able to keep up no matter how smart they think they are. So we need our machine colleagues to help us to be better doctors.

Bobby: Yeah. And so that's why I got fired up at the, about the space, why I reached out to you many months ago. And so for people like me that are.

In technology and interested about applying that technology, whether it's in robotics or artificial intelligence or other areas into medicine, what are some good places for them to go and start onboarding themselves on what's possible? AI med, I assume, is a good starting point.

Anthony: We spend a lot of time on the website. So it's mi-3.co.uk. And it has my ebook 3.0 version loaded for free. Has an annual academic magazine. That's fun to read. Yeah, we didn't want the traditional academic journal because even doctors don't read those anymore. And, but we didn't want it to be in an industry magazine either, because that's maybe not at the integrity level that we all like as physicians.

So I moved into the sweet spot between the two and we have substantive interviews and factoids and articles that are quoted and analyzed. So it's a fun magazine to go through. It is very colorful with lots of graphics, so it's not tedious to read like a traditional medical journal.

And then we have all the talks from previous years loaded for everyone to enjoy as well as the calendar of all the events are happening around the world.

Bobby: That's terrific. I would recommend the magazine. I did think it had, it was almost like if Wired in The Economist had a baby, it would look something like, we

Anthony: tried to have that impact. And I think, just get involved. I always say to the clinicians and the data scientists, they should just take each other out to lunch. Yeah. Perhaps coffee. Get to know each other's world. Yeah. There's a certain beauty in the other person's world that you may not be aware of.

Yeah. Just we could use all the help we can get to really make good use of this amazing technology.

Bobby: Yeah. I underscore that a hundred percent.  Dr. Chang, thank you so much for taking the time. Thank you very much for coming down to Southern California.