January 19, 2024: John Halamka, president of the Mayo Clinic, delves into the evolving role of technology in healthcare, particularly focusing on the integration of AI and the transformation of the CIO role. How is AI shaping the future of healthcare, from reducing administrative burdens for doctors to enhancing patient care through predictive algorithms? They explore the challenges and opportunities presented by generative AI, its impact on healthcare professionals, and the ethical considerations that accompany its adoption. They discuss the transition of healthcare from reactive to proactive, with a special emphasis on genomics and molecular medicine. How will these advancements redefine personalized medicine, and what does this mean for the future of patient care? This episode provides a thought-provoking look into the intersection of technology, medicine, and leadership within the healthcare sector.
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 on This Week Health.
(Intro) If you can reduce 10 to 20 percent of a doctor's administrative time so they can see more patients and deliver more care, hospitals are going to win, doctors are going to win, and the startups that enable that are going to win. It's going to require a partnership.
📍 Thanks for joining us on this keynote episode, a this week health conference show. My name is Bill Russell. I'm a former CIO for a 16 hospital system and creator of this week Health, A set of channels dedicated to keeping health IT staff current and engaged. For five years, we've been making podcasts that amplify great thinking to propel healthcare forward. Special thanks to our keynote show. CDW, Rubrik, Sectra and Trellix for choosing to invest in our mission to develop the next generation of health leaders. Now onto our show.
(Main) All right, here we are for another keynote episode, and I am excited to be joined by Dr. John Halamka, President of the Mayo Clinic Platform. John, Welcome back to the show. Hey,
Ed, thanks so much. And it's going to be a wonderful year. We learned a lot in the last year. I really look forward
to what is ahead. All right let's start there. It's always interesting to interview you. You do a lot of interviews, you do a lot of speaking and whatnot.
You've talked about the role of AI in healthcare, predictive generative AI limitations. Yeah you talked about ethical considerations for ai. Actually, I'm looking at my list and let's see, about 10 of the 12 things have AI in 'em. So I guess people are tapping into you to talk about ai.
I'm gonna rewind to November of 2022 and you posted This is our act of courage to create a healthcare model that turns data into wisdom that benefits patients everywhere. The revolution we're leading calls for a new approach. That radically transforms patient care while addressing the countless inequities built into the current system.
We are moving from reactive healthcare ecosystem to one that is more preemptive. We are shifting from symptoms based medicine to one focused on molecular medicine, which incorporates advances in genomics, proteomics, and emerging understanding. of the microbiome. Equally important, we are changing our focus from late stage to early detection, from depersonalized to personalized medicine, and from in person patient care to an ecosystem that enables patients to receive care anywhere.
And I'll close with this, at this time in history when the country is experiencing workforce shortages, supply chain breakdowns, and pressure on the bottom line. Embracing change is hard, but at the same time, we are at this rare moment in history when technology policy and urgency to change Converge to create a perfect storm of innovation.
And we must not allow this opportunity to go to waste. And my question to you is, this was November of 2022. How are we doing? So you
can imagine, all that's a journey. And that journey starts with data, because if you're going to develop these models, predictive and generative, and they're going to be appropriate for every audience, democratizing knowledge, you have to be able to curate the data of the world.
And so what have we done since November 22? Mayo, of course, did the de identification of all of its data, it's 10 million patients, and it was multimodal in nature, that is, it included structured, unstructured, it includes omics, it includes imaging, it includes telemetry and digital pathology. And the question, of course, is can you do that for not only Mayo, but other U.
S. based institutions, and then can you start doing it at country scale around the world? So that when you say, oh, I'm going to develop this algorithm that's going to help with early diagnosis of cancer, it isn't just U. S. What about Sub Saharan Africa? What about Northern India? What about Southeast Asia?
So what we've done over this last year and a half is travel the world and begun to sign up partners for what I'll call data reciprocity. It's a new notion of thinking. Instead of everyone hanging on to their own data, it's collaboration. at country scale. So any country can validate AI against any other country.
Anyone can take any innovation and locally tune it to their country. And so we started with Canada and Brazil and Israel, and those are in process now. But you'll see over the course of the next two quarters, we're bringing on Singapore, we're bringing on the Republic of Korea, Denmark, Sweden. And as we look to the countries of the world, it's interesting to do what we outlined in 2022.
You need to set a criteria, such as, is the country stable? Really hard to do a little, societal transformation if the government is changing every few months. Is there a regulatory framework that's going to support what you're after? Do they have a digital infrastructure? Internet connectivity, cellular connectivity, is there digital data?
So we actually looked at the world and found 21 countries, in addition to the ones we've already worked on, that meet those 12 criteria. And so you look at 2024, and the countries, I know this is a sort of lengthy answer, but it gives you a sense of what you need to do to meet the goal that is outlined there.
The countries are Nigeria, Ghana, Kenya, Rwanda, Uganda, Ethiopia, UAE, Qatar, Turkey, India, Bangladesh, Philippines, Malaysia, Thailand, Mexico, Chile, Colombia, Brazil, Dominican Republic. About 3 billion people. And these are countries that are ready for this kind of digital health transformation. And we've done a lot of homework in 2023 in preparing for visits to these countries, government, involvement in these countries and beginning to figure out how to deliver these novel solutions beyond the U.
S. and the Western industrialized European countries. The journey in 2024 is going to be intense
I want to go in a couple directions with that. I do want to talk about security a little bit as, because you're now talking about multi country, but where I want to start with this, is president of the Mayo Clinic Platform.
Is this a Mayo Clinic Platform initiative? because you do a lot of work with the on committees and rather in other aspects with the federal government and What drives this? What is driving this work for you?
Sure. Mayo is a remarkable institution with a long history of being a convener and a facilitator.
And the idea the Mayo Brothers had was they would travel the world, they would understand best practices, they would test those best practices at Mayo, and if they worked, they would disseminate them globally. And for example, the use of surgical gloss was an innovation that came to Mayo and then spread to the world from Mayo.
Although Platform is a catalyst, a convener, and a facilitator, as you'll see, that we'll try. to work in government, academia, and industry from that foundation of Mayo and try to make an impact at country scale and not say, Oh, this is Mayo's work. What we say is it's the patients that benefit no matter where they live.
Yeah. At the end of that article, I don't know if I have this quote in here anymore, but at the end of that article, you have the quote from the founder of Mayo. I can't believe I don't have it up right here, but but it was, essentially the best interest of the patient doing everything from the best interest of the patient.
I don't have the direct quote, but essentially that was one of the Mayo brothers who said that we do everything from the best interest of the patient, which is the I think very me and that, that call a hundred years ago is so important today because a A lot of these platforms that are being stood up are really being seen as a alternative business model.
We're going to generate revenue and those kinds of things. And it creates a different lens at which you approach those things. You're trying to monetize the data versus really unleash the data for the good of humanity. So how as we look at this, why would I go in this direction versus a Cosmos direction versus some of the other health systems have gotten together and put a data platform together, or are all these sort of working together?
I learned very early in my career to avoid the word or, and to prefer the word and. And so when we think about it, Cosmos absolutely has a purpose. And that is to empower research using a subset of clinical data, and to democratize access across multiple healthcare systems to researchers in healthcare systems.
And that's great. Truvetta or others, have their own approach and they have participants that have chosen to, again, contribute selected data elements. This is more of a centralized repository approach, de identify them, and then build an ecosystem. The platform approach is slightly different than both of those.
In that it is very decentralized and federated. And it gives governance and control of all data uses to those who created it. And again, every model is going to be a little different. Use the one that is going to meet your needs achieve your goals. What we're hoping is that if a country scale, countries are saying, yes.
We will participate in a global ecosystem of reciprocity with de identified data that we control and organizations can bring their innovations to us. We will never lose control of the data. We will never exfiltrate the data. We will never threaten privacy or security. But we can take an algorithm and say, Oh, it actually works on Nordic people.
Or take an algorithm and say, we deployed it in northern India and it was effective. And that's our notion here. It's a little different goals, a little different purpose.
identification is tricky. It's tricky to pull off effectively. And then it's tricky to ensure moving forward, right?
And so when you talk about the data not moving out, actually moving out of the model or having to be exfiltrated in any way. You're talking about creating this sort of clean room for operating against that data. How do you ensure that data is truly de identified?
Because we're talking about country scale here. There's gonna be a lot of angst around this. I would imagine as people talk about, they want to ensure this. So how do we ensure the de identification? And then how do we ensure that? Stuff doesn't leak in there over time. That seems to be one of the things that happens is without vigilance, without multiple checks along the way, eventually information gets in there that shouldn't get in there.
How do we make sure it's the identified? How do we make sure it? continues to be de identified.
Sure, and really good question, and I would tell you it's a process, not a project. Because, you do an initial DID, and that's based on the technologies of the day. And you say, oh, I will get external experts.
In our case, we use Dr. Bradley Malin, who is the world's expert on DID, re ID. He's from Vanderbilt. And he certified the de identified data of Mayo Clinic as 99. 6 percent de identified. And the fact that we keep it in a cloud container and we never exfiltrate it, we don't allow linkage, and we never expose patient row level data and that kind of thing.
He said that is sufficient for you to do AI validation and algorithm development. But here's an interesting question for you. Is the EKG, the heart tracing, with no name and medical record number, de identified? The answer is based on technology of today, probably. But Dr. Malin sent me a paper last evening.
Of how using new AI approaches, there is starting to be a question of whether the EKGs that each of us get may be like a fingerprint or a retina. That is, you could look, because what is an EKG? It's a vector collection of the electrical signals of your body. And that is a function of the 3D size of your body.
The shape of your heart, your conduction, right? All these other things. So you say in 2022, 2023, EKG's DID meant removing certain kinds of printed elements on them. But 2024 might mean, oh, you can only show 10 seconds of each lead, or you can only show two leads of the 12, and that kind of thing. I tell ya, you can't ever say you're dumb.
You have to evaluate your DID in terms of the threats, the risks, and the capabilities of the world at any given time. And that's what we do.
there's a recent ransomware attack. on Integris. one of the things that's different about the Integris attack is Integris refused to pay the ransom, and so they went directly to the patients.
And they set up this really slick customer service, hey, we found your record out on the internet, if you'd like us to take care of it for you, send us this in Bitcoin, we'll take care of it. Of course, it's we'll protect you from ourselves, because otherwise we will release it to, whatever.
But that's the direction this is There used to be I don't know, not as much concern about the reputational risk of losing data, but I imagine that was never the case at Mayo. You guys are very much at the center of this, but that has changed the game. These other models where the data is actually moving into a central repository, does that represent a significant risk of some kind to this kind of attack at, I mean you talk about nation scale, some of these databases are getting pretty large within the United States alone.
so our notion, there's this term of art in security called the surface area or the attack surface. And so as you say take some sort of very large database, Facebook, that would have a fairly large attack service where potentially a compromise could be a billion or 2 billion people.
What I've described is a very decentralized, smaller attack service approach where what you have is institutions participate in a federation, but their data is actually stored physically separately. And for example, Mayo happens to have its data in a de identified container that's fully encrypted inside Google Cloud.
Now, Mercy, our partner has all of their data de identified and encrypted in Azure. Totally separate, right? But the APIs are such that we can submit algorithms to Mercy for validation, or they can work with our tool set to develop a new algorithm, but we're actually never physically bringing the data together.
And so it's that approach of segmenting the data and ensuring that logical and physical separation prevents the kind of ransomware at, as you say, at country
scale. So talk to me a little about This feels like an interesting time in history. It's I went through, you might be a little older than me, but not much.
I went through the PC revolution. I went through the mobile phone revolution, went through Not the start of the Internet, but the Internet proliferation in the world. And those really had a profound impact on the last, 30 years or so. There's this quote that the future is here, it's just unevenly distributed.
And it feels to me like with genomics, with AI, with, all the things that we could talk about today with regard to health care. We're at that moment where it was early on, you had that first cell phone in your car and it was connected to your car and it was hooked up to your car.
And you're like how big of a deal is this going to be? And it turned out to be a pretty big deal. We're looking at these generative AI models, we're looking at machine learning AI models, we're looking at data stores that are massive that we can now tap into that are pretty exciting.
It feels to me like we're at the start of a new age. How do we Make sure that we progress responsibly and we progress with haste because there's a certain sense in which there's a lot to do with these tools that have been presented to us. So first, let me say,
William Gibson, of course, was right when he told us that the future is here, just not evenly distributed.
On average, and you've seen the literature, it takes 20 years for an innovation in healthcare to disseminate globally. 20 years. And the question is, how do we take that to 20 months? And so let me paint a picture for you of what we're trying to do. Okay, you curate the world's data, and it's not going to be perfect, and it's not going to include the whole world, but let's say we can get about 3 billion people into a federated, decentralized, locally controlled, de identified ecosystem.
Okay, so that means innovators around the world can start to develop algorithms and validate algorithms. How do you know that an algorithm is fit for purpose? How can you trust it? This is why in the JAMA paper we call for this notion of a nationwide network of AI assurance labs and the publication of all testing publicly so you understand for every product where it works and not.
So imagine this, you're in an electronic health record, anybody's electronic health record, could be anywhere in the world, and a guy named Bill shows up. Who's a 50 year old dude, and this 50 year old guy says I have a certain sign and symptom. You say, wow, I wonder what algorithm works for a 50 year old senior editor and journalist from the United States.
Go out to this registry and find what has been validated on 50 year old males that is likely to work in a certain disease state and then automatically that algorithm, that tool is run on your data and the result provided to the clinician. with very little effort, with little friction. That's the approach.
And so people may say, Oh, John, that's going to take us a decade. This week, the Coalition for Health AI is incorporating as a 501c6. And this week, there'll be nine Board of Directors members from government, academia, and industry appointed. And this week, we'll begin our process of outlining how the Assurance Labs of the country will be accredited.
And we'll announce JPMorgan, some of the work we're doing in creating a nationwide ready
to be much more rapid than ever before.
welcome to This Week Health, where every morning is an opportunity to transform your day with the power of health IT knowledge. Dive into our diverse podcasts on Spotify or Apple Music. Featuring shows like Today and Keynote, bringing you insights from the forefront of healthcare technology. But there's more.
Our daily Insight emails deliver the latest health IT news directly to your inbox, ensuring you're always one step ahead. And for those on the go moments, our weekly Clip Notes summarize key points from our content, making it easier to stay informed, engaged. and aware of what's going on in the industry.
And don't forget to visit our revamped website, especially the news section at This Week Health News. It's your curated hub for the most relevant news selected and summarized for your convenience. This Week Health is more than a platform. It's your daily partner in navigating the ever evolving world of health IT.
Subscribe, follow, and become a part of a community that's shaping the future of healthcare. This Week Health, where your journey into health IT excellence begins every day.
📍 📍 📍 I'm going to take you in a different direction because you're going to get to talk at many events and other things. You're going to get interviewed and you're going to talk a lot about these AI.
Aspects. You're going to talk about governance. You're going to talk about responsibility and all those things. And I would strongly encourage, there's just a great body of work that you've put out there around this. I'm going to do the opposite. I'm going to take you back to your CIO days. And I'm going to ask you to put yourself back in that role.
It was, You are known as really the geek doctor, the original geek doctor. You knew technology at a very deep level, and you understand medicine, and the business of medicine as well as the practice of medicine. And you really define what the role could look like, and a lot of physicians have followed in your footsteps.
Given what's going on, the challenges with regard to staffing, some budget pressures, and those kinds of things. One of the things that you're known for is having written your own EHR. As you're looking at these Emerging models. Let's just take generative AI. There's going to be advancements that are made by the big players.
There's going to be advancements that are made by garage startups. whAt would you be looking for from both of those, if you were in the role of the CIO right now? The big tech startups versus the garage startups at this point.
Sure, what a great question. So as you point out, as you look at 2024 and I talk to hospital CEOs every day, but I also talk to startups and It's a financially challenging time for everyone, right?
Hospital margins are negative, startup funding has dried up. You look at the runway of cash that a lot of these startups have, and it's short. What the answer is that we have to define the business problem. What is it that a CEO, a CIO wants? And then what is the role of an innovation to help them with a business problem like burnout, staffing shortages, access to expertise?
So Mayo, as you start to think about some of those, has said burden reduction through generative AI in documentation or administrative areas, certainly something to pursue in 2024. And, you're saying, sure, there are the big players, the Googles and the Microsofts and Epic working with Microsoft, but you're also seeing a huge number of startups creating creative products and platforms that will reduce the burden on our clinicians.
And so certainly, one would hope that these technologies are not technology for technology's sake. That they are specific tools that reduce burden, reduce cost, and enhance staffing. And it, I know it's hard. Because margins are tight, but if you can reduce 10 to 20 percent of a doctor's administrative time so they can see more patients and deliver more care, hospitals are going to win, doctors are going to win, and the startups that enable that are going to win.
It's going to require a partnership.
It's been an interesting number of startups that I've talked to recently that feel like they've found the Holy Grail, which is essentially what Nuance has been doing for years. But they're now doing it with generative AI. It's pretty interesting, creating that perfect soap note just with a natural language front end and that kind of thing.
And doing that in primary care is one thing. Doing it in some of the other specialties obviously is A lot harder, requires more learning and those kinds of things. But I found it interesting to me that that is one of the areas that oftentimes gets pointed at to say, look, if we can, if we could take this keyboard and make it a thing of the past, especially in the patient room.
And make it more intuitive, potentially put some computer vision in that room that sees something as opposed to the clinician actually having to type it in at the worst or even, say it that it can actually see it and understand it and do it. Those kinds of projects are interesting to me, but it feels to me like access to those tools is really going up. We used to not be able to roll this out to every clinician because it was too expensive. And now we seem to be making these more accessible and more affordable.
Yeah, I completely agree. So here's an interesting and maybe controversial question for you. Is there a need for a CIO going forward? And so let me answer it in the following way. What I have found, I was right in 96 when I became a CIO. I was the CISO, the Chief Digital Officer, the CMIO, and all the rest.
What I'm starting to see in organizations is that role of a traditional CIO of Infrastructure, applications, provision, keep lights on, respond to security concerns. A lot of it is being moved to the cloud or moved to providers where it's more now OpEx not CapEx. And the role of the CIO is being divided.
into multiple who are focusing on different areas of this. And so what that means is that, it used to be 18 months to do a procurement. Now you can turn on a function in an afternoon. iT's a cloud provided function. It may very well be part of your EHR. Or it may be something that is adjunctive to your EHR, a smart on FHIR app.
And you can roll this stuff out with a lot less cost, and a lot less lead time. And so I think it's a different kind of thinking. And the role of the CIO, as we traditionally know it, has probably come to an end.
Has come to an end. Wow, that's bold. had you on an earlier episode saying the role has changed totally.
And I think it has changed totally. Come to an end. It's interesting to me because where are you going to place the responsibility uptime, right? So even in the cloud, there's a security mechanism, there's a availability mechanism. There's all those things that need to happen.
And in terms of running those projects and making sure that we're adding the new capabilities. There's a responsibility there as well and as I look at it, what's happening is, I agree with you, the CISO takes part of that and the CTO is taking part of that. And the, a lot of the CIOs are picking up other consonants, right?
CDOC, whatever, CIDO, whatever innovation di digital. And so they're getting much more involved in the the day-to-day operations. There's a certain school of thought that says, look, we need somebody at the table who really understands technology, who is responsible for making sure that all those things happen.
Therefore, yeah, there is a CIO. Now, the CIO is doing a lot of different things than they did before, but at the end of the day we still need somebody to look at and say, your responsibility is to know that, Microsoft is rolling this thing out and you're to bring it to the organization so that we can understand it and incorporate it.
That's how I'm seeing it right now, but I'm curious when you say that the role has really ended where does that responsibility lie at that point?
So what I'm seeing is a number of organizations have taken what was traditionally the IT organization. And now labeled a lot of that as a procurement that is done to the cloud.
And they create a chief digital officer who is responsible for customer experience. And as you say, there still needs to be an orchestra leader of this. Because there's going to be a CISO, and there's going to be a CTO. And, as you suggest, the consonants continue to evolve. But, maybe as we look to 2024 Where the consonant is going is the chief digital officer as the orchestrator, and that's more about knitting together these services for a very positive, reliable customer experience.
the perspective of the patient, let's talk about the patient for a minute. What can the patient expect? We talked a little bit about research here. and research at a global scale or a nation scale is very interesting. But, from a patient perspective, I'm now wearing a blood glucose monitor on a daily basis.
And I, pre diabetic, I'm not diabetic, but, I guess pre diabetic is diabetic, but regardless. it's really been revolutionary for me to have that insight. I eat something, I look at it, it's not instantaneous, but it gives me that feedback pretty quickly. Are we moving to a different kind of relationship with a person and their health?
And the person in their healthcare system.
And the answer is, I hope so. And so let me just paint it in terms of my own health experience. So as I was the second human sequenced in the Personal Genome Project. And so my genome's been public for the last 15 years or so. And if you go to personal genomes.org patient two, you can actually look at my genome and you can see all the risks of disease I have.
Can we clone you? I did contribute stem cells and so my, I do have stem cells around the planet. It's a little creepy, I weigh 170, but there's 185 pounds of me on the earth because my cells are growing in labs, but we digress. And so when looks at that, you say, Oh, I've now sequenced the human.
I understand risks of disease. And, oh, John, this is all true, right? Carries BRCA1. If John has a daughter, we better sequence the daughter to make sure this daughter does not carry BRCA1 and have high risk of breast or ovarian cancer. In the case of my daughter, she doesn't carry BRCA1. All good.
Point to be made is that this was a proactive, predictive, wellness focused look at the disease I carry or the disease my family, my descendants might carry. I have a high risk of prostate cancer. Now, all men have a risk of prostate cancer, but I have a particularly high risk. My primary caregiver stopped doing cardiology testing on me because I have no cardiac risks and I'm a vegan and my blood glucose is 70.
thEre's all these things that say, oh, that's not a disease state I'm probably going to have. Oh, but the cancer thing. Yeah, we got to look at that. So my wellness checks every year include a PSA and full physical exam. Imaging as might be required. So it's focusing on keeping me healthy, not responding to disease at a late stage and predicting what I have and avoiding it, and that's going to be lower cost, it's going to result in a higher quality of life, and one hopes that becomes a more standard of care as more of these algorithms that we're talking about, help you understand your care journey and the potholes you want
As we have these discussions, and you're in some of these discussions at a national level and whatnot, I'm really comfortable with the algorithms that are looking at my blood glucose and giving me you know, feedback. They're not necessarily giving me clinical do this, take this drug, but they're essentially guiding me and coaching me.
And I think patients are more open to this. I'm wondering if clinicians, how clinicians are feeling about algorithms starting to step in and interact directly with patients and guide them in some sense.
There's a wonderful journalist at STAT named Casey Ross. And about a year and a half ago, he wrote an article, AI in healthcare has a credibility problem.
And so this is a problem that if a clinician has no transparency, no understanding of how a model was created, was it two people in a garage based on a hundred patients? Or was it a team of data scientists working on 10 million? You don't know. So the reason clinicians have had some challenges is because, I would tell you that having a Algorithm you can trust.
Credible algorithm requires that transparency, consistency, and reliability. And if we can tell a clinician, oh, actually for Bill, this glucose monitoring algorithm was developed on a million patients like Bill. And the physician can say, oh, wow, AUC of 0. 92 seems to actually, and it's, oh, it's transparent reporting work really well for phenotype, like Bill.
Sounds good to me. And that's where we're going to start getting to in 2024.
I want to take you back to the CIO role. why did you write the EHR? Was there nothing else out there? Was there, what was the catalyst for writing it?
Sure. There were a couple of thoughts.
In the 90s, there was this emerging technology called the World Wide Web. You may have heard of it. It had almost 500 users. And the question I asked, this is my MIT thesis, by the way, and it wasn't plagiarized, I promise. My MIT thesis said, could you take this emerging thing called the web and move 100 percent of our medical workflow to it?
And this was the mid nineties. And the idea was I could bring any device anywhere to anything and with security. And this is my thesis outlining the security measures, be able to treat anyone anywhere. And when I showed this to clinicians, cause I actually wrote the prototype as part of my thesis. He said, Oh my God, if you took legacy systems that back then were client, server, or, monolithic mainframe or mini or green screen or whatever, and wrap them with a set back then it wasn't called, APIs.
It was more a service oriented architecture. Remember back to If you wrap them in a way that I can now do my work on any device anywhere in a consistent fashion, that would be remarkable. And so that's what we did. We created the first fully web enabled front end for clinicians to do their work.
that in the back end was reading Sybase, and mumps, and cobalt based systems, but you never saw that. and then we kept going and going, and it was doctors creating software for doctors, nurses for nurses, pharmacists for pharmacists. The productivity gains were enormous, the satisfaction was absolutely off the charts, and the cost was very low.
And as you point out, there weren't a huge number of alternatives back then.
over the holidays, I finished Isaacson's book on Elon Musk. And one of the things that's fascinating about that is his desire to have all the software be written internally. And there's an interview out there by a Ford CEO was talking about why does Tesla have any an advantage over you?
And he gave this example. He goes, we have 130 control systems in a Ford, in the Ford F 150, 130 control systems. He said they were written by 65 different companies. And, so if we have a software update, we have to go negotiate with those companies. Then we have to, put in this new capability and then roll it out and all those other things.
So Tesla has those same 130 systems. They're all written by Tesla and they're all done over the wire And so essentially whenever he wants to roll something out they come up with a new capability. The company says hey, this is great We're gonna roll it out. They press a button and all of a sudden in everybody's garage around the country It rolls out when I saw that interview and as I'm reading the book, I thought about you because to a certain extent, having an EHR that's written internally by the staff, by the people that are there, first of all, it brings true to the, is it written for the clinician?
Yeah, if the clinicians are a part of writing it, it's designed for the clinicians, but the amount of flexibility that gave you. would assume, I know there's some downsides to it, but the amount of flexibility that it gives you to innovate and to stay ahead of the curve. It's interesting that, even in that case, I think that's been upgraded to another system at this point, right?
I guess in healthcare we don't value the ability to be rapidly innovative. We don't need to be rapidly innovative. Is that why we don't need to develop stuff ourselves anymore?
wOw, what a wonderful question. So back again, in the 90s, early 2000s, and people would argue it was a different time, we were doing 100 improvements to the EHR per week.
So you'd have an idea on a Monday and go live on a Friday. And, as you say, you controlled the experience, and if what you released didn't work so well, you revised it. And it was an extraordinarily rapid cycle, PDSA approaches to getting software out. Which, in today's world, people say, oh, that was risky.
What about quality? What about safety? What about regression testing? What about security? It's okay, fine. It's a little different time. appreciate that. But what I have seen in most hospital administrations Right or wrong, is they say software development is no longer a core competency of a hospital. And that it's pretty hard for any institution, even a large one, to have the speed of innovation of a vendor which has thousands of employees creating thousands of innovations in collaboration with thousands of customers.
I could see the 2024 compliance environment. But here's my, maybe the approach that I like that's hybrid. It gets back to that and rather than or. Mayo has been able to use smart on FHIR as a mechanism of creating a wrapper around an EHR that enables us to deploy any algorithm, any solution created by any vendor.
and we've demonstrated this. And so here's the interesting question. Okay. The EHR with all of its compliance and safety and quality and security and regression testing is done as a community. And it does so on a interval basis to upgrade, but the plug ins, the wrappers those can be instantaneously updated and upgraded through the use of a smart on fire.
application that is a sidecar. And that's what we're doing.
If you were still in that role given your background and whatnot, as you're approaching, again, we'll keep it to generative AI. I realize there's a lot of AI models out there and we talk about them like they're the same and they're very much not.
But let's talk about generative AI. A lot of CIOs who maybe aren't as technical as you who might look at it and go I signed the BAA with Google, with Microsoft, with Amazon, and we're going to roll it out in a pilot. We're going to have these people have access to it.
We're going to do some notes with it. We're going to do some trial responses to inbox and that kind of stuff. So that's maybe almost what I'm calling the CIO that has become a glorified buyer of technology. Some of the others are essentially are innovators, and they are tapping into the innovative community.
And I know you're going to say and to this. They're tapping into the innovative community, and they're going around and they're talking to various people saying, hey, what can we do? And they're thinking about, okay, our platform has to be able to, like Smart on FHIR, be able to bring these various capabilities in.
And still, yet, there's probably another And this is a really dying group within healthcare and it's a much smaller group that's saying, Hey, we're going to bring in the open source models. We're going to start to get some experience ourselves and play around with these things and figure them out.
I think your answer is going to be and, but would you be doing all three of those in the CIO role today? Yeah, and
Let me just reflect on Mayo's strategy because this is a strategy developed with the CIO of Mayo and platform and our strategy colleagues at Mayo. On the one hand, you want to leverage a commercial product that exists and if there's a great commercial product out there that will help you with documentation.
Great. Or inbox management. Great. Go for it. But those are going to be, at the moment, limited to more administrative kind of functions. Do you want to experiment with open source models to create something that may be a breakthrough? Some organizations will have the resources and focus to do that and others won't.
And that's, have to assess your own culture and your willingness to take a risk. But at JPM, we will be announcing the foundation models that we are creating from scratch using a curated Mayo Clinic data and a vendor partner. that is going to produce something that is completely novel and not based on any existent commercial product.
And I know we feel like we had to try it because it, maybe it will hallucinate less. Maybe it will be able to be given to our clinicians to test, and once proven, given to patients. All these things have to be tried. And yeah, as you predicted, I would say commercial where commercial is functional, open source where you'd like to experiment and expand the field, and Work with startups and do some early adoption and testing of their work.
Some of it will work, some of it will not. But this is the only way we're going to address these critical business problems of burnout. Staff
if they gave you a year off from the Mayo Clinic platform and put you in the role of CIO what do you miss about being CIO? I mean, I obviously you just said the role was gonna go away, , but what do you miss about?
And what I mean by that, to clarify, it's the role as we have known it for 40 years is going away and being replaced with a next generation and probably won't be called a CIO. But, what do I'm a teacher, right? That's been my nature over the last four decades. And so what you love is this ability to do cultural transformation by bringing novel ideas into an organization and working with stakeholders to overcome their doubts.
to overcome their resistance. And then to watch, it's like birthing a child. The idea go from, oh, two early adopters to 20, to 200, to 2, 000, and suddenly become part of the fabric of the organization.
And so now you're trying to do that in partnership at a global scale. the skill set applies, but it's very different, isn't it?
Yeah. And so
people say, do you like your job at Mayo? And my simple answer to that is it is the highest and best use of my lifetime skills imaginable, because it requires technology and medicine. Politics and Diplomacy, and it's internal, it's external. I've been working on this role for
40 years. 40 years.
John, it's always great to catch up with you. We only do it once a year, but we'll have to see if it's appropriate to catch up with you another time this year, because I love to Stay in tune with the stuff you're doing. And you're heading over to Davos here shortly. Yep.
Next Sunday we'll fly to Zurich and then head to Davos and 72 hours at Davos and 50 meetings and five dinners.
That's the nature of Davos.
you talked about JPM announcements, are you going to be at JPM as well?
My team is going to be at JPM, but, personal note, my daughter is giving birth to her first child during the JPM week, so I thought it best to be with her and not in
Thank you for displaying your priorities, I appreciate that. John, always great to catch up with you. Thanks again for doing this. Really appreciate it. Absolutely. 📍 I love the chance to have these conversations. I think If I were a CIO today, I would have every team member listen to a show like this one. I believe it's conference level value every week. If you wanna support this week health, tell someone about our channels that would really benefit us. We have a mission of getting our content into as many hands as possible, and if you're listening to it, hopefully you find value and if you could tell somebody else about it, it helps us to achieve our mission. We have two channels. We have the conference channel, which you're listening. And this week, health Newsroom. Check them out today. You can find them wherever you listen to podcasts. Apple, Google, overcast. You get the picture. We are everywhere. We wanna thank our keynote partners, CDW, Rubrik, Sectra and Trellix, who invest in 📍 our mission to develop the next generation of health leaders. Thanks for listening. That's all for now.