An Interview with Tim Pletcher

Brian: Hello, welcome to the Augusto Insights podcast. This is your host Brian Anderson. We’re continuing our series on health IT and learning from top industry experts. It is my pleasure to introduce you to our guest Dr. Tim Fletcher. He is the executive director of Michigan Health Information Network shared services, also known as MIHIN, the CEO of Velatura, the CEO of Interoperability Institute, and he is a visionary entrepreneur that is a contemporary thinker in the field of healthcare IT, and it’s my pleasure to have him as our guest today. Welcome, Tim.

Tim: Thank you, Brian.

Brian: Tim, can you just share some of your background? You’ve got such an interesting background. How did you get to this point in your life, where you’re the CEO of three companies that are making a huge difference in our world?

Tim: The short answer is I’ve always loved being at the intersection of technology and what helps people, and I’ve known all along that technology can make things better. And so I’ve always viewed my space as connecting technology to what is it that needs to change or could change to make things better.

I went to U of M. My undergrad, I started in engineering. I got down to the physics of how a computer chip works, and I was like, “Oh excellent, I understand that now” and promptly changed my major to combine computer science, engineering, business, communication, and library science, and created my own major.

That led me on a very interesting journey because the whole time I was paying for college, I was pulling cable to build computer networks, and really you know, there at the way the internet came together.

Just as I was going to go travel around the world and explore when I graduated, the folks over at the medical center said, “Hey could you come over here and help us do some of that computer networking?” and I really never looked back. That just took me on a whole career of standards and integration and tying things together.

I got into the data space when I started running a predictive modeling advanced analytics group a number of years ago, and I got to poke my head under the water of what was going on in all these Fortune 500 companies for predictive modeling and advanced analytics and data mining kinds of things.

At the time I was working at Central Michigan University, and I’d just helped them open a great health profession, and they recently started a medical school. I raised my hand and said “Hey we should get into this data sharing space.” It requires a lot of multi-stakeholder engagement. You have to work with everybody and be technically competent, and not annoy too many folks.

That kind of stumbled my way right into MIHIN, and along the way I did advanced technology, I ran revenue cycle kinds of things, all kinds of pilot projects. I used to run a group called Special Projects with the University of Michigan Health System, which was run around the health system and figure out what’s broken and try to fix it. So I’ve made more technology mistakes than most people have even had the privilege to spend money on.

Brian: That is probably the key to why you have such a deep understanding and vision for these things. All of these experiences have led you to this point, it seems like.

Tim: We’ve built a whole organization that’s learned to fail quickly. Embrace change, embrace new stuff, but get through it really quickly. Be afraid to fail, but don’t stay in that spot too long.

Brian: It sounds like in your education, you defined your own major, and the interesting one that popped out to me was library science. Why was library science important to your thinking back then?

Tim: Because those are the people who are the custodians of information. If you think about it, have you ever met a librarian who wasn’t actually interested in helping you when you go up and say, “Hi I need a book,” right? They’re like, “Oh let me help!”

So that combination of really helping people but also curating knowledge is very exciting. Later, the University of Michigan got rid of the School of Library Science, and they created a whole college called the School of Information, which took many of those same things and combined them together. So it makes sense if you look back.

Brian: When you started doing networking and you were kind of IT focused, right? I’ve worked with networking people before and people that were pulling cables and things, and it’s all about connecting so you can pass the data, but sometimes those guys aren’t as concerned about the data that’s actually moving back and forth, right? How did you move from that into data? Like how did you get to the point where you were able to get in and launch a company doing all this work on predictive analytics and advanced data work for companies?

Tim: So if you’re in the networking business, you know about the 7-layer stack, right? Where you start at the physical layer and sort of climb up. That’s exactly how my career went. I started kind of at the lower levels, but then I got into the networking protocols, but then I got into the web protocols and understood the standards, and eventually you get to health level seven. Here’s a quick test, Brian, what does health level seven stand for?

Brian: Oh my gosh, health level seven. Well, I don’t know! I don’t know how to answer that question! That’s like a trick question, wasn’t it?

Tim: HL7 is the seventh layer in the OSI stack, which is the application layer. And so the integration layer of healthcare applications. I literally just sort of climbed up that stack, then now in the healthcare space for how we integrate all of the different kinds of systems and that whole API network layer.

Brian: I could totally see now because I get the seven layers, and each one of those layers has a protocol and a standard for how you communicate, right, and like, HL7 is really just a standard of how you communicate in a standardized way across entire health networks. Is that the right way to think of it?

Tim: Yeah, in order to build early networks, there’s lots of politics, right? Even inside of companies.

Crossing these organizational lines is a set of skills. The different departments, first they don’t want to share data, they don’t want to be connected. Then everybody wants to be connected. It’s a very repeatable model.

I did a brief stint when I was at U of M that assigned me to the Consortium for International Earth Science Information Network, and that was to round up all the human dimensions data from the Centers for Disease Control in the WHO and marry it up with the NASA earth science information. And that was the time the web was coming out. There were a lot of things like GreenPages and markup standards for how you share and label large datasets.

So it was just a very natural progression for me to climb up to sharing data, and then, once you start sharing data, you actually like it to be useful, and that leads you very naturally into reporting analytics, predictive modeling, data science kinds of things.

Brian: Yeah, that’s super interesting. So with that kind of backdrop, maybe describe a little bit about what MiHIN does for the health community.

Tim: Sure. So MiHIN, the Michigan Health Information Network, started in December of 2010, so it’s been around a while now. It was created to be this focal point for integration for data sharing for all the healthcare information in Michigan. And it wasn’t necessarily to create a big pile of data, but to connect all the people in the healthcare ecosystem to be able to share data, which we often refer to as interoperability.

All of the hospitals are sharing information. It’s going out to thousands and thousands of doctors across the state. We have 1,400 pharmacies connected. And we take a lot of that information and we share some of it with the states for public health reporting kinds of things, as well as anybody who’s entitled to get that data.

In December, we integrated with another organization called Great Lakes Health Connect, and they had a longitudinal health record, so we also now have available a longitudinal health record that does store and archive some of that information. So we’re increasing the way data can be shared across unaffiliated organizations throughout the state.

Brian: Yeah, that’s super interesting, and there just seems to be a massive amount of opportunity to do good with that data, but it also comes with a lot of interesting concerns that maybe aren’t as big as in other industries like privacy and security. Maybe speak to some of that.

Tim: We’ve spent more money on lawyers than many people can even imagine. We have a very robust trust framework or legal framework, and our legal framework has this model for a master agreement that legally connects everybody, and then under that legal umbrella are individual, very specific, we call them “use case agreements,” that explain what kind of data is going to be shared, how it’s going to be shared, technically what the mechanism is, and what are the permissible uses. It’s a very mature and fairly disciplined approach to data sharing, but not many organizations take it to that level the way we do. But it keeps everybody very transparently understanding what’s going on with all our information.

Brian: Yeah, and then just the technology behind that too… To not only have those legal constructs, but to actually have the technology to enforce that and protect people and their information.

Tim: We have some underlying structures there. One was called the Active Care Relationship Service, which Brian, you guys know about since your team helped to work on our next generation of that. That helps track who’s connected to whom and where information needs to go, as well as, we’re a really big Amazon web services shop.

We have a very robust cloud infrastructure, and with that, we get some of the state-of-the-art cybersecurity tools. So we have a seven-person cybersecurity team, and they just focus on protecting information and making sure the right things happen. So it’s a combination of technology and people.

Brian: Yeah it’s very impressive. You mentioned Advanced Acres. That’s been an interesting engagement. We worked with you guys a little bit on the next generation one. That to me almost seems like a security ledger in a sense. It’s granting access to data. When people hear of active care relationships, how should they think of it?

Tim: The geeky people should think of it as a prospective ledger. So it’s aimed for “who do you plan to have a relationship with, or currently have a relationship with?”

The less technical people should think of it as a mass of who is connected to whom, and how is this patient connected to this health plan, this hospital, this emergency department, these doctors, these programs…

And that is a nice way to navigate the health system so when a patient goes to the hospital, an alert can be generated, and all of the entities that have an active care relationship with that person can be notified. Going forward, that’ll include your advocate; that’s a person you’ve assigned to act on your behalf when you can’t speak for yourself.

So it’s a very powerful piece of technology, but it’s also tied into making sure that only people who have a relationship with that patient can really get at, see, or be notified about certain kinds of information.

Brian: Yeah it’s like the map to the routing of where all the information’s at. One of the other projects that we worked on a lot with you guys is called Interoperability Land. That one is really interesting, Tim. Maybe you could just share some of your vision, like what led to Interoperability Land, and where do you see it going?

Tim: So, first of all, Interoperability Land is like a simulation. I don’t want to diminish it by calling it a computer game, but the metaphor is a sort of a computer-like game or simulation where people can practice interoperability. Very technical people or people just trying to learn it.

Over the last decade, MiHIN has had to get people to share data. And what happens is, when two groups are going to share data with each other, the first group goes, “Hey you go first” and the other group goes “No no, you go first” and they go, “Okay, well I can’t share my information because I don’t know if it’s going to work.” And I say, “Okay, well I can’t share protected health information with you because we don’t have a legal agreement.”

So what we do is we bring what’s called synthetic data, data that’s been generated by a computer. It’s not real. It looks real, but it’s not real, and it’s in the right format. And it’s sort of like we arm both parties. So we give each one this simulated data and they practice sharing with it. We’ve been doing this for a while.

We created a tool called PatientGen that generates this synthetic data, and then we started populating fake hospital EHRs, fake doctor EHRs, fake health plan databases, all these little simulated healthcare organizations, and we packaged all that up and created what we call Interoperability Land, and now these new standards like FIHR come out that take advantage of new types of technology like restful APIs, they can practice making calls and crossing organizational boundaries in this synthetic little space.

The way that happens is we create personas. Personas are like stories around certain fake synthetic people, and they go through their lives and get sick and get better and their data is represented in Interoperability Land, so you can kind of test what happens when these people go to the hospital and to the doctor and pick up the medication from the pharmacy.

Under the surface of the story there’s all of the data for the technical people to practice with. That really helps people come to a neutral place and figure things out.

Our vision is that each organization stands up these little sandboxes to try things out, and then the future generations will link those sandboxes together so that they can begin to do more sophisticated types of innovation.

Brian: Yeah it’s interesting. I don’t think people appreciate the complexities of that data, right? The ability to create this synthetic data is a pretty big feat on its own. In real life if I go to a doctor, and then I go to a specialist, then I go to a hospital, then I have a lab or I have an x-ray, that data is spread out potentially across five or six different entities that are sharing basically what my persona is, right? Interoperability Land helps you see that data across those different entities how it is in real life, versus like a lot of trial-and-error and limitations. You can build more sophisticated stuff because of the synthetic data, right?

Tim: Yeah, and one of the barriers we’ve seen is a big organization will try to get permission to work with the organization across the street. They have to go through the technology space, they have to go through the cybersecurity process. They also have to go through the legal process to get permission to do that. Now they’ve spent six to nine months trying to figure out, “Okay how can we just start doing things with the group across the street?”

You can all do that in Interoperability Land, and really quickly, but if you want to add a third group, that’s actually another trip back through cybersecurity police, back through legal, etc. That’s another nine months.

Everybody can kind of do that in Interoperability Land in a much safer, much easier way to understand, and we can kind of spin up in the Amazon web services a complete little mini healthcare ecosystem that’s just for that organization or that little community to play with. It’s a very powerful way to get rid of all the delays that usually occur when people want to begin to prototype a new concept or a new innovation.

Brian: Yeah, that’s really interesting. Is your vision to continue, through the work you’re doing at MiHIN and all of these legal constructs and security control and all this… Like right now, a lot of the people that are entitled to that data are participants in that health ecosystem, right? Do you envision that there’s a day when a normal app developer would be able to build things that help people in their daily lives and have access to this information. Where do you see apps or where do you see innovation going on top of all this data?

Tim: We’ve partnered with a company called Care Convene that does our telehealth service, and they have some very interesting applications that run on phones to help people do teleconferencing and virtual care.

During COVID-19, nobody’s going to the doctor unless they have COVID-19, or think they do. So a phenomenal amount of new care is being delivered by telehealth and virtual care. And that’s very consumer-facing, so we see a lot of that data and the role for apps, to answer your question, is really aligning behind these telehealth tools to help connect people to their doctor.

A lot of the energy around apps was to try to connect the doctor to the EHR, but what we see really happening as it goes forward, is that there will be more and more apps to help doctors do telehealth, virtual health kind of integration. Honestly, if you don’t need to go to the doctor’s office, even once COVID is completely over, why do you want to if you don’t need to? And it’s just a lot more efficient.

So we see a lot more apps happening in that space. That’ll be controlled with you and your doctor or your doctor’s office, and then your doctor’s office will have access to the data that we have, and will kind of flow back and forth, and we’ll facilitate parts of that communication.

Brian: That’s super interesting. It’s interesting working with some other clients and this idea of telehealth or video conferencing, it really has become a mainstream technology due to this pandemic. In my experience, even older generations who have been successful doing what they’ve been doing for a long time, they’re now becoming advocates for this stuff because they had to try it, and they realize that there’s a lot of benefits to it, and it seems to me you get a lot of the benefit of being a person without actually having to go through the hassle of traveling and being there in person.

Tim: Yeah, especially since if you go in person the risk of exposure right now is very high. What we’ve seen through the CARES Act is they’ve actually changed a lot of the rules, or at least temporarily suspended a lot of the barriers that are preventing telehealth and virtual care from flourishing.

And so one of the great things that will come out of this is many many people will be much further along in and actually taking advantage of apps and tools to interact with patients and deliver care. I don’t think we’re going to go backwards on that. I think that’s a one way trip where people are going to be very excited and really prefer that way of interacting with the healthcare system.

Brian: Yeah, I can totally see that. I know that’s true in my life for the most part, unless I really need to. That is interesting insights. In this world, with all this data, where do you see AI and machine learning, and all of those kinds of things fitting in?

Tim: So going back to Interoperability Land, I think we’ll be able to show lots of safe ways that artificial intelligence and machine learning can be introduced. As Interoperability Land talks to different services and people want to tell a story, they can interact where a particular persona is getting a standard treatment, but now you look at that standard treatment with some assistance of AI that kind of takes it back. You can see how the story branches or maybe he has a happier ending.

So I think in the short-term, there’s some very safe places where we can introduce machine learning and AI, and it’s not too scary, right? What I see happening is, and this comes from running a predictive modeling shop for many years, a lot of the energy goes into data science and making a better algorithm. What the rate-limiting step is, though, is not a better algorithm, it’s faster time to action. And faster time to action is about simplification and ease-of-use.

And so turning something red or green or yellow, and having a standard automation that says when it’s red, do this, when it’s green, do that, is a little bit lower bar than what a lot of us in the data science space sometimes think about, and so where I think a lot of the AI, and where I think a lot of the machine learning kinds of things need to go is they need to take a very complex world down to some very simple, actionable steps.

And so, what we’ve seen in the past is that a vendor usually likes to comb through the data, find the patterns, generate the algorithms. Then they want to take their algorithms, score the new data, make a prediction, and have a tool that now tells you what to do. And they want to sell you this whole piece. That’s probably not going to work in my opinion.

My opinion is that folks need to begin to really separate the tool from the scoring, from the generating the actual algorithm that you’re going to find. And so I think to the degree that those get broken apart, and creating sort of layers of APIs and separating the services, where these tools that people are going to use can take inputs from lots of different AI type services and machine learning and analytics services.

They’re going to be the winners, but historically people have tried to sell the whole thing as one big monolithic component. You need interoperability, you need AI type API services out there to be able to do that, and I don’t think in healthcare that everybody’s clued into that. They’re still looking for one of the big pieces, and so FIHR and some of these other services begin to crack that open a little bit.

Brian: Yeah, those are interesting insights. I know that you have an InterOpathon coming up. Can you tell us a little bit about what an InterOpathon is?

Tim: In our language, a hackathon is where you get a bunch of cool folks and they get lots of caffeine and they just generate cool apps, interesting things you hadn’t ever thought of. And then there’s “connectathons,” where you’re really trying to take an advanced standard. You’re really trying to see, does that standard work? Or what are the other feature functions we need to include in the specification for the standard.

The InterOpathon is you’ve got standards, and you’ve got examples of cool apps, and you’re now up here trying to solve problems, and specifically as you cross different organizations. So you’re trying to put all these pieces together to show how something that’s leveraging standards, taking advantage of cool apps, and crossing different organizational boundaries is solving a problem of some kind. So InterOpathon’s really focused on that.

The InterOpathon we have coming up is going to be very interesting because it’s looking at a number of standards. So the standards are like the DaVinci standards, which are a lot of a quality measure standards, they’re really a lot of how you get data from a provider to a health plan. The Karen Alliance standards, which are some of the new pieces of 21st century Cures Act, are how do patients get their data, the last five years of data from their health plans? Or how does one health plan exchange data with another health plan? And these are all sort of standards that are there, and then the last standard is called from the Gravity Project, these are all FIHR accelerators, and the Gravity Project is focused on the social determinants of health, and the social determinants of health are really concentrated on areas like housing security, or food security, or transportation kinds of needs. Start putting all those together and tackling a lineup of problems that use a couple of key personas that really shows how all that works under the hood.

Brian: That’s interesting. You mentioned these different standards, but those standards are also kind of use cases, right? There’s a governing body that goes all the way up to the top of the federal government that is kind of helping to drive a lot of these initiatives. Is that the right way to think about it?

Tim: They’re really kind of channeling into HL7, the standards organization, and really are FIHR accelerators to sort of get the protocols and the restful APIs, and really the technology and the specs built for all of that. And then what the federal government is doing, is it’s really trading incentives or regulations to encourage people to adopt those standards.

And so there’s a bunch of requirements that have recently come through under the 21st Century Cures Act that are requiring health plans to use some of those consumer-facing standards to be able to let patients get that information, or to encourage one health plan to be able to share data with another health plan when a patient moves, when a member moves from one plan to the next. And that kind of space is how the government intersects with the sort of HL7 standardization process.

Brian: Ah, okay. For the developer-minded like me, it’s like the W3 standards, right? The W3C Consortium. They’re publishing all these standards, it’s just specific to healthcare. The federal government might pass a law that mandates that something like this needs to happen, and the standard is already in development or it’s already done, and now you just need to implement it. Is that the right way to think about it?

Tim: Usually, it’s not as much a law, but a regulation saying “If you want money, you need to use these standards” or demonstrate these capabilities. But sometimes for public health reporting kinds of things, it could actually be a law saying, “You must report in this format.”

Brian: Okay, I’m glad you clarified that for me. I’m still learning. So how do people get connected to this InterOpathon if they want to participate in it?

Tim: They can register on our website. It’s more fun if you’ve got a whole team, but you don’t have to have a team. We can team you up with folks, and Augusto’s a sponsor, so if other people want to sponsor they’re more than welcome.

Historically, these would have been done in a big room over a 24-hour period, where people do all kinds of fun stuff and play games and things. This one is going to be completely virtual, where people are getting access to Interoperability Land.

What we do is each team gets their own instance of Interoperability Land, so they get their own complete little mini healthcare ecosystem packed full of synthetic data, a whole bunch of these personas, and a handful of external API sources that they can draw upon, and they can go to town and create interesting solutions to some the problems that get put forward during the during the Interopathon.

Brian: Great. Well we’re not only sponsors, we’re participating in it too. We’ve been working on Interoperability Land and all these use cases from a bottom-up perspective, and now we’re looking forward to kind of driving the car and working from the top down. We’re looking forward to participating in it too, not only just sponsoring.

Brian: Tim, thanks very much for your time today. It’s been a pleasure.

Tim: My pleasure, I appreciate it.