Q&A with Virtua Health Chief Digital Transformation Officer Tarun Kapoor, MD: Innovation “is never really done”

Since it was first introduced a year ago, generative AI has been the biggest topic in healthcare, and for good reason. The potential it has shown in improving medical education and clinical documentation has been nothing short of remarkable. In fact, when Tarun Kapoor, MD, was first introduced to ChatGPT, he had “a borderline out-of-body experience.”

Along with excitement, however, AI has also prompted a great deal of hesitation among leaders, according to Kapoor, who believes “an appropriate and healthy degree of skepticism” is going to be necessary as organizations get their arms around the technology. During an interview with Kate Gamble, Managing Editor at healthsystemCIO, he stated that AI literacy is one of the biggest challenges for leaders. But it also presents an opportunity to foster “deep, meaningful conversations” about the possibilities, as well as the hurdles that exist.

In the discussion, Kapoor, who serves as SVP and Chief Digital Transformation Officer with Virtua Health, talked about his approach with digital health initiatives, which starts by asking some critical questions. He also shared insights on the two sides of innovation, why the focus should be on product – and not project – management, and the human responsibility when it comes to AI.

LISTEN HERE USING THE PLAYER BELOW OR SUBSCRIBE THROUGH YOUR FAVORITE PODCASTING SERVICE. 

Podcast: Play in new window | Download (Duration: 28:45 — 13.3MB)

Subscribe: Apple Podcasts | Spotify | Android | Pandora | iHeartRadio | Podchaser | Podcast Index | Email | TuneIn | RSS

 

Q&A with Tarun Kapoor, MD, Chief Digital Transformation Officer, Virtua Health

Gamble:  Thanks so much for taking some time to chat. I know there’s a lot going on.

Kapoor:  There’s a ton going on everywhere, but thanks for reaching out.

 

Gamble:  Sure thing. I’m really interested in talking to you about your role and then what you guys are doing at Virtua.

Kapoor:  So, I’m an internal medicine hospitalist by training. I’ve been with Virtua for 15 years. I’ve been very fortunate; I’m now in my fourth or fifth role with the organization. I’ve been a hospitalist, an informaticist, chief medical officer of our medical group, and I headed up population health. In 2019, I took on digital health as a stretch assignment.

The best career advice I can give to anyone is that it’s always better to be lucky than good. So then March of 2020 came along, and all of a sudden digital health accelerated more in a few weeks than it had in previous decades.

Creating a digital transformation office

We specifically spun up a digital transformation office in 2020 to take advantage of the rapid changes and the rapid adoption that happened during that time, and to see how we could hard-code it. It was amazing; everything that people were doing and how much advancement we were able to make in a relatively short period of time, understanding that this pandemic would eventually ease. And so, with some of the urgency we had behind it, how do we create an environment where innovation can continue to happen?

The most important thing about innovation isn’t the experimentation; it’s the adoption. That’s the key piece. And so, we took the incubator and accelerator model from Silicon Valley and said that anything that a traditional operational division would normally do in two or three years, we would partner with them and try to help them pull it off in six months. That’s where the idea came from.

We maintain a portfolio and we rebalance the portfolio several times a year, just like with a 401K. We go where the organization needs us to go. That’s a little bit about the digital transformation office.

 

Digital doors

Gamble:  I want to get into some of those initiatives. Can you talk more about that?

Kapoor:  As I mentioned, we have a portfolio of domain areas; and they ebb and flow over the years. Again, our office was formally created at the end of 2020. We use the term digital front door, but I don’t like that term. I don’t believe in the concept of one door into a house; therefore, we have digital doors. We do a lot of work with clinical transformation, both on the inpatient side and the ambulatory front. We do a lot of work with our own colleagues and employees in terms of which digital tools they can take advantage of.

The consumer financial journey is another area, along with patient engagement — not so much in terms of satisfaction scores, but how to build tools that make a health system easier to use. Our opening slide for one of the projects we do says, why can you two-way text with your veterinarian and hairdresser — literally any small business on the planet — but not with your multi-billion dollar health system? That’s the type of engagement we’re talking about. It’s not just satisfaction scores; it’s how to literally make it easier to use your services.

 

The “triple consumer”

Gamble:  I would imagine when Covid started to ease, you had to step back and look at how all of this digital transformation is affecting not just patients but also clinicians, and how to ease that workflow a bit.

Kapoor:  You bring up a wonderful point. When we talk about consumers in healthcare, we try to talk about the triple consumer. Obviously, you have the patient/consumer. You also need to look at it from the perspective of the clinician consumer — physicians, nurses, advanced practitioners, etc. — but also your staff. Because in healthcare, one of the most immediate disruptions to staffing for us was Amazon coming to town. A bunch of folks figured working at Amazon would be a viable opportunity. These are lifelong healthcare people, but they wanted to leave. That’s why we talk about the triple consumer.

What historically happens, although not always, is that when we roll out something from a digital perspective, it’s advantageous to one of the consumers, usually at the expense of the other ones. And so, we try our best to see how we can create a win-win or a triple win. If you get all three consumer bases to benefit, that’s great. But if someone has to give up something, how do you minimize the downside to one consumer group if the other group is benefiting? That’s how we try to think about it.

 

Gamble:  It seems like it keeps coming back to that idea of balance. How do you address that, and make sure that a solution that helps one group doesn’t put undue burden on another?

Kapoor:  I’ll give you an example of one that hasn’t generally gone well, which is messaging clinicians through a portal. A number of health systems have stated that they are going to charge patients for answering those messages. Virtua is not there yet, but we may choose to do so. But the question that I think we have to take a step back at and think through is, why is the patient or consumer actually messaging to begin with? It’s not like they’re right. They’re not sitting there bored out of their mind with no one to talk to. They have a question.

Access challenges

We try to look at these things as a form of access. They’re not accessing an appointment, but they’re accessing the clinician’s mind to help with a problem they’re struggling with. And some of those messages don’t require a clinician to answer. Maybe they need a refill, for example. In that case, there are straightforward algorithms that we can put forward to say, ‘you have an appointment in X number of months from now. I see XYZ has been met.’ Don’t send an unnecessary message.

The uptick in messages (more messages, less time)

At the same time, you have to look at the uptick we’ve seen in messaging. Since the pandemic, the number of medical advice messages our clinicians are getting has gone up 300 percent. And so, clinicians are looking at this and saying, ‘Wait a minute — I don’t have any additional time to answer these messages.’

The questions are probably as complex as before because the easy stuff is all gone. We try to take the easy stuff away. That means by definition, the only messages left behind are the hard ones. So, what do we do to make it easier for them to answer those questions? What do we also do to change our processes to give them the time and energy process to answer those questions? In some of those cases, if a clinician takes 20 or 30 minutes to answer that type of question, I don’t think a patient/consumer would have much of a problem saying, ‘You spent a lot of time answering my question. You just saved me a visit. I don’t mind you being reimbursed for all that time you spent.’

That’s what it always comes back to. It’s diving into the nuances of, what is the problem that you’re trying to solve, and asking this question again and again. By the time you peel it back five times, you start to understand what the actual problem is. Where I think a lot of digital and tech has gone awry is they answer, they try to come up a solution for the first problem without peeling back and trying to figure out what was the fourth or fifth problem that was preceding all of that. That becomes hard to do.

 

Car, bike, or skateboard

Gamble:  Right, because you’re talking about changing the thinking, which is not easy to do.

Kapoor:  There’s a diagram I show at presentations that originally came from Hendrik Nyberg, who was a coach at Spotify and Lego, and does some consulting work. In this diagram, the customer says, ‘I want a car.’ And so, you build a car. But in another scenario, the person asks for a car, but instead of just building one, you ask, ‘what do you want it for?’ The person says, ‘I want to get from point A to point B.’ And so, the first thing you do is give them a skateboard, and they look at you like you’re crazy. But you say, ‘Use the skateboard for a week and get back to me.’

And then you start to iterate from there, with the next step being a bicycle. Eventually you may still end up with a car. Unless you live in Amsterdam, in which case you’d stop at the bicycle, because there’s nowhere to park your car.

I think that’s the key. Sometimes you keep iterating all the way to the end, and sometimes you stop early. And that requires the discipline of thinking a certain way.

 

Want versus Tolerate

Gamble:  Interesting. So I guess the big question is, how do you do that? How do you change the processes so that you can start smaller and iterate instead of going right for the big things?

Kapoor:  It takes time. First of all, you need a crazy group of people who believe in it. I’m very fortunate that I have a group of fellow crazies who live and breathe this concept of, ‘I know you think you want this, and I’m not saying no, but would you be willing to tolerate this, which is a fraction of what you thought you wanted?’ What ends up happening is you learn to compromise. Maybe the first time you don’t come back with a skateboard; maybe you come back with a bicycle. Because if they ask for a care and you give them a skateboard, you might get laughed out of the room.

And so, you compromise. You come back with a bicycle. I would argue that our team is still not fully in skateboard mode. We still come back with bicycles more often than we come back with skateboards. But you at least start to build street cred. Because the fear of the internal consumer is if you bring them a bicycle or a skateboard, you’re not coming back. That’s the final product.

If you start to show people, ‘I’m just bringing you this, but I’m not done,’ and you come back with the next iteration, you start to build that trust factor.

Innovation’s biggest killer

There are two sides to innovation: the experimentation side and the adoption/scaling side. The biggest killer of either one of those is fear, because people are fearful that what you’re coming with as your new idea and your new adoption of that idea could potentially end with a bad outcome. And in healthcare, that’s concerning. As a physician, I took the Hippocratic oath. One of the things at the core of medicine is the statement: Primum non nocere, which means, ‘first, do no harm.’

“Imperfect, yet safe”

That statement doesn’t sit really well with innovation. The way we’ve modified it for healthcare is, we will make mistakes. We will bring solutions that will not always be perfect, but they will be safe. And the two are different things. You can still be imperfect yet safe. That’s the mindset we bring. If we can prove to you that it’s safe, but it may be imperfect, will you iterate with us? And I think that’s where we found traction.

 

Product management

Gamble:  And once you did that, you were able to build on that trust and credibility?

Kapoor:  Yes. I’m not going to leave you alone. I think what we’re finally starting to get a mindset that has existed outside of healthcare for decades. And that is the difference between project management and product management.

There are a few examples I like to use; one of them is Tide. Tide is constantly being iterated on. Sometimes Tide is good enough for what it’s doing right now, but then someone notices something else that’s happening and says, ‘can we put Tide in a pod?’ Let’s bring it to the lab and work on creating a pod. Now, Tide Pods are in the marketplace. But do you know what else is in the marketplace? Old Tide. It’s part of a collection of products.

And then another thing comes along, like someone notices a conversation about washing clothes in cold water. Let’s go back to the lab and talk about cold water Tide. Now, it’s part of the product offering. That’s what I mean by project management versus produce management. People love to say, ‘I’m done.’ Unfortunately, if you want to get into this innovation world, you can’t have that mentality. The work ebbs and flows. There are maintenance modes and innovation modes, but it’s never really done. And that’s a change in mindset that we need. Just because it ain’t broke, it doesn’t mean it shouldn’t be fixed at some point.

 

His ‘eye-opener’ with electronic orders

Gamble:  True. You stated earlier that you are a physician. How did you get involved in informatics? What made you interested in that?

Kapoor:  I actually joined a startup right out of training. It was a great experience; I learned a ton. Then I was at another transition point, and I thought, where are the opportunities? I realized that I was still handwriting my notes. As a practicing physician, I was writing my orders out on a piece of paper and I thought, there has to be a better way to do this.

And so, I decided I wanted to learn more about healthcare IT and electronic medical records. I got my first role opportunity when I started doing some informatics work on computerized order entry. The entire process was writing the electronic order. That was the easy part.

What was eye-opening was the fact that we had to have to have a separate conversation. How many of these orders should we just get rid of to begin with? That became a recurring theme. I think we’re doing ourselves a disservice in healthcare in that when we say the word digital, we assume everyone is speaking the language. I have the word ‘digital’ in my title, and so, I’m contributing to this.

Three phases of digital

I think there are actually three phases of digital that we need to be talking about. First there’s digitization, which is literally taking it from analog and putting it into binary ones and zeros. Then there’s digitalization, which is changing the processes from the classic informatics world — the world I was living in.

The third component of that is digital transformation. How do you actually change the business? What services do you add or get rid of? To me, that’s the realm of digital. And you could be in different places.

In my experience, I was doing digitalization, not digitization. I was doing digitalization as an informaticist. Now, I’m spending a lot more time on the digital transformation side and thinking about how we can change the business. Being able to understand those three realms and be able to flex across them is a skillset that’s important for any CIO, CDO, or CIDO. That’s the key to understanding this.

 

Telemedicine’s limitations

Gamble:  Right. And it’s not necessarily linear, right? You don’t necessarily go from digitization to digitalization to digital transformation.

Kapoor:  I agree, one hundred percent. It can start with, ‘I have an idea for a completely different way of offering care.’ That’s a digital transformation idea. But the digitization may not even be done yet. And so, maybe you have to start on one side of it and work your way backwards and ask, how do we change the processes? And before we change processes, can we do this in a digital fashion? Can I even digitize this?

One of the ones we still haven’t figured out is telemedicine. How do I have any type of physical exam on you? I don’t. Now, there are some very interesting things coming down the pike with transdermal optical imaging where the camera can look at your face and pick up your heart rate by looking at the reflection of light through your skin. It can potentially look at blood pressure.

This is where telemedicine has hit some limits. Telemedicine is great, but if you look at the numbers, it’s only used by about 15 percent of the population. There’s a voltage drop on some things I can’t do easily through a telemedicine visit.

This is where I think, to your point, it goes in both directions. Sometimes the technology and the processes are ready, but the business and the payment model may not be ready, or vice versa. That working dynamic of understanding where to go after is also an important piece.

 

Physicians going digital

Gamble:  Looking across the industry, it’s becoming more common for physicians to get into digital transformation type roles. Why do you think that is? What are some of the unique perspectives can physicians bring?

Kapoor:  There’s no doubt that there are more physicians, nurses, and folks with clinical backgrounds entering this space. Is it a requirement? No, but are there potentially some advantages? Sure. Because I’ve there. I’ve been on the front lines, and I’ve seen some things. Trust me, the vast majority of doctors and nurses out there have probably said to themselves a dozen times, ‘how much better would it be if we could do.

I think the superpower for clinicians who are interested in getting into the space is being able to tell the story and walk between those two worlds. Can I do enough to understand and translate clinician needs to the technologists, but also go back and understand the technologist constraints? And then go back to clinicians and say, ‘I know you wanted the car, but what about that bicycle?’ If you can walk between those worlds, that’s key. That’s where I found a solid footing for myself in this world.

 

The speed of progress

Gamble:  There’s also the fact that technology is always changing, and clinical roles are always evolving. Is that something that can be challenging from your perspective as far as trying to keep up with that?

Kapoor: Yes. But what has been just as remarkable as the growth on the technology front has been what’s happening with clinical data. It’s important for folks on the traditional technology side to recognize that the same thing is happening on the medical knowledge side, but faster. Medical knowledge doubles every 45 to 75 days, depending on the source.

How in the world can we ask our clinicians to keep up with that? This is where AI disruption can come into play. If healthcare parallels what we’ve seen in other industries, it tends to disrupt the middle. And the two extremes, the highly humanistic and the highly technical, tend to stay within the realm of the human being. AI tends to sit in the middle of the rules.

AI and physicians working together

If I were to get a cancer diagnosis, would I trust an algorithm to match my genotypic profile with any of the thousands of studies out there that match my genotype and that trial better than a human physician would? Yes. Right. Because there’s no way a human physician can keep up. It’s not possible. But I still want to hear the diagnosis from a human analog voice. I’m not prepared to hear it from Siri.

Hyper-humanistic and hyper-technical

Siri may have talked to the doctor and said, ‘I recommend this trial for this patient, but that’s the hyper-humanistic side of, ‘I want to hear it. I want to talk about the side effects of the drugs — what is this going to be like for me and my family?’ The other end of it is hyper-technical. If you’re in a motor vehicle accident with a ruptured spleen, you don’t want a computer to come over and just start to operate. I’m not seeing that anytime in the future.

But I can certainly see the augmentation piece happening in the middle. The best line I’ve heard is that it’s not the clinician or physician who’s going to be replaced by AI. But the clinician or physician who doesn’t use AI is going to replace the physician who doesn’t.

The same conversation probably happened a hundred years ago when stethoscopes came out. Someone probably said that the physician who uses the stethoscope is going to replace the physician who doesn’t use the stethoscope. How is that any different? And for folks who support technology, the question should be, how do we help our clinicians become this? How do we teach them how to use the stethoscope the same way? How do we create tools that are digestible and usable in their world to help them with their output? That’s where it becomes very exciting.

 

Gamble:  It is very exciting, and it’s a good way to think about it. At its core, AI really is a tool.

Kapoor:  It is. My father was in agriculture. At that time, it was all about biotechnology. He used to say, it’s a tool to help you do something. Use the tool, but know that the tool isn’t going to do it by itself.

 

AI and automation

Gamble:  When you look at AI and all the hype, especially around ChatGPT, what’s your take? How it’s being used is what’s important, but still, it’s hard to contend with the expectations.

Kapoor:  I concur 100 percent. I don’t know exactly where we are in the Gartner Hype Cycle, but we’re still pretty much in the ascendancy of the first curve. I remember where I was when I first used GPT; I had a borderline out-of-body experience. It was unbelievable. The first thing I caution people is to make sure you understand the difference between automation and AI. We toss those words around as though they’re interchangeable and they’re anything but.

The simplistic way to put it is that automation is when a human sets the rules, and a computer executes it and repeats it. That’s the standard definition of automation. With AI, the human or the computer may set the rules initially, but then the computer is allowed to rewrite the rules as it learns.

However, there are cases in medicine where I don’t want the computer rewriting the rules. I’m fine with the computer helping me automate, but I don’t want you changing the rules as you go along.

 

Human responsibility

Gamble:  Right. That could be dangerous.

Kapoor:  It can be dangerous. And so, how do we bring an appropriate and healthy degree of skepticism as we go along? There’s another analogy I like to use that talks about the tools we have in cars right now like blind-spot controls and cameras. In the end, you’re still responsible for the car. Are we better drivers with these tools? Absolutely. I can see much more with my backup camera than I could ever with just a mirror, but I’m still responsible for the stuff that’s behind me. And if I hit something, that’s on me. I think that’s the way we need to think about these tools. Of course, they’ll make you better — and actually, if you’re not willing to use them, that says something else about you. But the human responsibility is still there, at least for the near future.

We’ll see what the next few years bring. I may be wrong, but for the foreseeable future, this is the way we should be doing it. To me, AI literacy is the biggest challenge we have in front of us. How do we get everyone talking about this, understanding it, and having deep, meaningful conversations and not just throwing buzzwords around. That’s the opportunity for the CIO and CDO. One of my biggest tasks inside of our organization is promoting the conversation around literacy of these tools. When’s the right time? When’s not the right time? When should we be skeptical? When should we buy in? That’s a conversation that has to be happening organization wide.

 

Gamble:  I guess there isn’t really a blueprint for that because organizations are so different in terms of size, scope, and where they are in their digital journeys.

Kapoor:  There are actually a few blueprints that can be used, particularly looking at high reliability and safety are critical. In a healthy organizational culture, they’ll say that safety is everyone’s responsibility. Of course, the quality department is responsible for doing this part, and another department’s responsible for doing that part. But everyone has to have that mindset of safety.

Maybe lumping AI and automation into the exact same bucket doesn’t quite fit. But everyone should be thinking about, what are these tools? They should be getting educated on these tools. You can’t just leave this to your IT department and say, ‘it’s an IT thing.’ It would be like saying only finance gets to use Excel. And only legal gets to use Word. We all have to have basic proficiency in these tools.

 

Gamble:  Right. And the ability to execute.

Kapoor:  To me, that’s what will differentiate the winners from the losers. To paraphrase a Steve Jobs quote, ideas are worthless. Execution is everything. Those health systems and organizations that are able to effectively execute will be winners. In healthcare delivery, there have never been losers historically. But that’s about to change.