There are times in which it’s wise to think small. Creating a digital health platform is not one of those times, according to Aaron Martin and Sara Vaezy.
It was one of many areas covered in 2022 Digital Predictions, which was published by Providence Ventures earlier this year. “We expect a platforming of digital health solutions — moving digital efforts from execution of tactical point solutions to development of strategic digital infrastructure within the health system,” the authors wrote.
It’s a trend Martin has already noticed at Providence, where he served as Chief Digital Officer for nearly six years before recently announcing his return to Amazon. “If you only tell someone like me the small story” — for example, a solution that focuses specifically on appointment scheduling — “I’m not interested, because I don’t want to have a point solution forever. I want to know where it’s going.”
In the report, he and Vaezy (Chief of Digital & Growth Strategy at Providence) shared insights on what they believe is coming down the digital pike, based on conversations with leaders from Providence and other organizations, as well as entrepreneurs and venture capitalists. Recently, they spoke with Kate Gamble, Managing Editor of healthsystemCIO, about their key findings — particularly around the workforce crisis; the use cases for artificial intelligence that they believe can be a valuable assist to clinicians; and how technology can be leveraged to help physicians use their judgement.
- For Providence Digital Innovation Group, publishing the Digital Predictions report “is a way of putting our thoughts out there, getting the industry to react to it, and gathering situational awareness.”
- One of the key findings? “The workforce crisis is front and center — that is by far the biggest challenge we’re seeing right now as a health system.”
- Getting a digital platform up and running is important, but without a plan to engage between episodes of care, organizations open themselves up to “becoming a very sophisticated price taker in the market.”
- What other industries often get wrong when trying to enter the healthcare space is that “it’s not just the cost of the technology. It’s the cost of implementing it and reimplanting it.”
- When leveraged properly, AI has the potential to become a true asset to clinicians by taking on the “boring, repetitive tasks” that can lead to burnout.
Q&A with Aaron Martin and Sara Vaezy of Providence
Gamble: Can you start by telling us how the report came about?
Martin: We do this every year. We also publish a lot of thought pieces around what we’re seeing in the market. Particularly during Covid, we published a lot of content around our thesis of where things are headed.
The reason why we do this is that on the health system side, we’re regionally competitive. We can share a lot of information, because of the 200 or so large health systems, we don’t compete with the vast majority. It’s a way of putting our thoughts out there, getting the industry to react to it, and gathering situational awareness — that type of thing. This is much more of a forward-looking version of that. Sara and I always say that predictions are always combinations of things you’re actually predicting and things you’re actually working on. It’s a little bit of both.
Guerra: That’s interesting. Can you talk about the process and how you gather information?
Martin: It’s a lot of conversations with our internal constituencies and with industry experts and other health systems. We speak with two to three health systems a week, just trading notes and understanding what’s going on. Sarah, what would you add?
Vaezy: We do this on a continuous basis. We’ll do a refresh, and if there’s something like Covid that was so industry-altering, we do a big event. To Aaron’s point, back in 2020, we interviewed over 150 people, both internally and externally. There was so much written over the last couple of years about digital, so it’s not necessary novel, but we tried to pull together our unique perspective from the vantage point that we have as a health system, as an innovator, and as an investor. We’re putting that vantage point against everything we’re hearing and the analysis we’re doing.
Just to add to what Aaron said, there has been a lot of action over the last few years. Digital health funding has reached completely sky-high levels, and I think one of the reasons why we do this is that health systems are so consumed with Covid; we’re so consumed with our own workforce challenges. But we think it’s really important for health systems to be able to influence this and to be forward-looking.
It’s really important we do that so we’re not just operating under the same framework that we have historically. Part of what Aaron and I try to do is bring those issues forward from a health system lens so that our colleagues can have that same perspective; so that it’s not just internally facing, but also externally facing.
Gamble: How did these predictions factor into the strategy at Providence?
Martin: Our approach is simple to say but hard to do. We focus on things that matter most to our health system. And then there are some follow-on effects that are very positive when we succeed. For instance, the workforce crisis is front and center — that is by far the biggest challenge we’re seeing right now as a health system.
Sara’s team is really digging into that to understand the problem, and based on those findings, we can hypothesize digital solutions to those problems. We go through a very diligent process around that, and there are follow-on effects. One is that the work Sara’s team does really helps the organization buckle down and decide what’s most important; we’re not going to go after shiny objects.
If we find a solution that’s investible with Providence Ventures, that, by definition, is a large market opportunity. We’re not the only ones having these problems. And so, if you find a solution, that’s big potential business. It helps Providence Ventures attract syndicate partners for their investments.
Gamble: Can you talk about the process your team goes through?
Vaezy: Absolutely. Through the interviews and analyses we do, we’re able to come up with common themes. We start with big buckets like the workforce challenge or behavioral health that are looming large, but not necessarily well defined or may have many different components, and then we disaggregate those component parts. As Aaron said, it’s easy to say but less easy to do. It’s a well-worn methodology.
And so, we disaggregate it into problem statements that are more specific and more quantifiable, and we place value against it. For example, how many patients are impacted worldwide, or what’s the economic impact of something like this? We take those opportunities and tier them according to a set of criteria and a framework, and once we’ve prioritized everything, we go through it again with our internal constituents to determine how what we’ve seen influences what we’re actually going to work on.
It’s interesting because some of the things listed in the predictions are not necessarily new to us as a system. But a few years ago, a lot of it existed more in fragmented products we were building, and now it’s coming together in terms of our overall digital platform. And so, it’s not always de Novo creation of new ideas; sometimes it’s how the pieces fit together, which is a lot of what we’ve seen over the last couple of years.
Gamble: One of the challenges that arose is engaging patients between care episodes. Can you talk more about that?
Martin: If I had to choose the most pressing issues that health systems need to deal with in the short term, medium term, and long term, the biggest in the short-term is the workforce crisis. The reason is that if you don’t get past it, there is no medium term. It has turned the P&L upside-down for all of these health systems; labor rates are rising faster than revenue rates — that’s just not sustainable.
The second medium-term issue is peer-provider integration. That’s the digital component of a larger problem of getting under risk. During Covid, we saw the devastation that happened on the fee-for-service side of the business. For health systems that also own a payer — as we do in Oregon, or as Presbyterian in New Mexico also does — you can survive it pretty well because there’s a stable set of cashflow from taking care of a population instead of being paid to do piecework.
That second part of that issue has to do with payers. In our market, the biggest disruptive force in the medium term is Optum, United, and some of the other national payer providers. They’re building networks and are putting a lot of energy and effort into making the customer experience more coherent so that it doesn’t matter whether it’s the payer side or provider side — it just functions in a seamless way. If we don’t do that, we’re in a boatload of trouble.
In the long term, it’s the engagement piece. And by long term, I mean five years, not 20. What that means is, if you get your digital platform up and running, and you’re really effective at attracting new patients and transacting with them online, you’ll give them a great first digital experience. But if you don’t have a plan to engage in between episodes of care, you open yourself up to investing a lot of money in becoming a very sophisticated price taker in the market. Because somebody else is going to create that relationship and engage, and then they’re going to be dictating the terms.
The analogy I always use is Amazon Prime. One of the reasons Amazon has grown to what it is today is Prime. It’s not necessarily its e-commerce capabilities, like creating a frictionless experience and being able to ship things to your home — it’s the engagement platform they’ve built. That’s probably the biggest and most successful engagement platform in the world. If health systems don’t build something similar, they’re going to be price takers.
Gamble: One of the other aspects I found interesting was the platforming of consumer healthcare. Can you talk about what that means and what it will entail?
Martin: It’s interesting. I agree with everybody who believes that platforming of healthcare will occur. I probably disagree with the ‘how.’ A lot of folks think they can come into the industry, throw several hundred dollars into it, and convince health systems to rip out their existing infrastructure and re-platform right. What they’re missing is that it’s not just the cost of the technology, which could be replaced at the right terms. It’s the cost of implementing it and reimplanting it. And the pain and the change management.
For instance, we’re just now getting through migrating to a single EMR with Epic. It’s been a five-year journey that I don’t think anybody wants to repeat. That’s not how platforming occurs. What happens is, these technology companies earn the right to become a platform, first by building a sharp-focused value proposition of how they’re going to pay back your investment in 18 months. It can seem like a small story, but then it moves into a bigger platform story.
From ‘small’ to ‘big’ platforms
One of our early investments was a company called Kyruus. When we first met them, we were customer number three — that was seven years ago. Now they have more than 200 health systems on their platform, and something like 60 percent of all physicians on their platform are employed by providers. And so, they executed on their vision.
But when they first approached us, the story was small. It was about our call center — it wasn’t about engaging online because this was seven years ago. It was, ‘you have a call center in which you’re getting inaccurate information about your physicians that isn’t the source of truth. We can fix that.’ That’s very small, but they executed on it. And eventually, we moved to taking the same data to build a platform for our web directory, to doing patient acquisitions through booking. Then, they acquired a company called HealthSparq that enables them to take that source of truth data and provide it our partners. That was always the plan.
If you just tell someone like me the small story, I’m not interested because I don’t want to have a point solution forever. I’m interested in where this is going. But if all you talk about is the big story, how am I going to rationalize making the investment? Health systems aren’t rolling in money.
Gamble: So there’s a balance you need to strike. I’m sure that’s challenging for a lot of organizations.
Martin: It is. I would be shocked if rip and replace becomes a thing; at least on the provider side. It’s too expensive.
Use Cases for AI
Gamble: We did a survey of our readers last year, and there was a lot of talk among CIOs and chief digital officers about going forward with artificial intelligence and making sure digital technologies are being leveraged. Have you found that to the case?
Martin: AI is going to be incredibly important for us. We’re seeing a lot of different use cases for it. One is navigation and the other is physician burnout reduction. There are two sides of that. With the Cures Act, a lot of information was made available to patients, which generated a ton of questions, and a ton of messages to physicians. You can get a lot of these messages automatically answered by a bot in an effective way. That was one use case.
Another case, which we talked about in the piece, was around navigation. Our CEO Rod Hochman says that today’s health systems ask patients to assemble the tinker toys of their care, by themselves. Things like, ‘you need to get a Covid test.’ Or ‘you need to get this checked.’ But the information isn’t presented in a conversational manner; it doesn’t tell people what’s the next step. You can imagine a bot doing that for you and putting together lists of what you have to do and here’s the right next best action from a navigation standpoint. It’s having a persistence so that you’re not constantly telling the bot who you are and giving the same information.
The third area, which I’m super excited about, is different diagnoses. It’s much more long-term, although we are starting to experiment with it on the platform today. If you put in a set of chief complaints, it will differentially diagnose you and say, ‘you have one of these three things.’ And then it will recommend, through navigation, how to get treated, whether it’s through telehealth or a primary care physician’s office.
“It becomes an assistant”
You can imagine a world in which that becomes an assist to the clinician. And eventually for low-acuity, algorithmically treated conditions, it becomes not just diagnostics, but also for treatment. That’s far down the line, but you see things like EKGs that are effectively read by the computer these days and cardiologists just read the findings. That’s a form of AI where it’s actually doing the diagnosis. The cardiologist will look at it and confirm it. And so, in some cases it already exists in the market today.
Vaezy: Fully autonomous AI is one thing, but I think AI is definitely out there, and can currently be used in a number of different ways to support and reduce clinician burnout on certain things and to aid self-service. It definitely exists.
Gamble: It’s really encouraging to see AI talked about as a way to reduce burnout. I think that as patients start to see this, they’ll have a better understanding of how it can help with care delivery.
Martin: And if we don’t have bots eventually — in like a 10-year time span — doing diagnosis and treatment and taking on some off the functions for physicians and nurse practitioners in low-acuity, low-risk, high-prevalence conditions, we’re in serious trouble, because the math doesn’t work. There won’t be enough clinicians to serve the number of patients.
“Boring, repetitive tasks”
The other thing is that no one likes boring, repetitive tasks. Try asking a nurse practitioner or physician if they like using their judgement or if they like seeing same thing they’ve seen millions of times, and they exactly what it is. The patients even know what it is. And it can be diagnosed and treated algorithmically, versus having them use their judgement to manage a patient’s overall care, which is what they went to school for.
And by the way, boring and repetitive is where humans fail the most. Because once you get into a boring, repetitive set of tasks, the error rates go up over time. Humans aren’t good at that.
Gamble: Right. We go on autopilot.
Martin: Exactly. Here’s the interesting thing. If you ask me if I can imagine a world in which I’d feel comfortable flying without a pilot in the cockpit, I’d say, ‘No. That’s insane.’ If you ask my son, who is 17, he’d say, ‘I don’t know; I’m worried he or she might override the machine, which might be better at landing the plane.’ The pilot might panic and grab the wheel, even though the machine is handling it fine. And in fact, most landings today happen through automation.
The point is, my son’s generation is going to be less trusting of humans and a lot more trusting of machines. It’s going to be interesting.