Episode 02: Driving down the cost of healthcare with AI

Sam Kina – People can be buying the wrong benefits for years and have no idea they don’t know it’s the wrong benefit, like how do you know, your health plan was the right choice? Or how do you know, you know, that, you know, the doctor you went to is the right doctor for you to go to it, it’s that you don’t that that feedback is a lot slower than you’d see maybe in some other product market areas.

Brian Colburn – Welcome to Creating Healthier Futures, a new podcast from Alegeus. I’m Brian Colburn.

Anna Lyons – And I’m Anna Lyons. In this series, we’ll reveal the state of healthcare consumerism through trends and research, our own data, and the health benefits experts that are driving our industry forward.

Brian Colburn – To download the materials discussed in this episode, or to learn more on the topic, be sure to visit Alegeus.com. Thanks for tuning in.

Anna Lyons – In today’s conversation, we dive into healthcare tech, specifically as it relates to how consumers select their coverage and pay for their care. While innovation is happening in every corner of the industry, adoption has severely lagged. But then, COVID-19 hit. Healthcare processes as we knew them came to a halt and people and institutions were forced to adapt on the fly.

Brian Colburn – This shift has provided a much needed spark for healthcare tech and an exciting opportunity for the kind of growth the industry needs. We spoke with two pioneers at the forefront of this change to understand how far the technology has come, particularly in the last few months, and how far we still have to go in helping people make better healthcare decisions.

Sam Kina – Hi I’m Sam Kina, I’m SVP of data science and economics at Picwell.
I am a health economist by training. That’s kind of the field I went to graduate school for. And that’s what I’ve been working in my whole career. So I immediately prior to Picwell, I was working as an economic consultant. And so working for a lot of companies in the healthcare space. And I’d also spent some time working at the Congressional Budget Office, whereas working on health policy, health care cost forecasting.

Brian Colburn – Could you share a little bit about the Picwell business and what makes it unique?

Sam Kina – Picwell is a company that was founded about maybe seven years ago now by a group of economists at University of Pennsylvania and Wharton Business School. And it was founded around this idea that you can apply a lot of data and machine learning or artificial intelligence to help people shop for health insurance. And it kind of came out of this line of research in the health economics field that, you know, in spite of the fact that health insurance is a really big purchasing decision that people make every year. They’re terrible at it, and are wasting a lot of money in the process. So they founded this company. And since that point, we started out just providing decision support for Medicare. But we’ve since expanded to go into employee benefits. And now that’s the bulk of our business, helping employees navigate their benefits, not just health insurance, but a lot of the additional supplemental benefits that come with that. And so, at the core of that is utilizing data, utilizing machine learning to help people really find the right benefit.

Nate Maslak – Hi I’m Nate Maslak, I’m one of the co-founders and CEO of Ribbon Health.
Ribbon is a healthcare enterprise API layer that powers other healthcare solutions to be able to get accurate and comprehensive data on doctors, which insurance plans they accept, and the cost and quality of care. So similar to how Picwell was helping somebody kind of understand the right decision across their benefits, Ribbon’s solution is helping other companies enable that for the choice of doctors. But rather than us having to interface and try to guide that decision, we’ll provide the data layer for a company like Picwell to then be able to even better optimize that decision.

Brian Colburn – Great. And thanks, guys, for joining today. I’m super excited about this sort of topic of healthcare and technology. And so maybe, let’s start at a high level for a minute. And, Nate, why don’t we start with you? How would you rate the effectiveness of technology and healthcare today? So maybe score of one to 10? And then a little bit of color around why.

Nate Maslak – Yes, I think we’re probably at like a two or three, because I think the industry is very much in its infancy. And while I think that technology and healthcare has done an immense amount for how people can think about their care decisions, figuring out how they spend on health care, and why they spend it the way that they should, I think there’s so much more that we can do. And I really feel like we’re in the first or second inning of this. And so that ten is on what could happen and what is achievable relative to where we are today.

Brian Colburn – Right, Sam, how do you see it?

Sam Kina – Yeah, I mean, it kind of in thinking about this, I have come to about the same number as well, I had written down, I think we’re at a two. And some of it depends on when you say technology, what specifically are you talking about? You know, obviously, for the provision of health care, we’re incredibly technologically intensive, we have some pretty advanced healthcare in this country. But in terms of like the information technology, the technology that can really help organize information and help people kind of understand what they need and what they’re buying. That’s, pretty poor. And there’s probably a number of reasons for that, that we can get into. But that’s an area where there is enormous potential to just improve all facets of health care from the delivery side to kind of the purchasing side and decisions around the benefits that help facilitate those purchases. I’d say healthcare lags behind a lot of other industries. And in that capacity.

Brian Colburn – All right, you guys are pretty consistent around the two out of 10. So help me think about two to three years from now? What do you think that number looks like? And what’s really changed?

Sam Kina – I think one of the things that kind of led me to say I think we’re at a two is, is that a lot of the delivery of health care, and then a lot of the technologies that are available to kind of you and me as patients or consumers are things that they’re not like these direct to consumer products, the decisions around taking up this technology, they’re made by institutions, and these institutions tend to move pretty slow. So I think there are a lot of organizational impediments to really rapid change here. So I’d say I think the number would be higher, but I don’t think it would be dramatically higher. Hopefully I’m wrong. But But I’d say you know, we could probably get to a four photos to market on That scale. I think that that’s twice what we’re at now. That’s pretty good.

Nate Maslak – Yeah, at risk of just agreeing throughout the rest of the podcast, I do think there will probably be like a three or four over the next two years. But, and I agree with the point that some of these larger institutions might be a little bit slower to move, but I think one of the inflection points that we’re seeing and why I’m optimistic that maybe five, seven years from now that rate of growth is exponential, is that you don’t necessarily have to work with larger maybe sometimes slower moving institutions. While I think they’re really important, we’re also seeing this new wave of entry coming in. New ways of care delivery, new ways of paying for healthcare and Consumer Directed models that I think five years ago would not have been possible. And because of just the rapid entry of these companies and venture capital dollars flowing in, I’m really optimistic that two to three years from now, the industry looks very different than what it looks like today. And I think some of the slower moving institutions will become much faster moving, because they’re going to have to, and then the other ones are going to play hard to catch up.

Brian Colburn – Yeah, makes sense. Yeah, I mean, from our perspective, one of the things that’s, that we’ve seen, I think everybody’s seen in healthcare is, you know, there’s this idea, if you build it, they will come. And obviously, that hasn’t really been the case in healthcare, whereas it has in other consumer sectors. And, Sam, I think your view was, and Nate, I think you sport is that part of the challenge here is you’ve got this intermediary of the employer. Right, who’s making the decision. What else can be done to get employers to move more quickly? Or what would have to be true, based on your experiences?

Sam Kina – Yeah, so I’ll start and I think, you know, Nate, one of the things you brought up that you’ve seen a lot more movement on kind of the smaller end of the market. So maybe it’s not the fortune 500, but you have a lot of these smaller employers who are maybe making decisions faster, there may be less averse to change, are, are adopting these technologies, maybe at a faster clip. And so I think that’ll help just promote awareness of these products, and also, you know, help really demonstrate that these technologies work. And I think, you know, when we think about, for instance, things like Consumer Directed health care, that’s something that’s it’s been around for a while, as a term, it’s been sort of, I’d say, it’s plateaued in terms of its growth, if you look at just the types of health plans people are enrolling in, but there’s some good evidence that it can reduce costs. And so I think that can be accelerated when you combine that kind of structure of a health benefit with tools, like Picwell, that’ll help people kind of see the benefits of a consumer directed health plan, but then it has to be paired with the types of technologies that Ribbon offers where you say, Okay, I’m a, I’m a consumer of health care. Now, I have no clue what I’m buying, I have no clue what it costs to go to this doctor, which doctor I should go to. So we really need to give people these shopping tools. And as you get it out there, as people use it more, we’re just going to have a huge body of evidence that we can use to really make our case, as vendors of this technology, creators of this technology to help it proliferate. So I think just getting it out there, being able to demonstrate that they work is going to help a lot.

Brian Colburn – Any differences for your side, Nate?

Nate Maslak – Yeah, I think so I agree that smaller employers are moving quicker than larger employers. But I also think that you have a pretty meaningful batch of large employers that are driving innovation in the space. And I would say more so than even some health plans. Walmart being a perfect example. I think it’s a company that can redefine how care is paid for and delivered in this country. And we’ve seen them do some pretty interesting and innovative things over the last two years, I would say, much more interesting and innovative than a lot of smaller employers because they have the pockets and the budget to be able to do so. So I remain optimistic that large employers who are looking at their kind of profitability dwindle because of how much they have to spend on health care, are really leaning into this and the ones that are going to continue to out compete are the ones that are going to be smarter about how they help their employees pay for care. I think the other thing that we’re seeing in the space is kind of drive adoption faster. What I think employers need is infrastructure. I think infrastructure could come in the form of tools that the benefits and HR department has in order to be able to help drive a decision. And I think even a layer of behind that and this is a shameless plug for Ribbon is the infrastructure that enables these solutions to be accurate, to see that how quickly they can be developed, how quickly they can be modified. We’re very long on that bet. And I personally believe that employers are going to be one of the core drivers of the shift to a different healthcare model.

Anna Lyons – Nate and Sam touched on where we may see the acceleration of healthcare technology in the most general sense. But “technology” can mean a lot of things. To gain a better understanding of where the healthcare industry is headed, we need to look at specifics like artificial intelligence – A.I. – and machine learning – M.L. We also need to dive deeper and make sure these types of technology are making a real impact and not just acting as buzzwords.

Nate Maslak – I think AI in healthcare is going to be at the base of almost everything that’s happening, because there’s so much information, so many decisions that need to happen. And I think we’re starting to see a lot of applications on ranging from computer vision and helping understand and MRI, all the way to the other side of how do you help our consumer choose the right health plan. And so I think that we’re starting to see where the value is happening. I think one of the issues that I think the whole industry is acknowledged by now but if you look back to a year or two ago, everything was AI, everything was ML, and AI and ML are a means to an end. And healthcare at the end of the day is between a patient and a doctor to drive the best care delivery model that we have all this other stuff that happens around it. That’s what AI and ML can help make that movement between a patient and a doctor more seamless than ever. And that’s where I get really excited. And then there are all these tools that get you to that point or help support that decision. I again, think we’re in the first thing there. I think that the difference is that the amount of hype that has gone into AI and healthcare has been just absolutely ludicrous and ridiculous, in my view. And I think that we are now back in the quadrants normalized to Well, what does that actually do? How does it actually achieve that outcome that we’ve been trying to achieve for 20 years?

Brian Colburn – And Nate, do you have any examples that you can give listeners of where it’s actually being used today versus just being talked about?

Nate Maslak – Yeah, I mean, companies like Ribbon that are trying to predict the right piece of information on a doctor or have somebody match a doctor, anything around decision support, I think there’s an amazing opportunity that we’re seeing that happen. And then on the care delivery side, I think we’ve seen it in areas like clinical research and areas around pharma and how to map a patient to the right clinical trial. Those are areas that are hugely impactful and just grossly inefficient.

Brian Colburn – Sam, how are you guys thinking about AI at Picwell? And any use cases that you’re kind of particularly fond of today or you think are pretty interesting.

Sam Kina – Yeah. So you may have talked about a few use cases on the delivery side. And there’s certainly a lot there. But also on the I guess the demand side, on the customer side, there’s a lot including what we do. And so when you think about the core problem people face when they’re choosing benefits when they’re trying to figure out which you know, what health care they need. It’s inherently this prediction problem. And then people are trying to make decisions with a lot of uncertainty. Like if you just think about what insurance is, health insurance or any insurance product here, you’re saying, it’s worth it to me to pay this money up front to manage some, some risk on the back end. And in most cases, that risk is people have no clue what their health care risk is. But we have just tons and tons of data to help people understand this. And so, you know, I think when you hear the term AI thrown around, you think probably you can get caught up in the hype and think it’s something that it’s not, but if you get into the kind of the specific sub domain of AI that we deal in with machine learning, it’s just it is a tool as Nate mentioned, we’re using it to predict raising it to predict risk and it just so happens that AI is better able to leverage all the data we have and generate better, more meaningful predictions than if we use some more older classic prediction methods. And so we use that we’re basically giving people these personalized risk prediction models. So then you can see, okay, here’s, here’s what your risk is, here’s what the risk of you or your family and then we can kind of layer that in with preferences to see, okay, okay, we know your risk now, how much do you care about protecting against it and, and so what we’re doing right now is looking at that specifically in the health insurance domain. But when you look at a lot of the benefits decisions, employees have to make more broadly, and just general financial decisions people have to make, it’s this big risk management problem that we have to deal with. And a lot of these risks are correlated. And we’re trying to manage that correlated risks by piecing together just a whole patchwork of different benefits.

Anna Lyons – Internal data from Alegeus confirms this disconnect between what consumers know about healthcare versus what they need to know. By our estimate, about 88% of Americans don’t know enough about healthcare to make good decisions, and that percentage hasn’t changed over the years. To address this problem, we want to move from the do-it-yourself healthcare model of today to the do-it-for-me model of the future, using AI and ML to get us there.

Brian Colburn – Consumers, it seems, are ready for this transition, too. Alegeus recently ran a survey around technology, where we found that about three-quarters of consumers were open to using AI and other technologies in their healthcare planning decisions, but only about a quarter of them had access to those tools.

Nate Maslak – I’m glad you kind of brought up the human component, because I think why we frequently forget when we talk about AI, in healthcare is that the end of the day, it’s meant to drive a consumer decision to do something better than what they could have done before. And so when I think about that disconnect, I think a part of it is that there’s a need to focus more on that end user. And employers are thinking about that end user, and employees are certainly thinking about that. And using those terrifying moments of vulnerability when you’re sick, or can even choose a new health plan for your family. I think those are the moments where AI starts to get applied more and more and more. And that’s where I think we’re so early on. And so for me, I think it’s helping bridge the gap of an employer understanding of what why does it matter that this algorithm exists and how they drive us It matters, because it’s going to save costs for the employer, and matters because it’s going to save costs for the employee, it’s going to decrease absenteeism, they’ll be more productive, they’ll just be generally happier. And I think that the industry went a little bit awry, when we forgot to connect what technology is doing for the human being who needs that technology. Because the consumer I don’t think cares if it’s ML powered or if there’s somebody on the other line, who’s saying, here’s the right time for you. Like they just want the right decision. They just want to go see a doctor when they’re sick. And so it’s all about how that gets delivered and a clean interface and make it simple. So that’s how I think that gets put back together and that disconnect gets bridged.

Brian Colburn – Yeah, I totally agree with that. And, Sam, if we build on that, from your perspective, do you think that that that employers are sort of yet convinced of the power of these tools? You know, is there enough data at this point? Because that seems to be one of the challenges that for years, there have been all these companies that have come along with promises. And to some extent, there’s some inertia that we see at the employer level, even though we’re 100% convinced of the value and effectiveness.

Sam Kina – Yeah, I think there’s a mix of opinions out there. Because there are tools that will help people navigate benefits that don’t use AI. And they’re tools that do, but at the end of the day, it’s what you want is your employees to go through and have an experience where they’re, you know, ideally, they’re choosing the right benefits. And it’s just an easy process that they’re satisfied with, but is the kind of situation where it’s not clear, there’s not an immediate feedback in terms of like, if someone picked the right benefit or wrong benefit, people can be buying the wrong benefits for years and have no idea they don’t know it’s the wrong benefit, like how do you know, your health plan was the right choice? Or how do you know, you know, that, you know, the doctor you went to is the right doctor for you to go to it, it’s that you don’t that that feedback is a lot slower than you’d see maybe in some other product market areas. So, you know, in some regards, there is this problem where you were, without that immediate feedback, how do you kind of demonstrate and generate evidence that AI is the right tool for the job, because it’s not like, you know, just give people AI say, Hey, this is great, you’re giving people a solution that is facilitated by AI. So one of the things that that we can do is, is track over time and see, you know, are we saving money, that’s kind of like a bottom line that, that people care about. And so if you’re looking at health benefits, you can see how much our employees or it’s the employer spending on benefits. If you’re looking at things that help them consume health care, you can see, you know, are they getting the right care? Are they getting more preventative care? Are they getting cheaper care, and those are things that you can track and build up a lot of evidence to help people really see the value of these technologies.

Brian Colburn – And do you guys see any consistent themes at Picwell, particularly around plan selection, like mistakes, that seem to be really prevalent or consistent?

Sam Kina – Yeah, I mean, we see it all the time. And it’s basically people over insure, and they do so to a massive extent, like they make mistakes that, you know, so to some extent, you want to pay more to manage your risk, but the level of risk aversion that would be required to justify the choices that people are making, are really far out of whack with risk preferences that you see in other areas of your life. And so that leads us to think that this is probably irrational behavior. And so even after we adjust for those preferences, we’re finding, you know, about 90% of people should be in high deductible plans with a health savings account. But you know, you have the adoption is going to vary from place to place, but you typically see it in the 20% 30% or lower than that. So there’s this big mismatch and, and a lot of money wasted. And, and the, the money people are foregoing that just gets compounded over time.

Anna Lyons – In the same Alegeus survey, an overwhelming number of respondents said that the cost of care is the most frustrating aspect of healthcare and health benefits. Technology has the potential to jump in and solve these problems. Because let’s face it – we can’t all afford to have a healthcare financial advisor or rely on a friend in the healthcare field to help us make the best decisions. The beauty of technology is that it scales efficiently and can manage costs effectively, democratizing access to better decision-making.

Nate Maslak – I think Sam gave an awesome example, the fact that people are over insuring, which quite literally means they’re just overpaying for insurance plans that they don’t necessarily need. So that’s one good way to take costs out of the system, and for a consumer to save money. I think another one where we spend a lot of our time with our cost effectiveness and quality products is around helping somebody find the right doctor. So if we start with just helping somebody go to an in network provider, that is one of the most predictive drivers of how cost effective somebody carries in network visit might be $50 to the patient and out of network visit might be $2,000 to the patient. Or somebody going to a telemedicine visit when their child has an ear infection would cost them 25 bucks, the emergency room could well cost 4k, and they’ll probably get sicker while they’re in there. And then have even more cost with all the tests that happen in there. So those are very visceral moments that we see happening, and especially for employers who are on the hook for that that cost of care. That’s, that’s a really important thing to control for. And then as you start to think about elective procedures, things that may not be quite as urgent, that’s where consumers have an opportunity to shop. So a knee replacement surgery for somebody in a high deductible health plan where they’re on the hook for their care, that could be as low as $5,000 for somebody at a good doctor who does that a lot. Because of the lower facility fee, see them in a different surgery center. Seeing that right back there. It could also be $60,000. And the consumer that might have a $10,000 deductible, they’re going to pay the first $10K and employers pay another 50,000. That is suboptimal for basically everybody. And in an industry where cost and quality of care are entirely uncorrelated. And if anything, sometimes negatively correlated. You’re also likely to end up getting a worse outcome when you’re paying more money for something. So nobody’s really winning in that scenario. And so I think those are just a few of the examples I could go on.

Brian Colburn – Yeah, so talk a little bit about quality versus cost, because that’s one that consumers often get wrong, they assume more expensive equals better. Can you explain why that’s not always the case?

Nate Maslak – Yeah, it is frequently not the case, because of how healthcare costs are incurred. So what we find is that the more frequently a provider does a procedure. So if you’re getting a knee replacement surgery, you probably want to go to someone who does a lot of knee replacement surgeries, not someone who does a lot of shoulder replacement surgeries, they’re also more efficient at delivering that care. Also, because they’re a higher volume provider, they might be more comfortable with a lower reimbursement rate from a health plan, because they’re getting value. The patients getting better care, the doctors making more money, the health plan is paid for less money. So that’s how you take costs out of the system. That’s one example of there were many times where cost and quality are not going to be correlated.

Sam Kina – Nate, I’ve actually got a question for you about this process of choosing a physician, which I think is really fascinating. Because when you look at how people feel about their physicians, people are really attached to their physicians, that’s one of the most important things that that they look at when they’re choosing health plans is my doctor covered. But a lot of times, that’s kind of like, reflecting a lot of inertia, a lot of you know, built in, you know, personal relationships, that resulted from an initial decision that probably wasn’t well informed. It’s just we’re creatures of habit. And, and so it’s not clear, you know, like for the doctor, I have that I went through some really rational, smart process to do this, just like this person was close to my house and accepting patients. So, you know, from your perspective, where you’re taking a data driven approach to this, you know, what are some of the mistakes you see people making, or at least information people are ignoring? And how is using technology, curing those decision mistakes?

Nate Maslak – One of the things that we also see, when people are choosing a plan during open enrollment, the number one driver of the plan they choose is the doctor that they currently work with. And if they’re switching plans, is that other plan going to be in network for their provider. And I do think that people get really attached to their providers. And sometimes that’s a very good thing, if you spent 15 years going to the same doctor, you have trust in that provider that they have your interests in mind. And I do think that that is important. But I think that one mistake that people make is over leveraging things like patient satisfaction scores, or physician readings, especially in specialties where that’s not predictive or important. So similarly to cost. patient satisfaction is uncorrelated with outcomes and surgical silence. In fact, it’s probably a little negatively correlated to outcomes, because it again comes down to experience when somebody’s choosing a provider, it’s really important to be able to pair the right information for that type of doctor. With that decision. On the primary care side, patient satisfaction is hugely important, you want to have a relationship with your provider, you want to know that they have your best interests in mind. And so that’s just one area where we see people start to make some mistakes. Location is usually the number one driver. It’s convenience, it’s access to care. And sometimes it actually makes sense for somebody to drive a little bit further to go to a provider, that’s going to be more cost effective and higher quality for them. But I think the reason that people don’t do that is they’ve never really had the information to know that. And so without any information, you assume that you can only control for the one piece of information, you know, which is how far you have to drive to get to a doctor.

Brian Colburn – That’s really helpful, Sam, you know, as I think about where we are, in COVID, kind of heading into what promises to be probably a really interesting and unique open enrollment period. Can you talk a little bit about the tools that Picwell makes available, that can actually help people navigate the system in a more remote type of environment?

Sam Kina – What Picwell is built for is giving people guidance and education around benefits remotely online. And that’s where we’ve kind of seen an increase in activity over the last several months, because there’s just much more demand for that. But essentially, what we’re doing is providing a place where people can get all the information they need about their benefits, but also get very clear guidance about what benefits they should be choosing and what benefits are going to provide them with the best value. And so, you know, the I think the highest level, we’re helping people understand what health insurance plan should be right for them.

Brian Colburn – The adoption of healthcare technology has been sluggish over the past decades, but as we’ve learned in recent months, consumers are more than ready for change. The technology exists to make finding, paying for, and managing care more affordable and more efficient. Now, it’s up to employers and benefit administrators to embrace this technology, push for engagement, and make clear the tangible benefits of using it.

Anna Lyons – Our thanks to Nate from Ribbon Health and Sam from Picwell, for being part of this episode and sharing their perspectives. If you would like to learn more about this topic, our guests, or download the research discussed in this episode, visit Alegeus.com/podcast. And, be sure to subscribe on Apple podcasts or wherever you listen to podcasts. While you’re there, leave a rating or review. We’d really appreciate it.

Brian Colburn – You’ve been listening to Creating Healthier Futures with Alegeus, the industry’s leading benefit administration platform.