What role will artificial intelligence play in healthcare? How can smartphones make healthcare more accessible? And what’s the future of tech giants such as Amazon and Google in the world of health?
Those are a few of the issues we discussed with a panel of health tech leaders during a special live recording to launch Season 3 of GeekWire’s Health Podcast, on location at Premera Blue Cross, the sponsor of this new season of the show. The discussion set the stage for the topics we’ll be exploring this season, looking at the many ways technology is changing the healthcare landscape.
We were joined by Ranjani Ramamurthy, principal physician scientist at Microsoft, who has a background in both medicine and engineering, as an MD and a computer scientist; Oron Afek, CEO and co-founder at healthcare technology startup Vim; and Robbie Cape, CEO and co-founder of startup 98point6, a Seattle-based entrepreneur who sold his previous company, Cozi, to Time Inc.
The discussion was kicked off by Dr. John Espinola, executive vice president of healthcare services for Premera Blue Cross.
Listen to the episode below, continue reading for highlights, and watch the full video below. We’ll have new episodes every month during the upcoming season, starting this month. See more episodes here, and subscribe to GeekWire Health Tech in your favorite podcast app to catch future shows.
Todd Bishop: Can you each give us your elevator pitch for what you’re doing?
Robbie Cape, 98point6: We are addressing the primary care crisis in the United States. On the one hand, we acknowledge that there’s going to be a shortage of literally 20,000 physicians by the year 2020, and that number is going to rise to 30,000 by the year 2025. On the other hand, we all know here, especially that primary care saves lives. In fact, if you introduce a single new primary care physician into a population of 10,000 people, you will reduce the mortality rate of that population by 5.3 percent.
A single individual who has a relationship with a primary care physician is 5 percent less likely to be hospitalized. They are 6 percent less likely to end up in surgery, and they will save on the order of 30 percent on their healthcare costs over the course of their lives.
When we look at these two competing statistics, what we clearly see is that we have to introduce dramatically, dramatically more primary care physicians into our ecosystem. We’re doing that with technology. We are delivering an on-demand primary care service that is supported by artificial intelligence and deep learning that enables our physicians to treat on the order of 25,000 patients in their panel … Any of you could be sitting here today, download 98point6 and conduct a visit with a 98point6 physician, including our automated assistants.
Ranjani Ramamurthy, Microsoft: I’m a part of Microsoft Healthcare, as a part of the larger clinical sensing and analytics group. Within Microsoft Healthcare, we have two pillars, and I am super proud of being a part of the pillar where our primary focus is empowering those who are in the frontlines of healthcare.
We think that building technologies that are frictionless, seamless, and amplify and assist those who already know what to do well. We’re not trying to replace any caregiver or any provider. We’re just saying, “Let us help you, and let’s do it in a way where it really provides value to you.”
Personally, I lead a team called Empower MD, where we’re building an intelligent virtual assistant for physicians at point of care. What this means, as all of you here know, physicians have a challenging relationship with technology. The biggest ask we’ve had has come from physicians who’ve wanted us to really leverage speech and natural language processing technologies to give them an intelligent scribe or a digital scribe at point of care.
Oron, I have to tell you, you’re a little stealthy, Vim is. It’s hard to find a lot of information on what you’re doing. My impression is that it’s a little bit like an Apple Wallet for healthcare. You’re bringing together lots of different services into a central place.
Oron Afek, Vim: I think what Apple brought to the world is, for the first time ever, integrated experience between software and hardware. When I’m thinking about selling the partnership … with leading high-value providers, the ones who are really good in delivering better healthcare at lower costs, more efficiency, I think about it like Apple connecting the payer and the provider seamlessly, creating liberational data, seamless referrals, prior authorizations, scheduling, those kind of things …
I think it’s going to be trying to focus more on the member experience, which is a good byproduct of that, if everything is integrated.
Where is the seam today between technology and humanity in the delivery of care, in access to care?
Cape: I’d like to share a couple of statistics first, which blew my mind away as I learned them because we often get asked, “Are consumers ready for the digitization of care?”
Let’s just talk about Americans. 80 percent of Americans begin each and every healthcare journey on Google.com. Five percent, one in every 20 searches that happens on Google, is related to healthcare. … These are phenomenal statistics. What this tells us is that not only are people ready, but they’re doing it. They’re doing it. They are starting on Google. Who is our greatest competitor to fixing the healthcare system? It’s Google search.
And why are people going to Google search first? They’re going because one, it’s in their pocket. They can go there when they’re standing in line at the grocery store … And guess what else? That visit to Google is free.
It’s very simple. People are quite predictable. When the technology is delivered to them and it’s priced right and it’s delivered in a way that delights them, they use it … Our big issue is awareness. It is not willingness. When you put it in their pocket and you make the price right and for us, our perspective is of course you have to pay for the service, but the marginal cost of a single visit has to be zero. In other words, there is never a financial trade off on the margin to check in with a doctor. And when you do that, you can focus on your health.
So, can an artificially intelligent agent deliver the care … is it possible that AI agent will actually be making decisions that are today made by physicians?
Ramamurthy: It could be assisting physicians in making decisions, and I can give you an example of a use case that was done by Microsoft with Children’s Mercy Hospital. It’s called the CHAMP Program … There are babies born with hypoplastic left-heart syndrome. These are children who have very pressing medical needs and usually have to wait for a period of time before they get surgery.
Now, with Children’s Mercy, Microsoft had this project where they actually gave the patients’ families a Surface tablet where they could actually log the children’s vitals. The temperature as well as oxygen saturation. Put this into the cloud so that their providers could be notified earlier on when the children were not thriving as they should. And they found this to be a huge success, because in the past what parents had to was to do take this big ring binder and document all this information and then call the nurse manager at the end of the week and give her the information.
But not replace, augment?
Ramamurthy: Not replace, augment … the physician is always in control.
Does AI play a role in what Vim is doing?
Afek: I think there’s obviously a lot of things that can be done to fix some of the engagement paradigm using AI. And I’ll just give a few examples of some of the things we’re working on.
When I’m thinking about healthcare, I’m thinking about healthcare as any other engagement paradigm that we’re familiar with. So let’s think about Uber and Amazon. From a behavioral perspective, I like to think about it as a two-axis diagram. On one axis you look into the frequency of use. On the other axis you have the ticket price. Uber is very high frequency, very low ticket price. Amazon somehow high frequency, a bit kind of bigger ticket price depending on what you’re buying.
Healthcare is very high ticket price, low frequency, again engagement paradigm. And we see it in Silicon Valley a lot where there are many well-funded companies building consumer engagement tools that has brilliant tech behind them but lack the engagement. As I think Robbie said, everyone’s just going to Google. I think the way to fix it would be using machine learning and AI to infer when would the member need me, instead of educating the member to go to my website when they need me. That’s going to be really expensive.
Instead, if we can infer when does the member need us by predicting or by ingesting real time feed and understanding behavior, then I think we can more successfully engage in that journey. And that’s just one of the things that we’re working on at Vim that will maybe put us in that quartile of AI and machine learning.
Cape: On the primary care front, 80 percent of cases fall into 10 pretty straight-forward categories of diagnoses, in terms of what’s coming in the front door. So it really is true, that common problems are common. That’s exactly where AI tends to work really, really well. Now you have to build a system that is capable of augmenting the physician, because in our current regulatory environment — and I think it’s important for us to talk about regulatory in the context of all of this — in the current regulatory environment, you have to be a licensed physician in order to practice medicine. To deliver personalized advice of any kind to an individual in the United States, you need to have a medical license. And people are pushing those limits. I think that’s a mistake. I believe that so long as that’s what the regulation says and I think that’s what the regulation says for some very good reasons, we need to keep a physician behind every decision that gets made.
In time, some of those regulations will change for very specific cases. Like, you could imagine for a flu diagnosis .. We might be able to imagine in five or eight years from now, the FDA approving an algorithm that is built on top of machine learning, that is capable of diagnosing the flu and prescribing the right medication in the right circumstances.
Oron, you use the analogy of Uber and Lyft. So I have to ask, as we’re talking about this, when I get into a Lyft, it is generally a driver that I never met before and I only know by his or her rating in the app. Is there a risk that we are going to be transactionalizing healthcare in the same way? What is the risk to the relationship between the doctor and the patient in this new world?
Afek: I think there’s a huge risk. It’s pretty easy to look into how millennials how accessing healthcare today. More retail clinics .. And I think it’s going to just continue and evolve because people just want to have access over a relationship. And I think the risk here is that single-time encounter might be good for certain people, if they are very determined and they are going to go and research everything and are going to be smart.
The majority of people still need guidance and I think that’s not going to change very fast. They need to have a trusted advisor that can lead them to the right place. I think me not having that kind of care champion but trusting the single time encounter can eventually lead to more fragmentation and more downstream cost that is going to be hard to prevent.
Cape: What we want to do, and this is a little controversial and is probably the hardest thing that we have to do at 98point6, but we’re incredibly focused on it, is build a relationship between our patients and the 98point6 service. Not the physician who happens to be behind the delivery of care, but the service. The service knows them. The service understands them. The service knows why they came to the clinic the last time. The service is reaching out to them proactively, to engage them in their health. The service understands who their family members are and what that might mean for their health moving forward.
Ramamurthy: That’s what I was thinking about in terms of how technology can be built, to mirror some of the best practices that you see for care delivery … It’s important, because there’s enough research that shows that if the patient knows that they’ve made this connection with the provider, they are much more likely to actually generate the plan that they will actually adhere to … Why can’t we take the best practices we see in clinic and feed that into the technologies that we build?
Currently we have Seattle area tech giant, Amazon, partnering with Berkshire Hathaway and JPMorgan … what are the prospects for Amazon specifically making a serious change in healthcare beyond its own employees?
Cape: I think Amazon is probably the single company in the world that is best positioned to have a deep, deep impact on the industry as we know it today, and there are two specific areas. One that we haven’t talked about at all, but I literally refer to our prescription crisis as an utter fiasco. If you dive into what’s happening in prescriptions, we literally have for-profit organizations — I believe, and this is just my editorial, slightly overdramatized view — almost stealing from consumers.
They’ve acquired PillPack.
Cape: I think that that is the tip of the iceberg to literally fixing the prescription problem in the United States, and then around the world. And that will have an enormous impact not just on access, but also on cost. Second step — also this is speculation on my part — I believe that the next step related to cost savings and health is diagnostics, where I also believe that Amazon can have considerable impact.
Look at just one specific problem, the flu. The flu costs the United States 10 billion dollars a year. One ailment. Amazon could, by virtue of operationalizing the distribution and the testing associated with the flu, and then operationalizing the delivery of the medication associated with the flu, they could likely cut out a material percent of that cost. People literally would not need to leave their home to be tested for the flu, and then to get their medication for the flu. And, I’m just talking about one little issue. Across those two areas, the impact could just be phenomenal.
We’ve seen smartphones do so many different things, from diagnosis, to monitoring, to communication. Where should the line be drawn? What are the limits today, and what will we see in two, three years in terms of healthcare delivery, access on our smartphones?
Ramamurthy: In my opinion, using mobile apps to track and manage chronic disease conditions is definitely something that’s happening right now. What I think about when I see this profusion of mobile apps that are targeted to the patients, I look at it in terms of, one, does it inform care that a physician gives, or is it just generating a lot of noise for the physician? And two, if it does inform care, how do you integrate it into the care delivery pipeline such that it actually betters outcomes for patients?
So, I think there’s definitely a place for it. I’ve seen some very interesting ones. And, again, some of them are really transferring some of the best practices from clinic into the app, and they’re interesting but I think with a grain of salt in that it has to inform care.
Afek: I think it’s easy to look into other insurance industries, like car insurance industry, and how they’re leveraging data from smartphones to give me feedback about how I drive, and perhaps also price my premium in accordance to the feedback I’m getting there … I think whether it’s going to be on a mobile phone, or other implements that I’m going to be wearing in the next 10 years — probably three years is still mobile phone — I think this data is going to be invaluable, and could be leveraged in healthcare.
Cape: As a technology guy, I’d say my perspective on this is a little bit controversial. The data that is generated by these devices is complex, and at the end of the day a lot of the data ends up implying what is essentially personalized healthcare advice to the individuals who are perceiving that data. And, personalized healthcare advice is the definition of the practice of medicine. You’re supposed to have a license to practice medicine.
I think that when that data is being generated, and is ultimately getting reviewed, and understood, and interpreted, and counseled with the expertise of a Board Certified physician, it’s incredibly interesting. I mean, we’re talking about taking diagnostic devices, and essentially democratizing them, and generating that data for doctors to understand. That is phenomenal. However, if you take that data, and you kinda-sorta make your way around the rules, and end up delivering advice without that license to practice medicine, I think you are ultimately playing with fire.
Yes, certainly, in 40, 50, 60 percent of cases, you might be empowering people to have some insight that they might not have had before, but it’s not really any different than the democratization of healthcare information on Google. But what you end up leading to is people who are not qualified to interpret that data are interpreting it, and coming to conclusions about that data that they should not be coming to. And, I think that’s dangerous.
For physicians or for consumers of healthcare, what does the world look like in 10 years if things change in the way that you hope they do?
Ramamurthy: In short, I’d say physicians would go back, or clinicians, any provider would go back to doing what it is that they know to do best, which is taking care of patients. And, I think that that’s just it. Just taking care of patients, patient-centric care, that’s why they went to medical school, that’s why they went to nursing school. Let them do what they know to do best.
Afek: I think the business model would change significantly [over the] next 10 years. From a reimbursement perspective, on the physician side’s going be much more value driven … I’m talking about large portions of the book of business is going to be driven mainly by performance, which in my opinion will be a mixture between fee-for-service or enhanced fee for service with performance.
I think what we would start to see on the patient side is more personalized care plans from A to Z. So, if I’m a diabetic patient, I would have a health plan for diabetic patients, and I would have specific centers of excellence that would treat my conditions. And, I’m not going to just buy the same product other people are doing. So, I think there’s going be a way to differentiate, and create very very personalized insurance products.
Cape: It’s a combination of what’s been already talked about. It’s physicians who are practicing at the top of their license, who love their jobs day in and day out. It’s patients who are getting what they need, exactly when they need it. So, what that translates into is, for the easy stuff, that technology is playing a very substantial role. And, that there is this transparency across the system that enables those patients, in the event that their care needs to get more complicated, and hence more expensive, there’s a transparency across the system that enables them to get that care in the place that’s going to lead to the best outcome, where the quality is highest, and the delivery is the most affordable. Everyone will be happy.
Editing by Frank Catalano. Video by Bootstrapper Studios. Note: The name of the CHAMP program has been corrected since this story was published.