Solving a $150 Billion Dollar Problem with Machine Learning with Benjamin Fels, Founder and CEO at macro-eyes

No comments exist

Solving a $50 Billion Dollar Problem with Machine Learning with Benjamin Fels, Founder and CEO at macro-eyes

Hey Outcomes Rocket friends, thanks for tuning in to the podcast once again. As a leader in health care, you have big ideas great products, a story to tell, and are looking for ways to improve your reach and scale your business. However there's one tiny problem. Health care is tough to navigate and the typical sales cycle is low. That's why you should consider starting your own podcast as part of your sales and marketing strategy. At the Outcomes Rocket, I've been able to reach thousands of people every single month that I wouldn't have otherwise been able to reach if I had not started my podcast. Having this organic reach enables me to get the feedback necessary to create a podcast that delivers value that you are looking for. And the same thing goes if you start a podcast for what you could learn from your customers. The best thing about podcasting in healthcare is that we are currently at the ground level, meaning that the number of people in healthcare listening to podcasts is small but growing rapidly. I put together a free checklist for you to check out the steps on what it takes to create your own podcast. You could find that at outcomesrocket.health/podcast. Check it out today and find a new way to leverage the sales, marketing and outcomes of your business. That's outcomesrocket.health/podcast.

Welcome back once again to the outcomes rocket podcast where we reach out with today's most successful and inspiring health leaders. Today I have Benjamin Fels as a guest. He's the founder and CEO at Macro-eyes. He leads teams that innovate and build systems that recognize predictive patterns and macro-eyes, they use machine learning to increase access to care. Their live clinical product and a two leading academic medical center institutions they develop supply chain analytics product for one of the largest value based healthcare systems in the U.S.. Today it's super important that we find ways to apply artificial intelligence and machine learning to healthcare. Because let's face it we're not able to scale the number of patients that need care with a number of providers that can give it. And so that's why that'd be so timely to have Benjamín on the podcast today to talk to us a little bit about what they do it macro-eyes as well as the concept of machine learning and how we can apply it in healthcare so Benjamin. It's a pleasure to have you on the podcast sir.

Thank you. The pleasure's mine.

So Benjamin did I leave anything out in your intro that maybe you wanted to have the listeners know more about.

No no just kidding. That's not I. I don't love talking about myself. So maybe some other pieces of the backstory of Macro-eyes as a company or from my own history of working with my friends will sort of come out through this conversation. So I'd love to just dive into it.

Beautiful Love that you open that loop there. And so you know before we do dive into that I'd love to hear what got you intrigued or interested in the medical sector to begin with.

Well that's a good question because I didn't train as a doctor as a company we've been working in healthcare now since since 2014. But it's I bet it's a pretty unconventional route. So sort of go backwards in time here to answer that. So..

All right.

I've worked on what you could broadly call pattern recognition for many years and I believe in and I pretty sure that most of the people who are listening would agree with me that health care poses the most complex and at the same time most important problems in pattern recognition. This is where pattern recognition matters. So OK now let's go to sort of how I got to this point. So I graduated with a degree in the history of art and the way that I see the history of art as a series of exercises and pattern recognition how one artist perceives, reflects on, interacts with art that came before art from other places. And there's a way to see that transformation refraction of pattern. When I graduated college I then went to go work for a quantitative hedge fund first in Chicago then in London. My job was to predict patterns and data faster more accurately than our competitors. That was it. It was very clearly defined. I mean there's that great beauty in that in that clarity. I led teams that traded global markets 24 hours a day. I led teams also built autonomous agents so what I now know is called machine learning. So basically machines that looked at patterns and data in markets and looked for signals that we might have missed and acted upon. And financial markets as a whole are decades ahead of healthcare in terms of designing and implementing infrastructure for I guess what we could broadly call data driven decision making. So designing systems that learn as reality on the ground changes. So it's everything in financial markets is always responding reacting learning. There's a very powerful very heavily incentivized feedback loop there. Nothing is ever static. So essentially I left this hedge fund to found a company so that I could take this approach this mindset and a bit of new ideas for technology to where it is needed most. And we believe that's healthcare. And again this concept that. Pattern recognition the ability to detect patterns that are meaningful and large amounts of data where that is the most important is in both the delivery of care and the practice of care and I'll just speak a little bit about some of my colleagues in this company because it's certainly certainly not me doing this all on my own. So I found it with two other people a chief design officer and a chief AI officer. Maybe we'll get a little bit later in this conversation to why I think that mix is important. So Sebastian Cowper's and Sebastian is our chief design officer and he has had a really a career long commitment to thinking through how healthcare organizations can best use data. And this is from huge global pharmaceutical companies to one of the first personal health data dashboards which he designed for a company now more than 10 years ago and surprise our lead AI officers since read is a world renowned expert in large scale machine learning and optimization and a professor at MIT. So I hope I answered that question about sort of how how I got to healthcare.

Absolutely. And you've now focused on this area. You've got a phenomenal executive leadership team there at your company. You're tackling Data in Healthcare and making insights so as it relates to that Benjamin, what would you say a hot topic that needs to be on every medical leaders agenda today and how are you guys approaching it?

So I gave away this answer a bit already and I hope this is the answer that many other leaders in healthcare would come to but absolutely machine learning or AI and me and maybe to talk a little bit later about the distinction between those two. But again this is medicine is pattern recognition and the delivery of care should be pattern recognition at scale, at speed. And the thing that is so unique about healthcare as an industry is that it is unparalleled in the richness of the data that it holds that describes almost every one of us in incredible detail, meaningful clarity and no other domain has such an impact on human life and has this incredible picture of each one of us. And I mean all of us have experienced this for better or for worse every time we go to see a doctor or nurse. They spend a lot of time mentoring in data and speaking very, very, very broadly and I'm going to ruffle some feathers here. Very little of that data is effectively put to use to build what we could again say very broadly our learning systems for help very little of that data is learned from in a systematic structural way to personalize care to make care more efficient. And I see absolutely no reason why that should continue to be the case.

Love your passion for this. Benjamin and I agree. You know a lot of the data that we shovel into EMR is stays shoveled into silos. That is not accessible by anybody outside of the system. Definitely a problem right. And so you guys work to offer this solution to health care. You've decided on the macro-eyes. So can you give us a little bit more detail about what macro-eyes is focused on and what's the problem what's the solution you guys are providing?

Sure. So I'll tell you a little bit of the evolution of the company and also the problem that we're most focused on today. So we founded this company in 2014 and we spent a number of years refining and deploying core machine learning at a leading academic medical center in York City at Stanford at one of the largest health systems in the United States and at a number of federally qualified health centers across the country.

And what came out of this experience was a couple of a very robust technology for understanding patient behavior and understanding patient behavior multi-dimensional and what I mean by that is at many points in healthcare, the understanding of the patient is very or at least the classification of the patients is very limited. Fifteen year old male diabetic but there are hundreds if not thousands of other data points in dimensions which are going to inform both the care that is most appropriate for that patients and how we should think about risk and also opportunities and our expertise is that ability to build these very rich pictures of patients in time and the other important thing that came out of this experience largely focused on clinical decision support during these years were working with physicians, physicians scientists to answer clinical questions and from a business perspective probably the most important piece of this is that we we've got an understanding of healthcare as a business and as everybody listening here knows it's a very complicated business it's very difficult to to understand. And I think we got an understanding of the problems that are solvable and the problems that are less solvable particularly for a small company like ours and one of those problems that came to us again and again and again and from many different perspective this is really twofold. One schedules that don't work schedules that aren't predictable a day that a provider has which is chaotic. So it's a balance between having 10 patients in the waiting room all waiting for the same slot and other periods of the day where four of the five patients who were scheduled don't show up. So it's it's sort of feast or famine and there's a significant financial impact to that. There is a clinical impact that this is important to patients, it's important to administrators, it's important to physicians. Scheduling is really the front door to care. And our own response to that was to spend the last. Now more than a year developing and implementing and refining a product called Sybel and Sybel is software for intelligent patient schedule and what that means is that Sybel identifies when each patient is most likely to show for an appointment and uses that insight to build a better schedule, a schedule that is more predictable, a schedule that increases access to care and a schedule that reduces the number of times in the day when there are expensive gaps in the schedule and reduces those periods in the day when there are many patients waiting and waiting and waiting in the waiting room because they've all been booked for the same timeslot.

That's fascinating and through the work that you've done you found that this is one of the biggest problems. And you're right it's definitely a huge issue. And so you've deployed a solution to help fight this problem. Intelligent scheduling.

Yes. Yeah. And I want to emphasize and this is sort of one of the things that drives us as a company and so when all of us think about what machine learning, artificial intelligence, innovation in healthcare probably what comes to mind, Robots, self driving cars, things happening on Mars. And one of the strong beliefs in our company and that I really want to emphasize is that where innovation is both needed the most and has the greatest likelihood of actually making an impact are these issues at the very foundations. Of care issues that touch operations supply chain scheduling. Think of that as this is the roads and the airport of healthcare right. This is the core building blocks that when they don't work the whole system suffers. And when they work effectively now you have a base that you can build on and you can build something which is much more efficient. And we would argue also were personalized and enables both better access to care and more personalized care.

And I love that you've focused your efforts in this very niche area that frankly a lot of health systems health executives I mean if you're listening to this you're probably like yeah I'm definitely struggling with this. It's a billion dollar problem. There's a big loss of revenue due to people not showing up to their appointments. And so I think it's interesting that you guys decided to just niche down to this particular pain point.

Yeah and part of it has to do with again this from our perspective and this is what really, really exciting this is a problem that we believe is solvable. And again I'm gonna say maybe some some controversial things here but there are a number of issues in healthcare that particularly when you're new to healthcare as an entrepreneur and you look around and you see things that just don't make any sense. And if you're looking at that from an entrepreneurial mindset. These are why should start a company to address this and this and this and each one of these is maybe 50 billion dollar problem. But I think if you are humble and you observe carefully you start to notice that many of those issues you cannot solve them as an entrepreneur even if you are enormously successful because perhaps at some level they are the result of policy or they are the result of a certain structure that's in place. And you know we could argue whether that's a good structure or a bad structure but it's very often something that you cannot move even if you are immensely successful as a company. And what our job as a startup and as a startup that wants to make an immediate impact and I would argue that that's the job of every startup. Our job is to find those points in health care where we can affect change we can push the system towards operating better operating more efficiently giving greater access to care to patients because that's that's what we get excited about. Right we want to do something where we can have an impact.

Absolutely. And listeners by the way if you're curious if Benjamin has has hooked you at this point which definitely probably has hit pause and go to gosibyl.com. You'll see a little quick video on what the software does for your scheduling. So it's G O S I B Y L.com. Check them out. They're definitely doing some very intriguing things to use A.I. to help you keep those patients in those slots that are scheduled for whether they show up or not. So this is fantastic work that you guys are up to here. Benjamin looks like you guys even received Grand Challenges exploration grant to deploy a version of it in East Africa by the Bill and Melinda Gates Foundation.

Yeah that's really really exciting. So I'll talk a little bit about that. So we described here that this evolution. You know we we spent years in where we're up close to clinical questions. Our core technology has analyzed several million medical records and has learned from all of this and then out of this came this manifestation of that this product which uses insight into patient behavior to build a better schedule to better predict demand and to build better schedule so you can make best use of existing resources and offer the basis of that. We were awarded this very prestigious funding from the Bill and Melinda Gates Foundation and USA which is development and its government to design the first predictive supply chain for vaccines and at a certain level. And bear with me here as I as I explain this I see these really remarkable similarities. So our aim with this work is to increase the coverage so increase the number of children who can access vaccines and significantly cut the amount of vaccine wastage. And this is also a many many many many billion dollar a year problem. And it is also a problem which you can measure in terms of human lives just like in the United States where if you have to wait weeks and months to access care because the scheduling doesn't work and if you are particularly ill your illness will become far more grave with that period of time. So if a child is brought to a facility and they've run out of that vaccine that's an opportunity that you might lose forever. So our job is to analyze data which describes these different facilities and use that to predict exactly the right amount of each type of vaccine to be delivered to each clinic. No more and no less. Because if you deliver too many vaccines what happens is you're significantly increasing the likelihood that there will be wasted vaccines are very very fragile. They live in these delicate glass files they have to live uninterrupted in a very very narrow temperature range there just really easily breakable. You deliver too few vaccines to a facility and then you have this issue of people are coming traveling to this facility to be vaccinated and you have to turn them away. So this is really a case of get where you want to exactly nail demand, you want to get a perfect sense of how many people are going to show up if you can get that right. You can make best use of existing resources.

Yeah you know it's super interesting project and talk about hypersensitive matter with the vaccines. I mean you guys are definitely doing some cool work Benjamin. Scheduling Vaccines is the Bill Melinda Gates Foundation. This is super cool stuff that is making an impact. So I feel like as health leaders, we often learn more from our setbacks than the things that we've done right. And if you could just share one of the setbacks that you guys had and much you learned from it to make you guys stronger?

Sure. So I'll go again back in time to the beginning of this company. And our first customer was Stanford. And at the same time we were working with a leading academic medical center in New York City and in both at both institutions and in both cases these are these are world famous organizations. These are these are places that are at the very cutting edge of care and in both cases we're working with with brilliant physicians and physicians scientists and and I'm going to describe a bit sort of our again our our founding thesis and how we changed that idea but our basic concept again is that medicine is pattern recognition. So let's pretend here that I'm a patient and I walk into your office and you're a brilliant physician and you look at me and you look at my chart and clicking away your brain is this a version of what we call patient similarity. So where have I seen a patient like Benjamin before have I read in the literature about a case like this. As a colleague in the hallway mentioned you know I saw a patient three weeks ago with this this this and this and then that happened. And that is going to guide almost every point on the journey a pair prognosis diagnosis a notion of risk a notion of which medication to prescribe when. And our founding idea was let's bring scale and depth and muscle to this. This pattern recognition that the good doctors do and let's take it across every record that sits in that organization so they can get the best sense of when have we seen a patient like this before and what happens and what we underestimated. To a great degree is the extent to which physicians have been forced to deal with technology that asks an enormous amount of them and delivers almost nothing and because of that. Rightfully so. And I will support them the whole way here. Physicians are overwhelmingly wary of working with yet another piece of technology that will guide them during this complex process of clinical decision making and I think the other thing that we learned is that if we want to support that process of decision making we need to earn trust and that trust has to be built up incrementally slowly. It might take years to build that trust. Being inside that institution and we believe innovating and improving on the very foundations of care because the other thing that we like about the points where we have focused is that very often there are very clear metrics that you can present and you can point to the impact again and have a very clear way and that helps to build trust. So just to reiterate here I think that the big mistake we made is we thought well this is transformative technology. Of course physicians will want to use that and we just didn't understand enough about the day to day reality of what physicians have to do with technology and how how much they dislike that interaction. And again speaking very very broadly for all of us on the side of the table who are building technology for health care. This is something that we have to think very carefully about how do we gain that trust and how do we deal with a community of users who are skeptical and again rightfully so from the very beginning.

And you've brought up some great points. And I've done over 400 interviews now and the topic of adoption you know and getting clinicians to adopt a technology you it just continues to resurface. And it sounds like after the journey of figuring that out you've really gotten a clear idea of how to you know number one address it and number two just meeting them where they are. And so what would you say the best way to do that is today. Have you guys done it. It's a problem for a lot of people.

That's a good point. So I would say I mean part of our answer to that is we are trying to make that day of every physician better, easier and largely by working in the background. So our product Sibyl is not something that a physician necessarily will interact with. Now I've worked with many physicians who are extremely involved in scheduling because it impacts every second of their day and they have very strong opinions about how that schedule should be structured. But our job I think first and foremost from a physician perspective is to make their day predictable and I cannot overemphasize how important that is. So let's put aside for a second the financial actual impact. And again this is a hundred and fifty billion dollars a year are lost in the U.S. alone just scheduling that doesn't work. That's a big number that just so happens to be what I believe to be the global cost.

I knows it was in the billions that it knows 150 billion it's High.

This is more than what the world spends to care for patients with cancer. So we're we're talking big, big, big numbers here. And let's put that aside for a second and let's put aside the access to care piece and let's just think about how a physician goes through their day and imagine trying to be efficient. And imagine trying to do your best which is what every provider of care wants to do when if you have for instance five appointments scheduled for every day, you have no idea which patients are actually going to show up which means how do you prepare for that. Secondly that means if we back out from that how do you allocate additional resources when we should. What types of support when should which types of nurses be available. Which of the supply chain kick-in in winter certain goods necessary and just that feeling of going to work every day and not knowing how many people are going to walk through that door is very difficult to work with. And that's something that we learned an enormous amount about during this experience of deploying and working with different organizations and that one of the things that came up over and over and over and over again is that this is literally driving our providers crazy. In one of the things that they talk about all the time is just how destabilizing and difficult it is to do your job when you have no sense of what is going to happen in the next day, hour, two days. And the more predictable you make that schedule, the more in control of their day providers are the better they can prepare and the more efficient the whole system becomes.

Love it. You definitely honed in in a big way. Benjamin so kudos for you and your team for being so hyper focused. It's definitely what's needed to move the needle in this space for sure. So within all of the things that you're doing today what would you say an exciting project or focus is for you.

I'm very excited about these early deployments of civil just because every time we work with an organization we learn new things and this sounds cliche to say but it's absolutely true. I mean one of we recently sat with some schedulers in Alaska and they shared their scheduling care that is very complex in many levels. They're scheduling sequential care they're scheduling care oftentimes for a family because the distance traveled to the clinic is so great that maybe everybody is going to get in the car or try to see this physical therapist this occupational therapist on the same day. And so they have to find maybe four or five different types of appointments which are just certain order and balance that with the availability of providers. And they shared with us that this can take some 45 minutes to put together a schedule and we didn't build our software for that used case we didn't think about oh well this is going to save schedulers time. Our goal is to increase access to care for patients and build a schedule which maximizes utilization. But we realized well if we can cut that down to a minute for a scheduler that means the schedule will have more time and that means the scheduler could then spend his time her time in what we believe is the best use which is engaging directly with patients. So imagine that the scheduler could then pick up a phone and personally call the patients who they're the most concerned about not showing up and a personal phone call is so much more meaningful than getting a text message before moving an automated reminder or an e-mail. There's just as human beings we respond very strongly to human beings. And that's something that you can't do if you're spending 45 minutes to schedule an appointment.

It's a great call out and the journey is exciting and Benjamin if people want to engage with your software if they're curious about it they're listening right now and you're like OK just tell me how I could get involved. Where do they go?

They should definitely go to the website you just mentioned, gosibyl.com and send me a note. So my first name benjamin@micro-eyes.com, and I'd love to have a conversation.

Love that. So folks the website. Benjamin's email. All those things. I'll have them for you here on the show notes. Just go to outcomesrocket.health/sibyl and sibyl is S I B Y L. So outcomesrocket.health/sibyl and you'll find a way to get a hold of Benjamin and get started with this phenomenal scheduling platform. Benjamin this has been a blast. Time flies when you're having fun. Would love if you could just share a closing thought and then the best place where the listeners could follow the work that you're doing.

Well I think that if I could share a closing thought all a little bit more broadly to just sort of what we've observed with working with a number of health care organizations. Big, small and the lesson that I'd love to get across to health care leaders is embrace risk. So understand risk. Think about it carefully. But embrace the right type of risk. And there are very often opportunities to work with innovative companies like ours and we're certainly not the only one out there where the downside is very limited. But the upside is almost immeasurable and maybe I'm sort of putting back on my hat from when I used to trade derivatives but that's exactly the type of risk you want. Right. You know the worst that can happen here is nothing and that's it right. Yeah nothing collapses. No one gets fired. But the best that could happen here is transformative. Right it's a transformation of how we deliver care. And I would love to see more of that intelligent risk taking.

I love that call out and folks I know we've we've talked a lot about Sibyl at gosibyl.com but also check out Benjamin's company macro-eyes macro-eyes.com. You'll see some of this thought process that he has shared. He's a thought leader in this space applying what he did with derivatives into health care. We're always trying to manage risk and I think it's a good opportunity for you learn the philosophy that him and his leadership team are leading with here. So check them out macro-eyes.com. But again you could get that link and all the rest just keep it simple. Go to outcomesrocket.health/sibyl, S I B Y L and you'll find everything there. Benjamin, truly appreciate the time you've carved out for us and we're excited for you and we're excited to stay in touch.

Thank you. Thank you for a wonderful conversation.

Hey Outcomes Rocket friends, thanks for tuning in to the podcast once again. As a leader in health care, you have big ideas great products, a story to tell, and are looking for ways to improve your reach and scale your business. However there's one tiny problem. Health care is tough to navigate and the typical sales cycle is low. That's why you should consider starting your own podcast as part of your sales and marketing strategy. At the Outcomes Rocket, I've been able to reach thousands of people every single month that I wouldn't have otherwise been able to reach if I had not started my podcast. Having this organic reach enables me to get the feedback necessary to create a podcast that delivers value that you are looking for. And the same thing goes if you start a podcast for what you could learn from your customers. The best thing about podcasting in healthcare is that we are currently at the ground level, meaning that the number of people in healthcare listening to podcasts is small but growing rapidly. I put together a free checklist for you to check out the steps on what it takes to create your own podcast. You could find that at outcomesrocket.health/podcast. Check it out today and find a new way to leverage the sales, marketing and outcomes of your business. That's outcomesrocket.health/podcast.

Automatically convert audio to text with Sonix

 

Best Way to Contact Benjamin:

Email: benjamin@micro-eyes.com

Mentioned Links:

gosibyl.com

MACRO-EYES HEALTH

Episode Sponsor:

Leave a Reply

Your email address will not be published. Required fields are marked *