Talking Healthcare Analytics with Wout Brusselaers, CEO of Deep 6 AI
I recently had the pleasure to sit down with Wout Brusselaers, founder and CEO of South by Southwest Accelerator Pitch Event Winner Deep 6 AI, which at the time of recording was known as Deep 6 Analytics. Deep 6 AI offers a software solution based on natural language processing (NLP) that matches complex clinical trial criteria with qualifying patients, and we discuss the landscape of healthcare analytics and what it takes to be successful as a start-up in healthcare. Listen below, or scroll down to read the transcript.
DeciBio is a strategy consulting and market intelligence firm focused on clinical diagnostics and life science research tools. Learn more about our services and products, or send us an email at [email protected].
Author: Anthony DeBenedetti, Associate at DeciBio Consulting, LLC
Connect with Anthony on LinkedIn
Disclaimer: Companies listed above may be DeciBio clients and/or customers
Thanks for listening to the DeciBio podcast. My name is Anthony DeBenedetti, I’m an Associate at DeciBio, and I’ll be your host. Today I’m joined by Wout.
Wout Brusselaers, is that how you say your last name?
That is better than most people say it, so I’ll take it.
What’s the proper pronunciation?
Wout was perfect, Brusselaers is the last name.
I definitely was not going to get that.
Wout is the CEO and co-founder of Deep 6 Analytics, a healthcare analytics company that matches clinical trial criteria with qualifying patients. Wout has an interesting background.
I have a background in philosophy, economics, and math. I started my career as a diplomat in the Middle East. I joined McKinsey and Company, I ran a little extreme sports start-up community, became the head of international relations for a global security firm.
It was then that he met his co-founder and Chief Data Scientist, Brian Dolan, and they got started on what would become Deep 6 Analytics. I’ll let Wout take it from here.
We jointly applied to a US government contest to deal at peta-scale data levels with real-time decision making. We won that contest, beating out IBM Watson, HP Autonomy, some other big names, which was kind of nice. And that led to contracts with the US intelligence community, so for about two years they were our major clients. We developed a really cool platform in what is arguably the most complex and challenging data environment in the world. So, that’s how we got started, we decided that wasn’t really our thing, we had too rotten a sense of humor to deal with the US government. The pace was a little bit too slow, so we moved into healthcare.
And you said that transition, you were mentioning before we started recording, was just within the within the past year or so, that you started getting into healthcare.
Absolutely. In 2016 we applied to the TechStars Accelerator with Cedar Sinai, and that really helped us pivot. We had built a prototype before, so we kind of leveraged all the technology we had built for the US intelligence community and turned it into some kind of prototype which was more in the realm of decision-making support. Our initial idea was that data in healthcare is fragmented and is years behind the curve. The US government had spent about 160 billion dollars on digitizing all the medical records into EMRs, and we felt like at the end of that $160B, there was no big data in healthcare. What you have is lots of little data locked and fragmented in all of these different medical records, but they don’t talk to each other, they don’t inform decisions. There’s just no way that really changes the way that healthcare professionals work today, and that’s a missed opportunity.
Yeah. So, what would you say is the short-term goal of Deep 6 Analytics?
So right now, what we did is we built an alternative data model that allowed us to use NLP, natural language processing, to mine all of this structured and unstructured data for patients, all the clinical data, and we turn it into a graph, like a multi-dimensional vector that has tens of thousands of data points with all the symptoms, diagnoses, outcomes, treatments, social demographic information, omics information, lifestyle choices, everything. And we thought first, we’ll match it with other patients for clinical decision support. That didn’t really fly, so instead we saw a big opportunity that has a much shorter ROI for clients, and that is matching those vectors patients against the eligibility criteria for clinical trials, which was a huge opportunity. And us being the village idiots of healthcare at the time, we had never heard of that. “Oh, is that a problem? How big of a problem?” And then we looked and said, “Well, in 2015 that 2 million US patients participated in clinical trials. What does that really mean?” Well, just all of the listed clinical trials that are posted on clinicaltrials.gov, which is a federal website where companies and sponsors have to post their trials, just for all of the trials from that year they were trying to recruit 6 million patients. So barely 1/3 of all the patients that were required just to get all of those already approved and started clinical trials to get those going. You had a 60% gap of patients, actually closer to 70%. So when we started thinking about it, “That is a real big opportunity.” And as you know, a clinical trial typically for a major drug can take up to 3.5 years to get it done across different phases. 3.5 years for drug that’s under patent, where patients expire in 20 years, if you can shave off one year of that, you can make that drug or the company that’s sponsoring that drug an additional 5 billion dollars. So the stakes are really high, and not to mention human lives. If you delay bringing a new drug to market, there are thousands of people that are effected. So that’s our goal, if we can make a difference there, I’ll be very happy.
So, you have very specific criteria that you put in, in laymen’s terms, into your proprietary search engine, and then it gives you a list of patients that would qualify for that trial?
Yeah, and we even parse all of the criteria. So literally you just copy and paste your criteria, you probably do a couple of clicks, and within minutes you see your results. We will show you patients that match very complex clinical trial criteria literally within minutes.
That’s what we said.
That’s very cool.
So what would say were some of the greatest challenges you had shifting from US government applications of this platform to healthcare-oriented applications?
There’s a few. The original use case, if you try to find a parallel to the intelligence community, is find bad actor, remove liver, and feed to the dogs. In this case, it’s find sick patient, remove liver, and find a transplant. So, there are similarities but it’s also very different in the end. That’s a joke, by the way, we never found bad actors and removed their livers. Or at least we wouldn’t talk about it on a podcast. [laughs] One of the issues is that in healthcare things move fairly slowly, but for a good reason. One of the issues that I brought up about clinical trials, the ground truth or the ground fundamental reason for the existence of clinical trials, is that once you’re in healthcare you’re affecting human lives. So everything you do, you have to prove that it works. And in a way, that is our key value proposition in healthcare. We want to speed up innovation in general in healthcare. If you have a new drug, a new device, a new procedure, even if you want to change out the pillows in a hospital, you have to prove that it improves the lives of patients and doesn’t make them worse. That is something that is very unique to healthcare. The good thing about it is that it’s very data-driven. And without going off on too much of a tangent, in our current environment where not much in public life seems to be based on facts anymore or on actual data, it’s nice to have that in healthcare.
Yeah, that’s interesting. I mean there’s obviously a lot of talk about the role of regulations in society today. What are some of the issues that you think other tech start-ups don’t have face that you experience being in a healthcare start-up?
It depends on the industry that you go in. Something very specific to healthcare is a slow decision-making process because everything has to be tested, everything has to be tried. Because the organizations also because of consolidation both on the payer side on the pharma side on the provider side, lots of consolidations, you’re dealing with very big actors. And their decision-making process is slow, it’s just a world of difference from a small start-up. For a start-up, the key truth for the first couple of years probably is that money equals time. Every dollar you have buys you a limited amount of time to keep on trying, to keep going to the next level. For big organizations in healthcare, time is not their issue. Time is not their concern. They want to slow things down so they’re really sure about what they do. That’s the issue. You’re at a fundamental disconnect.
Very much in conflict, yeah.
Absolutely, and that is something — you need to be able to either get funded, which is making a deal with the devil in a way because you’re diluting yourself and you’re now at the mercy of somebody else but it’s probably necessary, or you need to find a model that can have a very quick ROI and a shorter sales cycle.
Is there such a model in healthcare where you can get quick revenue generation and then be able to use that to fund your pipeline dreams?
Apart from robbing drug stores, I’d love to hear the alternatives. [laughs] One of the things that we really focused on and that led us to our current business model was showing quick ROI. We did not want to sell a platform, we wanted to sell an application. That was very important. In a platform, typically you leave it to the user to figure out what to do with your technology and where the benefits are. If you have an application that clearly tells you, “If you use our application, you’re going to find patients in minutes, not in months,” it’s very clear. It’s very simple. We were in the accelerator with our technology, we had a lot of different opportunities, a lot of problems we could have solved. But we really rigorously went through the process of analyzing where we had the quickest return on investment for our users, where we could make the best case that the sooner they get this, the sooner they start either making money from additional revenues or saving lives or saving costs. And in healthcare, strangely enough, that is really important. Money is really important for them. We were very idealistic, but in healthcare money is the mission in many ways. We had to really think about that and show that by working with us, they could cut costs or make more revenues very quickly.
Could you expound on that for a just a second when you say that money matters a lot within healthcare?
Yeah, it’s basically that healthcare is going through a big transition. In the US, me being from Europe with an outsider’s view, the incentives are not so well-aligned. Payers, providers, and pharma all have different incentives. And because of that they’ve created these fiefdoms and these workarounds that are fairly inefficient. That’s one of the main reasons why the cost per capita of healthcare in the US is the highest in the world, and the results are not better than a lot of other industrialized countries, and in many cases, they’re slightly worse. So there are a lot of inefficiencies, there are a lot of ways to do things better, and I feel like fee-for-value rather than fee-for-value is a drive that, if it continues under the current administration, will see the total expenditure on healthcare reduced. As a result, there will be less money going around in healthcare I hope, or there should be. That means that a lot of actors in healthcare have to start saving, they cannot just spend money on everything. Not every doctor can have a full-time assistant, you cannot just charge MRIs and do things all the time. Knowing that, a lot of the actors in healthcare have to think, “Where can we spend money and how can we find alternative ways of revenue?” And it’s interesting that that’s something we were not even as aware of. “Big hospitals are doing billions in dollars, sure they can spend money on good technology.” No, they don’t. They’re very concerned about where they spend their money, what technologies they embrace and what they don’t, because they know there’s lots of money that will not be coming in in the near future.
So then being in your shoes, or in the shoes of any other kind of healthcare analytics solution, do you think that’s a positive or a negative, the fact that there will be potentially less money going around in the future?
Overall, I think it’s a positive, absolutely. I think there is a lot of waste, there’s a lot of inefficiency in healthcare, and I think trimming that down is great. At the same time, there has to be enough room for innovation. People are saving lives every day. Healthcare is one of the most important and critical industries in your life as a consumer and as a patient. You want to make sure they spend the money to get you to the next level and to heal patients. As a start-up person, at the same time, it really puts the emphasis and focus for us on showing that we deliver value, which I think is great. That’s why we spent time when we were in the accelerator on making sure we had a really strong business case and we had a really strong return on investment for our clients. We could say, “If you do this, it really helps you. This is how it helps patients, this is how it helps you cut costs and bring more revenue outside of a fee-for-service model to healthcare.”
I’ll get you back to my conversation with Wout in just a second, but I wanted to say a quick thank you to Javier Suarez for our music. Also, if you have an idea for an episode you’d like to hear in the future, if you’re interested in learning more about DeciBio’s products and services, or if you’d just like to say hello, feel free to shoot us an email at [email protected]. Now, back to the interview.
In terms of healthcare analytics, do you think we’re at a phase as a community where it’s still nascent and we’re looking for these big, huge developments? Or is it going to be more incremental as we move forward and we do need to learn to appreciate the incrementalism?
I think it has a combination of both. I feel like there’s a lot of drivers in digital health that are, like you said, so nascent, a lot of things are still happening. One is pushing a lot of data to the actual patients, to the consumers, so they can own their own data: having portable data, having insight, and helping them become more of an active steward of their health. Today, typically patients are still at the mercy of their providers and their payers. They don’t have the data, they don’t understand the data to make their own decisions. And in digital health and clinical decision support, by sharing things with patients and translating that data into actionable insights, telling them, “This is what’s the matter with you. These are the options that you have. This is how many other patients that are very similar to you have responded to this, these were the options,” it makes them more knowledgeable. It makes them more informed and it starts more of a dialogue between the patient and the doctor. The care decisions are no longer just being imposed upon them.
One thing that I could see potentially being a challenge there is, I could see a scenario wherein ultimately it’s their physician that decide whether or not that data gets shared with the solution-determining platform, and they might be reluctant to engage in that because in some ways they could be working themselves out of a job or turning their job over to an algorithm, so what do you think needs to be done to make sure we’re moving process forward in that sense?
I think you raise a really good issue, and I think there are a lot of miscommunications and misperceptions about that issue. I was on a panel a while ago where some people from big consultancies talked about the future of healthcare, and there were a lot of medical students in the audience, and one very well-informed consultant said, “To all you aspiring doctors, you should all become data scientists first to understand the data.” And I was thinking, “No, that is absolutely not true. That is the wrong approach to data.” Data should become available to people who are not data scientists. It should be really easy to interpret and deal with data, both for physicians and for users. I think technology should remove the barriers between a patient and a physician, and should empower both the patient and physician to have a dialogue again. It should not replace the physician, it should not try to give full control to the patient, but it should make sure you can talk about these decisions and that you have all the data available to make them smart and informed. You will always need a physician.
With Deep 6, have you had the opportunity to interact with any physicians or medical directors, anything like that, hand-on with your solution?
What’s the messaging that you need to give them, so that they see – well I guess yours is a little more focused on clinical trials – but how do you have that conversation with them so that they understand that this isn’t replacing care that you’re giving but it’s ultimately improving the care that patients will receive?
Oh in our case, our users see that immediately, for two reasons. One, we spend a lot of time on our workflows and on our UI to make our tools very intuitive and very easy. It’s not another frustrating EMR system where you have to do drop-down dialogues and click on stuff and spend a lot of time. They immediately see on each page in our tool, it’s one or two clicks and you get to results, so that helps. At the same time, they also see that what we’re doing is actually empowering them; we’re not replacing them. There’s a whole process of validation, and that’s typical for any type of a human intelligence and an artificial intelligence. The best results come when you work together. You’ve probably heard about the whole thing with Kasparov, how he was beaten by Deep Blue. But even a mediocre AI with a good human player can outperform any powerful AI. It’s the same thing in any application. Our tagline for a while has been “human intuition at machine speed and scale.” We always want to empower human intuition and let that be the driver of all the decisions, but just make sure all the information, all the data is available for the user to drive those decisions in seconds.
Right, it’s very much an example of “the whole is greater than the sum of the parts.”
Absolutely, it’s an augmentation, it’s not a replacement.
Is there anything besides the lack of structure in healthcare data that you would say is really limiting the ability to embrace healthcare analytics for improving patient care?
There’s a lot of things. One thing is just the data entry process. You know the paradigm of “junk in, junk out.” When you take data entry for patients and push that into the hands of highly qualified doctors who have better things to do, or of nurses that are also very busy, a lot of data that gets entered into current systems is fragmented. It’s incomplete, or it’s copied and pasted. There’s a lot of junk in there. Finding better ways interact with systems so the data entry becomes more seamless and simpler is going to really increase the value and the quality of that data. And once you have that you’ll have much better outputs.
What advice would you give to a young start-up or an entrepreneur with an innovative idea for the healthcare space?
The advice I gave to someone else is “marry rich.” [laughs] That gives you the breadth and everything you need to just stick it out. Maybe more appropriately, partner up. Try to reach out to professionals, to providers, to PCPs, to pharma, to payers (depending on your business model). Get them involved early on so you have the right product-market fit and you have the institutional support to test your thinking and validate it. Unless you’re a very experienced healthcare expert and you know everything about the market, you’d be surprised what you get out of that. And interestingly enough, almost every pharma company, hospital, or payer today has some kind of an innovation platform. They have their own accelerators, they have their business contests, there are so many opportunities. Use them.
You mentioned earlier that you wanted to make it very clear that you were a solution and not a platform. I went to a presentation at TechStars at Cedar Sinai and I’ve gone to a couple other entreneurship-focused events, and that kind of tends to be a common theme, that the industry as a whole isn’t super interested in platforms. They want to know what your specific molecule is or what your specific application is, what your specific solution is. How did Deep 6 make that transition from, “We’re a platform,” to, “We’re a solution?” Was it more externally driven or was it internally realized?
A combination, I think. From the start, when we started working for the intelligence community, we built a platform because we had very smart users and we could help develop solutions on that platform. But then taking those and reselling them somewhere else was really hard because they’re so customized and they’re so specific. When I was looking then at other applications, and I looked at a lot of other companies that are doing data science platforms, they were all solutions looking for the right problem to solve. Honestly, that’s the story of IBM Watson. It was like, “Well we can do this, or we can do that, and we can do this.” But there’s no clear application for it. “We just throw something against the wall, and we’ll see what sticks.” And I was reading so much about analytics, and it was all the same thing. “We turn data into actionable insights.” But that’s generic, that’s fluff. “We find the voice of the customer.” Maybe 3 dozen companies all had the same verbiage in their marketing materials and there was never any clear evidence of what they were trying to do. I wanted to be able to say, “This is exactly the problem we solve.”
I think that covers just about all the questions that I had. Are there any other final thoughts, anything we might’ve missed that you wanted to comment on?
Not really, we talked about a lot of the issues in healthcare. Again, what I think is exciting for everybody in healthcare is that there is a lot of innovation happening.
And where can we find Deep 6 either online or in-person.
In-person, at all the dive bars in Pasadena, or on the bikes. Online, at https://deep6.ai/. I think we have a fairly solid Twitter presence. I’m not involved, so it has to be solid, and LinkedIn. We’re going to be at South by Southwest as part of the accelerator program.
We’re going to be at a couple of conferences. And we’re always just an email or a call away.
Wonderful. Well thank you, Wout, for coming in, it was a pleasure talking with you.
Thank you, my pleasure.