Episode #5 –Accelerated insights from data for preventative health innovation

Ellen Sukharevsky  0:29 

Hello and welcome to the BC Platforms podcast. My name is Ellen and I will be your moderator today. BC Platforms is a global leader in providing a powerful data discovery and analytics platform as well as data science solutions for personalized health care. BC platforms enables cross functional collaboration with our global federated network of data partners. Today's podcast will focus on the topic of accelerated insights from data for preventative health innovation. The podcast sheds light on industry challenges and expectations as well as innovative approaches to collaboration. This discussion is led by Tonu Esko, BC Platforms SAB Chairman and Vice Director of the Institute of Genomics University of Tartu, where he also holds a Professor of Human Genomics position. He is head of the Estonian Biobank Innovation Center and focuses on public private partnerships and innovation transfer. Dr. Esko is also a research scientist at the Broad Institute of Harvard at MIT. He acts as one of the senior leaders for Estonian personalized medicine programs and serves as a scientific advisor for several companies. Our speaker today is Michael Geer. Michael Geer co-founded Humanity along with Peter Ward. Humanity is a health tech startup that enables you to find your rate of aging, and then what you can do to slow your aging down. Michael Geer is a serial tech entrepreneur, having amassed over a billion users, Badoo, AnchorFree, and even more revenue over his years building meaningful consumer focused technology. He wakes up every day working on how to best enable people with superpowers. Now, I will hand it over to the speakers for a brief intro to begin the discussion. 

Tõnu Esko  1:59  

Hello, my name is Tõnu Esko and I will be the host on this podcast and I kick it off with ice breaking question to you, Mike. How long do you want to live? Is it forever? Or is it, I don't know, 500 years from now?

Michael Geer  2:31  

Yeah, I really enjoyed life. In my career, I've also done things to basically always try to give people as much freedom and options as they can. For myself, I would want to always have the option to live longer in a healthy way. So it's more about the optionality, I guess, then that I would like them to pick a particular number today. But yeah, at this point, really enjoying life. So I would love it to last 1000 years and not 100 years.

Tõnu Esko  3:01  

Yeah, that's a good one. A key learning or insight is that it doesn't matter how many numbers, but what's the quality of those years and how fulfilled and happy you feel. That's why I'm really glad that the co-founder from Humanity is on this podcast, and today we are talking about big data, but how data plays an important role in the startup life and how the data and questions around data actually either can accelerate or break a major innovation. So if we kick it off, what are the main obstacles in doing innovation in healthcare, in your mind Mike?

Michael Geer  3:49  

I was giving this some thought. I'm a product person and have been since I got into consumer tech. I see everything and kind of funnels of different sorts. Basically, my whole life, is always a funnel. I think what we see right now is, you have these, starting with the goal. What's the bottom of the funnel, what's the goal of the funnel you want to have for Humanity. And I think for a lot of health tech startups, we want to give meaningful extra years of quality, healthy life to everybody on Earth. But to get there, as you rightly point out, it starts with data at the top. Then you need the ability to analyze that data in an effective manner. You then need to actually have, basically at the same time, the ideas or the ways that you might be able to leverage that data, right? And then you need to be able and this is the crucial one, you need to, once you get those learnings, you need to be able to distribute that out to millions and billions of people. And I think that's where we sometimes or a lot of times expect all those things to be within the same group of people or the same kind of organization. And I think that's what we're seeing now, is that the ability to organize cross organizations and to open up that data is the thing that will lead to all this impact, right? If you take that from the top, you have, and let's start inserting companies, and then you have, let's say the data part, right? You have the Estonian Biobank, just as examples, then you have BC platforms, which has the infrastructure and allows for that data to be cleaned and sorted and interactive, you have the analysis skills. I think, honestly, if you could open up that data to the entire world, you probably have at least a million people, probably more, that have the machine learning and data science skills. That number is probably more, like 50 million people, right? And then the ideas, if you can open that up to 7 billion people that have potential ideas, right? So the more you can reach those 7 billion people, let's just call it 100 million people that can really have access and can leverage their ideas, you can just imagine the impact. I think why Humanity and why Pete and I, as consumer tech folks, decided to jump in, is because of, one we're good at identifying good ideas, we don't always come up with them, a lot of our ideas are really bad. But we can identify the good ones, right? And we can distribute stuff out to those millions of people. And so interacting across these organizations, and making sure that data is actually accessible is the key thing. I mean, I'm sure you've seen that with the Estonian Biobank, and with BC platforms. What are the biggest blockers right now, do you think?

Tõnu Esko  6:32  

I think the biggest blocker are these two facts. One is that the data is siloed. And in those silos, it's also in a random silo specific format. So it's a pain in the neck to actually join those different data sets or even replicate your scientific insights that you're playing from one silo to the other one. Because it's just the internal structure of the data is just so very much different. And the other one is actually quite strong ownership of those data sets. If you think about healthcare institutions, or public biobanks, it's still very, very complicated to get access to that data. And then of course, because it's why you should open it up, it's all data. We put sweat into, get it together, and so on, and so forth. So that's actually, the structures of the data and the accessibility. And then of course, I'm not even going into all the different ethical and regulatory obstacles, doing international research and open data sharing. And I think it's also fair, because health data and molecular data is this special type of personal data, which actually carries a lot of loss of information. So those are my own insights. And if I take it further, then I really liked your concept of being, this data community and having 7 billion people with potential good ideas, and 100 million folks with excellent machine learning skills to harvest this data and make some or generate some insights. So I really, really like this idea. So what you might think is a good way to open that data or how to leverage this knowledge. How have you done it in Humanity, in order to build this rather complex health guide using different data types?

Michael Geer  8:50  

Yeah, I completely agree with it, the angles that you were talking about there. I don't think I'm being overly optimistic, because we've seen this happen to different industries, like banking. I think the dam will break, you touched on a lot of these. Birds can be killed with the same stone, not to use too much violent analogy. If you can keep the data private and the ownership and even the location of the data kept where it is, with federated learning and you can make sure that when the models are trained and that learning comes out, that there's sort of, whether it be differential privacy or some kind of security layer to make sure that the individual records data doesn't come out. I think that will allow for both on the regulatory and ethical side. It gets rid of a lot of those hurdles, because you are accomplishing the goal of making sure that patient or person owns their data and that the privacy of that data is kept but you're able to get all these learnings out. I think it also allows for less gatekeeping. And so it's instead of saying, hey please propose your research idea. And then you need to check your backgrounds and all this gatekeeping, that needs to happen now, because the data is not really in that structure. Once you get it into that structure where the privacy can be preserved, I think you can let the gates down quite a bit. And that then allows the person we talked to offline about this, that allows the person sitting in their living room, with that good idea and a little bit of machine learning skills or data science skills, that they can interact with that database, they can try out their idea. And I think that basically unleashes a lot of people out there. That won't solve, necessarily the distribution part, which is why Pete and I feel is very important for Humanity to be in there, and especially in the health span space. But it will very quickly amplify at least 1000 fold, the amount of breakthroughs and learnings and valuable data models that then can be applied to people to improve their health. So I think that is the key part, because after that we won't be able to stop the innovation. It's like, stop the internet. Once it was able to have these interactive protocols that anybody could connect up to, without any gatekeeping, there was no way to stop the unleash power that was created. So I think that's where I would like to see more people concentrate on. What I do see, is a lot of that, which makes me very optimistic. What I also see through, which I think we should probably do a little bit less of, is the same people that are thinking about what they should do, to open up that data, feel like them, or a particular person they're talking to that week should also be the people that then decide what to do with the data. If you get what I'm saying, they're keeping this more old school of thought that we're going to structure the data, and our colleagues are going to be the ones that leverage the data. And I think that, once we can separate those a little bit more, I think it will be a lot more clear what needs to be done. 

Tõnu Esko  12:02  

Yeah, I think this democratization of data or even broader democratization of knowledge, and a good analogy could also be the sequencing of the human genome. Part of this initiative was actually a private effort, which aimed to put this information behind the paid firewall, and would have a major impact on innovation and an impact on where we are now on understanding the genome The folks who gather the data and guard the data and also revere your ethical background and scientific rigor, they also decide who of the rebels get access to the data and who doesn't. So I think this democratization of data, but also the federated access, and this privacy preserving aspect of it, I think it's of crucial importance. And to be honest, I don't know, maybe, or even any of these platforms that would enable different data collections to connect, and then allow this privacy preserving way to mine the data. I'm quite surprised that there isn't anything like that, or if there is, it's actually not prevalent. There has to be this gatekeeping now, because it's really, really very sensitive and potentially impactful data. If we can have this privacy preserving way to mine the data, then we would put gates down, and then people would just have fun and just innovate without no borders. But until we don't have that platform, it's very tricky, because maybe 40 years ago, it wasn't very important if you gave away your genetic data or material that you could obtain a genetic profile. But today, you may already start to think, but maybe tomorrow, you wouldn't like to do it, because then you could just get cloned or something like that.

Michael Geer  14:21  

People should have the choice. I've put my genetic data to George and anybody that has a good research project, but people need to have the choice for sure. They need to, it's there. It's information about themselves, they need to have the choice around it. I think the good news is from my interaction with a few different companies that are working in the space of federated learning and differential privacy. A great one is OpenMind, which is working on trying to democratize the technology. They'll take some libraries that Google has created or MIT has created and try to make them more readily usable by other folks. I think what we'll see is the BC Platforms is kind of a roadmap. For platforms like BC Platforms that already work with data and the interaction with data, these are features that are added on. It’s not some brand new thing that needs to be created. And it really ends up being quite low weights as well, not to get into coding or anything. But when you add, you're talking lines of code within a system, you're not talking about some change of how we actually store data. It's more metadata and layers of layers of code on top of what's already there. So I'm quite optimistic about it, I think it's that focus. What always is hard in business is going into a board meeting, or going into a strategic meeting and trying to convince people of the opportunity cost, which is basically saying, what would happen if we did something, because it's natural for the human mind to be very focused on what we're doing right now, what is real right now Then when you go into a meeting and say, hey, actually, we could 1000 times speed up the innovation that we have in science, if we do basically democratize the data, it's very hard for people to wrap their head around that. It's much easier for them to look at the manual, we have this data in this organization from this university that has this great idea of how to leverage it. That's just easier for us to grasp, right? And so, yeah, it's understandable why sometimes we don't do it as quickly as we should.

Tõnu Esko  16:33  

Yeah, another two keywords just on top of this is patronization. So again, you have some folks in university sign, this is a valid idea, or this is, holy enough idea to pursue. The other one is actually the only one that shouldn't be even considered, because it's so academically uninteresting. I think the other one is, what should be also quite a lot is this open consent for the data owners, or the citizens who actually own the data themselves. Maybe I would like to give all my data to Humanity, but I wouldn't maybe like to give it to some other companies. So, there also should be this possibility for flexibility and all that. But now, we have retreated into this very philosophical concept of data and how to share it. But if we jump back into the core of this discussion, where did you or how you manage to get either the same data sets, or the scientific insights, or algorithms to build your products?

Michael Geer  17:54  

Yeah, I'll give you the short version of this, but you're included in it. So it'll be an interesting story, I hope. So we went on a journey, me and Peter. First we had fairly tragic health events happen to people close to us. We then wanted to find out how we could actually apply ourselves to the space as preventive health space is probably the best way to put it. Was looking at early detection, and then started meeting a lot of people like George Church and others. I started to get a signal that actually early detection wasn't the best, wasn't the best that was possible, we could actually look upstream of this, and actually understand how far people were from disease. So it's the spectrum of completely healthy, and then you start moving towards disease, and then eventually, you would get diagnosed with it. And one of the first people that really turned the light on in my head and basically led to, what we ended up building with humanity was Kristin Fortney, who you know well. Then Kristen runs a company called BioAge on the Valley. And one of the greatest things she did early on is she worked with the Estonian Biobank with you and with other biobanks. This idea that you could actually, even at the surface level of the data that was already in the biobank, you could go in and see this amazing thing. You saw these past biomarkers, and then these future health outcomes of these same people. And because of that, because of the high correlation, you don't need causation, necessarily, you need high correlation, you could actually predict the future health of people once you got these same biomarkers, and that whole concept. And then just working with a ton of scientists over the next five years was seeing that they spent 90% of their time really trying to get the data, and only 10% of their time actually doing the thing that ended up making them famous and the analysis, right? So seeing those two things is that first of all data can predict the future. And second that all these even the top scientists that I could find in these different fields are spending 90% of their time, just trying to get the data set made me understand that that's probably where you want to break the break the dam, to let the flood of innovation happen is just get rid of that piece and make sure that that piece is a given and that the person in their living room has that piece. And that's, five years of hard work by one of the top scientists in the world kind of thing. And so I think that's where that's, that's where I came to, and why there's such a concentration with Humanity on. This access to data and not just for us, but for everybody that wants to that's mission aligned in space. And that's one of the things we actually want to do with Humanity, as well as work with different biobanks and different data sets. And every time we touch a data set, we want to actually be able to add more information to that data set. So everybody else that then comes in contact with that biobank, or with that data set in the future, actually has access. Because again, we're not going to be Peter and I and our team, although we'd like to think that we're smart, from time to time, we're not going to be the people, the only people on earth that come up with great ideas to extend people's health span. We don't expect that to happen, we don't want that to happen, it'd be kind of sad if it was that way. And so that's kind of it all, it all revolves around the data and giving people easier access to it

Tõnu Esko  21:12  

It's important to not just consume the data, either in our customers data or some data custodians data, but actually try to give back something unique, something that this technology provider can provide a painkiller. Because, sometimes or usually, for those data custodians, it would be almost complicated to get, for example, physical activity information from your phone, if you don't already have an API and technology, how to extract it, how to maybe structure it, and how to get some insights from there. So I think it's very, very, very important. It's also one way to open up this state of communities that it's two way, the benefits go two way. And I think we always need to get the citizens on top of our priorities, because we're doing it for the community, we're doing it for the patients and not ourselves.

Michael Geer  22:21  

The one thing that we also wanted to do with humanity is from day one, we very much limited ourselves to ideas that would bring value to the N Equals One. Because you can think of a lot of other ideas, but that would help Humanity possibly, but not necessarily the person using the app at that moment. But I think what happens a lot of times is that you get abstracted and you start doing things that don't interest the user. Because they can understand the concept of maybe this will help with research later on. But, but if you're not bringing value directly to them on that day, or that week, or that year, first of all, you're not going to have them stick around. And so that is longitudinal. Nature of the data is not going to be a length of time that you have measured the same person. Secondly, I think it's important to wake up every day and just think about, like you're saying, okay, the patient or the person, however you want to call them, is certainly the citizen, as you said, is the center of this, of this whole thing. And starting from what can we do today to make that person healthier, will always keep you a bit more focused on the goal, then if it's, okay, here's a great project we're going to do around data. And then, oh by the way, we think this will be a value that a citizen might get from this. I think it's very hard to keep on a straight direction, if you have that angle. Starting with the citizens is definitely the way to go. So I agree with that.

Tõnu Esko  23:53  

Yeah, and I think this is a concept that is very well aligned within the startup community, because you need to obey customers, the customer is literally the king or the queen. So you have to please them as much as possible, even a little bit more than possible. But  you can think of this general research community or the state, they need to be a studying community or other tribes driven by this intellectual curiosity and the citizens. It's more of everybody in the background, but if we have time, we will try to come up with something useful for them as well. But they should nonetheless, give us all the data and then record it. But we're a little bit running out of time with this very interesting discussion. But maybe the last point to cover is, I know that you in Humanity have different types of data that you consume and from the customer side and then try to give some feedback based on data. So what's your data, how detailed should this feedback be? And if you have, for example, physical activity from the phone, and then some biomarkers and then some sleep stuff, you have this very complex data model. There is quite a big chance that the customer gets lost in all these tables, figures, stuff with all kinds of check results. How do you manage this complexity and an information overload this year?

Michael Geer  25:30  

Yeah, yeah. We set out to be very radically inclusive, which means that we need to interest the mainstream person and not, whatever you want to call the person, biohackers. We very much focus on having two main scores in the app. One is your daily Humanity score, which are just points you get for your actions during the day. And then the other one is your biological age and your rate of aging. And we try to keep very focused on that, obviously, if you can click down a couple pages and see more and more detail on your data. But I think it is important to people, for people to focus on, just a couple of numbers. Because fortunately, or unfortunately, I think it's all worked out fine. But we definitely did have 15 years of wearables just telling us step counts and other abstracted data that never really got leveraged into real change and people's behavior, or at least not as much as it could have been. I think that's changing pretty quickly. Now, Humanity's definitely leading one of those efforts is to basically allow you to really, and longitudinal data connects it, allows you to really actually know are these actions that I'm doing today, are they making me healthier and are they adding healthy years to my life. I think there's other amazing stuff when you start. The great thing is, if working in this field, is that I started and we geeked out on this very shortly on our offline was, there's this whole math and statistics of causal inference and basically looking at observational data sets and being able to get actual cause and effect out of out of that space. I think the importance of that is you need to actually be able to start to guide the user or give them options, and let them see if it's working out. And that feedback loop needs to be pretty tight. I think, just presenting data without giving some interpretation of it, as in that the user can understand what it means, is fair, not to overstate it, but it is fairly pointless. Because you need to have that interpretation layer. And then obviously, you need to keep within the bounds of what's actually scientifically true. I think there's more and more players that are capturing those two components, and we hope that humanity is a real poster child of doing that.

Tõnu Esko  27:44  

Conceptualizing the information is really, really important. And it's the same with the genome, it's made of almost 4 billion base pairs, and we all have at least 4 million unique mutations. So we are all literally mutants, but none of us is imitating later in the night, or in the talk, or maybe until they were the lucky ones. So I think with this, we can conclude today's podcast and give it over to Ellen to guide us out.

Michael Geer  28:15  

Yeah, thanks for the conversation.

Ellen Sukharevsky  28:17  

Thank you to our speakers for joining in and to everyone for listening. Speakers, do you have any final comments? 

Michael Geer  28:23  

I mean, I guess the only thing is, as you can tell from what I focused on during the talk, I think any mission aligned folks that believe in opening up the data and that have a skill set that allows for that, or if you're a data custodian, and you're looking for solutions, I would love to talk to you. And I'll help connect you with different folks that can help you actually bring this to reality. And with Humanity, we have about 10,000 people on the waiting list, but we're trying to get people on as quickly as possible. So if you go to humanity.health, we'd love to start to try to bring your health to a higher level and obviously learn from all the feedback that you can give us.

Ellen Sukharevsky  29:01  

All right, and with that, I think we can conclude. Thank you for tuning in to our podcast and thank you Tõnu and Michael. To connect with our company and learn more email sales@bcplatforms.com or visit our website bcplatforms.com. Thank you so much, and we hope to stay connected with you.