Singularity Summit 2012
For more transcripts, videos and audio of Singularity Summit talks visit intelligence.org/singularitysummit
Speaker: Linda Avey
Transcriber(s): Ethan Dickinson
Moderator: Our next speaker is Linda Avey. She is cofounder and CEO of Curious, Inc., a personal data discovery platform. A biologist by trade, she also co-founded 23andMe, the leading personal genetics company, and worked at Applied Biosystems during the early days of the Human Genome Project. Time Magazine named 23andMe's saliva-based DNA-testing service "Invention of the Year" in 2008, and the service has played a major role in popularizing the value of personal DNA testing in the American public consciousness. It is my pleasure to welcome to the stage Linda Avey for a conversation about genetics, 23andMe, and her new company, Curious, Inc.
Moderator: So, welcome.
Linda Avey: Thank you, it's great to be here.
Moderator: It's great to have you. You are most famous for founding –
Linda: [laughs] I don't know if "famous" is the word.
Moderator: Hey, you're up here, there's a reason for that.
Moderator: You're most well-known, at least, and the first thing on your Wikipedia page is, the founder of 23andMe. So I know everyone is going to be wondering the state of the company today. You left in 2009, and we'll talk about what you're up to now in a minute, but I know that you're still keeping close tabs on the company. Can you tell us a little bit about where 23andMe is today, and what the state of personal genetic testing is today?
Linda: Well, I'm always curious in a crowd like this to see a show of hands of how many people actually have their 23andMe data, just out of curiosity? We figured it would be really high [laughs] in this crowd.
The company's moving along as planned. We knew that it wasn't going to be a YouTube- or an Instagram-type takeoff, because there was that cost up-front that people have to pay to access their genetic information. To see that over 150,000 people now are on board and a high percentage of them are contributing additional information so that we're able to do the research on the platform, that was really the intention from the beginning, so it's really gratifying to see that continue. They're experimenting with pricing and some other things, but I do believe it's on a great path, and the publications are coming out now and the American Society for Human Genetics meeting is happening soon, and I think they're presenting at least four if not more papers and posters. It's great to see the research side of it really taking off.
Moderator: 150,000 is great progress. One of the reasons that maybe that number isn't higher, is people are often concerned, and you've spoken about this many times, with the privacy of their genetic information, the privacy of their data, and might it be used against them at some point in the future. Genetic information, of course, is hard to anonymize, so there's a somewhat fundamental tension there. Do you see any resolution to that tension in sight, or will genetic information and privacy always be in tension in your view?
Linda: I think that's always going to be a tension, and it seems to be pretty ingrained in people, whether or not they're concerned about it. I was always struck by how immediately people had a response to the notion of 23andMe, or just having any access to genetic information. It seems that there are people that are just so ready to go down that path.
The people who don't, or who are really afraid or have concerns, the good thing to tell them is, "You don't have to do this, it's completely optional." That relieves a lot of the stress and the tension, I think. I think some people have come back to say, "Oh, well, maybe I will give it a shot." Then when you talk to some people initially they have this concern, but then they might have an event in their family where there's a child born with a disease or something, and then all of that completely goes away. It's more important that they see the value of accessing the information and being able to share it with a researcher or something. Things change, and people have different perspectives on that.
Moderator: You've also been kind of a living example of sharing this data and putting things out there. You've stated publically that you are a carrier of the APE-4 gene, correct, do I have that right?
Linda: APOE, yeah.
Moderator: APOE-4. A gene that carries an increased risk of Alzheimer's. You've said that that's usually still quite late-stage, which is encouraging. How did you decide to make that information public? Would you advise others to make information like that public, or do you not try to advise at all? And what do you see as the benefits and the risks of sharing that kind of information with the world?
Linda: I never even thought twice about it. My husband and I both are carriers for it, interestingly, and it's actually not that rare, I think it's about 25 percent of the population is a 3-4 combination. We all have an APOE gene, but it's just what combination of these mutations do we have. I also look in my own family, and I don't see incidence of Alzheimer's, both grandmothers have lived to 100 and really never had any signs of it.
So there's clearly so much more going on genetically than just this. So in a way it's a way to defuse concern that it's so informative. It really isn't. By opening up our own data, and hopefully if other people are interested in doing the same, then we'll learn more as we just get more data and we can start looking at it. Not just APOE, but all the other genes that are now starting to look like they're implicated in some ways. There could be some really interesting protective genes, that's what I'm hoping for, that that's why I haven't seen this problem in my own family.
Moderator: Gotcha. For people who are thinking about this, the risks I think that usually come to mind are things like, "The insurance company's going to discriminate against me." You've said that that's not such a big concern because we already have some legislative protections in place.
Linda: With at least health insurance, yes. The GINA law that was finally passed, and passed overwhelmingly in the Senate and the House, and President Bush did sign that. It was really the first way that – it's not perfect, there are a lot of holes in that legislation, but at least it was a first step. Because there was such overwhelming support for it, it does seem that there's a recognition in society and in our government that we need to protect people so that they can access the data. I was really shocked at how much support the bill got.
Moderator: Cool. Let's talk about Curious, which is your current company. It's described commonly as a personal data discovery platform, but you guys are still in stealth mode, so not that much is known about it yet. I was just on the website last night and signed up to receive more information. The URL for those who want to sign up, and I'm sure that many people will want to do this, is wearecurio.us. Just spell that out, and then put the dot between "wearecurio" and the "us", so the "us" is the suffix.
What does it mean to be a personal data discovery platform? That could be a lot of things, and maybe it will be a lot of things.
Linda: Yeah, and we're still defining it, of course. The interest we have is where we see all of these new devices coming on the market, where we're going to be able to start placing them on our body in all kinds of different and interesting ways. Right now our bodies are really a black box, and 23andMe was just one way of tapping into these reservoirs of data that we have.
I'm very interested now, and luckily I found two co-founders of similar mind, that if we could start to get this data out of ourselves and into a single place, a platform where we can now start to probe it, look at it, look for correlations, look for things that might be going on. Not only for ourselves, of an n of one, but n of many, where let's say a whole group of people come together who get migraines, and they really want to start to experiment, "Do we get them because of what we eat, because of the environment, or genetics," or what have you. It finally is a place to do that all together. If we want to bring our sensor data in, and we want to bring in our food, or what the weather was that day, it's really going to be an opportunity to combine all that together.
Moderator: When I hear you talk about people, for example who have migraines coming together to try to solve their problem, the lightbulb goes off and says, PatientsLikeMe has that community aspect. It sounds like you're looking to build community, plus a big data solution.
Linda: And very undefined in letting individuals come in who have a theory or a hypothesis that they want to test, and bring whatever type of data they want to bring in. Some days they might be more interested in their psoriasis than their sleep. We're looking at people very holistically. You're a whole person, you have all kinds of issues potentially, or maybe you don't have that many, maybe you really just want to optimize your marathons. There are always questions about, "Should I eat more carbs before I run, or should I lay off on the carbs?"
We have so many questions – and that's actually our tagline, "We've got questions" – that we just don't know and we don't have, not only the information, we don't have the data to really support different theories. Hopefully, the supposition is that if we do have the data we'll be able to publish those and then share those with the rest of the world, so that other people can come in and look. It won't be Yelp! in the sense that it's "Oh, it's an opinion about something," it's actually data-based, it's informed by data rather than just opinions.
Moderator: Yes, I was just going to ask you about the dividing line between data and opinion. Somewhere between the two are things like, "I felt great today," or "I felt not so great." How do you guys approach that? Is that something that you're working on in the very beginning, the subjective, self-reporting data slash opinion? Or do you think that that's too gray and maybe should be saved for later?
Linda: Well, that's certainly important for an n of one. If something makes you feel really good, and you can track what that was and nail that down, then great for you, if you want to repeat that and really prove it to yourself and then share it out to a community of people who have a similar issue that you were thinking about.
We totally believe in anecdotal data and something that makes you feel better, even if it is just a placebo effect. That's still important, and it's still meaningful, because you felt better. Then being able to dive into that more, "Well why did you feel better," and really now pose another question of "Was it because I slept better? Or was it because I took that supplement from Whole Foods, hmm. Let me stop taking the supplement and let me just track now." So really trying to peel the layers back on the data and get to the bottom of it.
Moderator: The statistics on this obviously sound incredibly complicated. Can you give us even just a teaser as to what are the methods, the technology that you're developing? I know it's still very much in development, but how do you begin to tackle something that has so many free variables?
Linda: I hire lots of really smart people. [laughs]
Linda: I'm so lucky that my cofounder, our CTO, has just an amazing background and way of looking at these things and thinking about them. He spends a lot of time really looking at various data. We've been doing our own tracking, and he's already starting to think about, "OK, how do we feed this back to people through auto-correlations to say, 'Huh, looks like this sort of is correlating to that, is that interesting to you?'" Always posing it back as a question. We'll refine that. It's going to take a ton of work, but yeah, we'll definitely be looking to hire some really smart people, who I know are here in the audience [laughs].
Moderator: That's cool, and you guys are based in... ?
Linda: New York, for now.
Moderator: New York City.
Linda: We will be hopefully building a team here, probably backend people more in the Bay Area, and frontend people in New York is how we're looking at it.
Moderator: OK, cool, so keep that in mind. And again, the URL for those who maybe want to check it out, even right now, wearecurio.us.
You've speculated, I've seen in the past, that certain public health phenomena such as rising autism rates, the seemingly meteoric rise of gluten intolerance, is not, as some have said, just a measurement effect or overly concerned parents or whatever, but you think it actually may be legitimately connected to things in the environment. How do you think about that, and how do you see those sorts of things connecting with the Curious platform?
Linda: If you give people a way to start tracking things, and really trying to measure what is going on in their lives, and then bringing the genetic component in as well, to start looking at are there predispositions, or certain genetic profiles that we see that are showing up more in people who seem to have these intolerances or what have you?
It's just an opportunity, we think, to really give people a place to ask the questions and start to probe more. Because I do think there is a bit of an overfocus on that right now, but the thing is people stop eating some of these foods and they feel a lot better. To be able to quantify that in some way, and then be able to say, "OK, now what do we measure? Do we start adding things back in?" Almost like you do with allergies of, you bring it down to the very basics and then you slowly start to add things back in. Then can you start to measure, "OK, that's when I started feeling bad again."
We don't pretend to have any answers about these things, but we hope our platform will give people a way to start figuring it out, and then just hopefully feeling better. We're all about prevention. I was involved for most of the end of the last part of my career in the whole personalized medicine space, and I'm so done with that [laughs]. I'm so ready to be in a world where we're trying to prevent people taking medicines altogether, and what can we do to feel better, and deal with things before we get sick and before we have to end up in an emergency room or seeing a doctor.
One of the big things we're excited about is, can we redefine disease? So many of the disease labels we have nowadays are so antiquated from the 19th century in some cases, and they don't really seem to fit. I think that's where we're starting to see these problems like chronic fatigue and the overlap with fibromyalgia and all this, it's such a huge gray area. If we can just have people simply tracking their own data, starting to take advantage of these amazing technologies that are going to be coming out, and now does that data help inform and describe disease in a much better way? Will we have people who thought they all had MS or something like that, now suddenly parsing into other groups, and just seeing these things redistribute once we get more data?
Moderator: Cool. What are you personally tracking, if you'd care to share? You said you're doing your own tracking. What sorts of things are you looking at in your own life?
Linda: Do you really want to know? [laughs]
Linda: I'm 52, and I have hot flashes like you can't believe. I am tracking those, because I'm very fearful of hormone replacement therapy, I don't know if there are other women out there who are in my boat, but it's a really scary time. My grandmother and my aunt both had breast cancer late in life, which I don't have the information, but I assume that they were probably on hormone replacement therapy, and so I'm avoiding it like the plague.
So then I deal with all the ups and downs, sweating and all of this. It affects my sleep terribly, so I'm really looking at my sleep patterns as well. Then of course there are all these supplements on the market for this that they claim that it makes it much better, and do they work or not? Nobody has any idea. I'm really interested in partnering with other women and using this platform to start to track these things.
There's a company called Proteus that has a patch. They actually have a platform where it's a pill with a chip on it, you take the pill and then as the pill degrades in your stomach it sends off data to a patch you wear on your arm. The patch is also a standalone that takes temperature, heart rate variability, galvanic skin response, all of this interesting information, that if I can just wear one of those and be able to show when this stuff's going on, I think it'll be extremely interesting.
Moderator: Let's talk about sensors. That one is obviously right on the cutting edge in terms of something that goes all the way through the digestive tract, and emits information along the way. Most people here are probably much more educated about this than your general audience, who may only know that Nike can help you track your runs. I think you probably could give us some other good examples too, of sources of data, types of data, things that we may not be aware of that are starting to become possible, and just give us a sense of the type and the amount of data that we could generate if we all went home and bought a handful of sensors, or made a small investment in doing that.
Linda: On the sensor side – and I'm advising a company out of the Rock Health program here in the city called Sano Intelligence, and it's basically a microneedle patch that you put on. It barely scratches the surface of your skin, but the microneedle is able to go into the interstitial tissue and pick up a lot of the molecules that you normally would have to measure through a blood test. This is real time flowing data. Their tagline is they're "The API to the bloodstream," which, it's going to be crazy to start getting this data where you start seeing your chloride, your magnesium, all these things that you maybe get once a year when you go to the doctor, and to see how those fluctuate throughout the day.
I think especially for elite athletes, this will be the first place where they really use that data to optimize their performance. Because a lot of times a runner will have a run and they'll feel great, and then three days later they go for a run and they feel terrible and they don't really know why, "Why is that? What was going on with my energy level?" Maybe this data will be able to inform on that.
Then on the bio side, things are getting really crazy. Of course with the whole-genome sequencing and the next-generation sequencing platforms coming out, driving the cost down precipitously, but then there's also our microbiomes. This has been in the news a ton lately, where if we can start to access that data, whether it's through our guts, or on the surface of our skin, and be able to say, "Do all these little critters living all over me have some influence on how I'm feeling, how I'm sleeping, how I'm eating, my weight, all kinds of things, my skin?" There's any number of things that we can start to ask by having access to that. So that's something we're really excited about.
We all got our telomere length read through a company, where it tells you how the ends of your chromosomes are looking. Apparently stress can shorten them, and that leads to the aging process. That's a reversible thing. That's another datapoint that we'll be able to give people access to and then bring into the platform, and say, "Now, look at that, and then over the last three months, have you had a really stressful go of things, and is that influencing that number?"
Gene-expression profiling is another one, where maybe every three or four months you go in and just do a swab of your cheek – in fact there's some ideas going around of having dentists be the frontline for prevention and doing some of this work. Because you probably go to your dentist, if you're pretty healthy, more than you go to your doctor. Every time you get your teeth cleaned, they do a quick gene-expression profile of your saliva, and they're able to pick up now some very early signs of cancers. It's work being done down at UCLA.
But all that cumulative data right now, there's really no place to put it.
Moderator: Does that data all have a roughly equivalent form? It all boils down to a bunch of matrices, or is there anything that you're finding that's a little bit difficult to put into the traditional paradigm of big data?
Linda: I think the streaming data is going to be the most exciting and interesting but also the most challenging. Because there's a lot of noise in it, we're going to have to put a lot of thought into "How do we make that meaningful?" where spikes in that don't look like false positives or things going on. The genome is static, and there's a lot of that type of data. Your telomere length is one number that you get. Because it's across the board like that, then also merging it all together and making sense of it is very challenging, but again we feel like we're starting out in a more simple way, and then as all these technologies start coming on the market, hopefully we'll be ready.
Moderator: You touched on the idea of spikes in the data, and being cautious about overinterpreting them. Is that something that you worry a lot about? I just think about the prostate cancer phenomenon, where the drum was beat for a long time, "Gotta get the test, you gotta get the test." Well all of a sudden it turned out, no you don't really have to get the test, the test often just generates a lot of false positives, people get unnecessary treatment, and you might just be better off if you didn't ask that question.
There's obviously a failing there that's not just in asking the question, but how do you think about that, and how do you think about people using Curious, but making sure that they are not using it in a way where they're shooting themselves in the foot by overdiagnosing or things like that? That seems like something that will be a challenge.
Linda: I think with bigger data we'll get there faster, to these questions of, "Is this something to be concerned about or not?" Part of the problem with our existing culture and our existing medical system is we don't have access to our own data, oftentimes. This whole new model of people, not only getting their own data, but having access to it and being able to look at it more, and then putting communities on top of that, where people can share and talk about it... We don't know, we don't have all the answers yet of how this will all play out. But we think it's worth experimenting with individuals now being able to have more control and more say on how their data are interpreted, and how that fits in a broader community by using social technologies that allow them to share and compare.
Moderator: I've heard you say in the past that you've talked to folks, even at Facebook, about working to try to find health information, and that they always get a little scared. [laughs] They have their own privacy concerns and they may not want to be associated with the brand, but there's certainly a huge dataset there.
You also had mentioned athletes as a possible user base. It seems that in general, athletics is the first thing that we, humanity in general, pushes to the fore. I'm not sure on exactly why that is. Do you have a view on how useful Curious can be to athletes? And is there any reason why the same sort of push for peak performance can't be applied to intellectual domains? It doesn't seem to be as rigorously done there for whatever reason.
Linda: You're right, I think we look at athletes sometimes as the guinea pigs. Certainly they do a lot of testing on themselves and give themselves a lot of things that they shouldn't be probably [laughs]. Lance Armstrong comes to mind.
They are so curious and they have so much interest in really – it's like a high-performance car, right? If there's one thing wrong with it, you see it right away, as opposed to an old clunker, you may not notice that one of the sparkplugs – I don't even know if there are sparkplugs in cars anymore, but... You start to see these things right away, because one thing being wrong, the utmost performance is off.
I do believe that there will be opportunities to move this platform into a lot of different areas, and certainly Mitsu my co-founder is especially interested in making sure that we're not just focused on health. Anywhere that massive amounts of data could be informative. He sees this as really building information beyond what we have now. The Internet really is a lot of documents, and it's just accessing documents and links to those. Whereas, if we could have an Internet that's based on finding data that's been assembled and really accurate and precise in hopefully curated ways, that it opens up a whole new way for us to do searching, and finding relevant information.
Moderator: That kind of leads into the next question for Curious. Maybe it's too early to have any answer for this, but what do you guys think of as making it into a business? Is there a business model? Have you put any work into defining that, or is it pure technology at this point, and the business will follow on from it later?
Linda: We certainly have all kinds of – there's the freemium models, and lead generation, and all sorts of ways that we think... They're all in our pitch, in our deck.
The one thing we won't do – that's what we're focusing on more, is the things that we won't do. For instance, we will never patent information that we derived from our customers, our users. They're sharing it with us and they're letting us hold onto it for them, and we're giving them tools to get better information out of their data. Things like gene patents, I'm very opposed to. You take information that you've derived from a community, and then you go off and sequester it through a patent. It's good for us to really put that out there first, rather than what we think we're going to do, what we know we will not do.
Moderator: That's great. I know that Twitter has also taken a lead in terms of saying what it won't do with patents, and I know you're a big user of Twitter and on there all the time. I saw that you had been weighing in a little bit on the upcoming – I don't know if it was specifically in relation to the upcoming ballot initiative about genetically modified foods, but certainly on the topic in general. Given that that's something that's so now, would you care to share your thoughts on, not just the upcoming ballot initiative, but the issue in general?
Linda: This is just a minefield. The sentiment of Prop 37, I can understand. I personally just feel like it's going to open up a can of worms there as well. I tweeted that the other day, "Define GMO." What is that? It's so ill-defined right now, and I think the general population has no idea really what the science is behind this.
One of my good friends, Mike Eisen at UC Berkeley, is just rabidly opposed to Prop 37. Then somebody challenged him on Facebook I think about, "Well are you getting funding from Monsanto?" That seems to be the automatic knee-jerk reaction, but he's just looking at it purely from a scientific perspective, that we're not there yet. To force this on food providers is really going to probably add a lot to the cost of food, and it really is a misplaced sentiment right now. Because again, I feel very strongly that we need to know what we're eating, but this is not the right way to go about it.
Moderator: Gotcha. We've got a little bit of time left, I have one question that actually did come from the audience, and then we may have the chance to take one or two more.
Genetic information, obviously, as it proliferates and we get more and more of it, some have worried that it will give rise to, or give opportunity for, the creation of superbabies, or genetically modifying ourselves. How do you think about that? Is that something you worry about?
Linda: Yeah, it certainly came up when I was still at 23andMe. In fact we were asked to go into a fertility clinic. Because apparently, it was a couple where they needed an egg, and the soon-to-be-hopefully-father walked in and said, "Well I have my 23andMe profile, I want any egg that we're going to use to also have the 23andMe profile." So these doctors were pretty stunned, and weren't quite sure how to deal with that.
On the positive side with this ability now to do PGD, which is preimplantation genetic diagnosis, where – let's say a hypothetical. A couple gets their genetic data from 23andMe, and there are all these carrier reports that you get. If they find out that they're both carriers for the same rare disease, then certainly I think it makes absolute sense that they would want to choose an embryo for implantation that the child was not going to have this very rare disease. This doesn't come up that often, but that's very doable, and it's a very positive thing I think for society to be able to use this information.
Then you go on the other end of it with "Gattaca," and that movie still comes up all the time. [laughs] It's a fabulous movie. But the possibility of being able to say, "I want my baby to have these 15 different things...." when you do the math of how many times you have to put an egg and a sperm together to try to derive that? It's utterly impossible. It makes for great movies, but it's so impossible to get it to that point. I think it'll be used for these positive things, and I think the other stuff will probably play out.
Moderator: We are just about out of time, maybe we can just take one question from the audience. Let's do a quick one right down here in front.
Man 1: What was the name of the startup focusing on telomeres?
Linda: Telome Health Sciences is one of them, and SpectraCell is another that you can... I think you have to do that one through your doctor, and I don't know if Telome Health is out in the public yet, I don't know that they've launched.
Moderator: Cool. Unfortunately we are out of time, I'm sure there's so many questions that the audience would like to ask.
Linda Avey, founder of Curious, formerly of 23andMe, thank you so much for a great conversation, it's been a real pleasure to have you here.
Linda: Thank you.