As part of TBI’s new Innovator Series we interviewed Professor Michael Snyder from Stanford Healthcare Innovation Lab. Professor Snyder is also the Chair of the Genetics Department and directs the Center for Genomics and Personalised Medicine at Stanford, and is currently conducting a study to ascertain whether data collected from wearable devices can be used to predict the onset of an infectious disease such as Covid-19 before symptoms start.
The Institute has written a briefing on this in collaboration with his lab, while a video of the interview can be found here. The following is a lightly edited transcript between Professor Snyder and the Institute’s Head of Technology Futures, Benedict Macon-Cooney.
Benedict Macon-Cooney: Thank you so much for taking the time. Just by way of introduction, tell us a bit about you, the innovation that you're pursuing at the moment in wearables and what role you think it has in the fight against Covid-19.
Professor Michael Snyder: The project we're focused on now is really to try to do early detection to Covid-19. It stems from our wearables work, where we discovered you could actually tell them people are getting sick before they realise that because their heart rate went up and other physiological measurements change.
What was special about it was that you do it all from a smartwatch. So in fact, we discovered in my Lyme disease of all things from the fact my blood oxygen dropped, my heart rate went up and I later learned my skin temperature went up. Some of that was from a smartwatch. We also set a pulse oximeter at the time. So then we went on and figured out that actually every time I got ill – even when I was asymptomatic – my heart rate and skin temperature went up and we picked it all up with a smartwatch. It was true for other people as well, certainly the heart rate. Now, during the pandemic we're really going out full force to try to basically do this at scale. The goal is to do it for millions of people.
And what we want to do is first training our algorithms we've now pulled in a lot of data sets of people have had COVID-19, and we're training our algorithms to get them optimised. We've shown that about a third of the time we can pick up disease early before they're symptomatic, another third of the time it's around the same time as they’re symptomatic, and there's a third we're a little bit late on. Some may pick up at time of diagnosis, some after that. We’re still trying to improve the algorithms but nonetheless even if we catch two thirds at or before they're symptomatic it's a big deal.
In fact, one case we found nine days, nine and a half days before they felt symptomatic their heart rate went up. It's pretty clear they've been sick for nine and a half days before they knew it.
BMC: One of the things from the public policy perspective is how difficult it is to stop transmission happening if people are asymptomatic. Even if you have smart sensors and thermometers, it feels like it's never going to be complete unless you kind of have some form of individual monitoring based system.
Professor Snyder: Yeah, we believe that 100% because what's special about smartwatches on the wearables is they're measuring you 24/7. They literally make hundreds of thousands of measurements each day and so they're following you all the time. Right now, our window is six hours. We think we can tell when you're getting sick in about a six-hour window.
What we do, it's all built on the individual: you try and figure out people's baseline measurements or their resting heart rate, and then when they start going up from that in a statistically significant fashion we call a signal that will say ‘look your heart rates up more than it's ever been in the last two months’. Something might be going on. That's the alert function.
We're still optimising and I hope within a few weeks we're actually going to start rolling this out initially to a small number of people, maybe a few thousand, and then the goals is to millions.
BMC: One of the things that I've been so struck by is how behind the times medicine is compared to technology. This example you use of temperature is fascinating because our perception in the public is based on a German physician in 1851 sampling 25,000 patients. Whereas you've talked about this a lot, but there's a quite a wide range and actually it's not even the range that matters so much as your deviation from your baseline norm. Can you talk about that within the application of wearables, but how this also fits into your view of individual-based rather than population-based medicine.
Professor Snyder: Yeah, so the way medicine is done today nearly every decision about your health is based on population based averages, meaning this is what the population does as a whole and so therefore this is how we're going to view you and treat you. The classic is oral temperature. The party line is your oral temperature is 98.6°F. But it turns out if you actually look at studies, it's not 98.6°F is 97.5°F. But the bigger issue is it's a spread as you just pointed out. It ranges from 94.6°F to the 25th quartile, so 25% of people around 94.6°F, and the 75th percentile the 75th is up at 99.1°F.
That's quite a spread and this is their normal healthy temperature. So in today's world, if your normal healthy temperature is 94.6°F, and you walk into a physician's office and they measure your temperature at 98.6°F, they tell you you're healthy, everything's fine. But you're at four degrees Fahrenheit over your background, I guarantee you're not healthy. You’re ill if you’re up four degree Fahrenheit. So we just need to know people's resting baseline, the baseline for everything.
And that's the new mantra if you really want to understand your health, you got to know your baseline.
You can't miss shifts from the baseline that's what our heart rate is all about as well. We know people's resting heart rate, and then when it goes up for some prolonged period. Something's wrong if your heart rates up and it only has to be a few beats per minute. We think it's about two beats per minute higher than average and you'll be able to tell that pretty easily.
BMC: So the algorithm that you're developing for the wearables is based on heart rate?
Professor Snyder: The algorithm we're building is dependent on heart rate, although you do want to bring in heart rate variability and other measurements as well. To the extent we can get some different watches we are going to be watch agnostic, so to speak. We're going to work with Fitbit, Garmin, we are already with Apple Watch, and they do measure different things. They all measure heart rate, heart rate variability, so we'll be measuring those.
But some of them will measure SpO2 (Blood Oxygen Saturation) to delta, meaning a change in your blood oxygen. Some of them will measure skin temperature, some don't. We'll bring as many measurements as we can. We know heart rate is a good one, because when you get ill, your heart rate goes up, even when your skin temperature doesn't. We think it actually might be a better measure for illness than skin temperature, although I'm sure the two together will be better the one alone.
BMC: There's varying degrees of measurements and metrics that wearables can take today and so there's varying accuracy within the different elements. But I guess this is quite similar to different elements of research and analysis: you want a sort of a multivariate reading for it to be as accurate or precise as possible and try to get that perception that something might be wrong with someone. So you don't want to just take heart rate, you also want to look at blood oxygen and these are some of the things that we've seen around Covid-19, particularly on silent hypoxia. Can consumer wearables begin to do all of this in the near future to an accurate degree or is it that some elements are a bit further ahead of the curve than others at the moment?
“We're using very old technology for a 21st century problem. We're using oral thermometers and that's just incredibly Byzantine in today's world.” Michael Snyder”
Professor Snyder: Yeah, great question. I absolutely agree we need all the measurements we can to get the most accurate picture of someone's health. Some of them are quite accurate right now: resting heart rate and heart rate in general is quite accurate off a smartwatch, as is heart rate variability.
Some of them aren't so accurate like skin temperature and blood oxygen. Their absolute values aren't accurate, but it turns out that they can still pick up changes, so even if it's not quite getting your skin temperature quite right, when you get ill, it will shift and you can still pick that up. So we need to think about these things more in terms of their deltas, their change from normal, and that part can be quite accurate, even if the absolute measurement is not.
So, that's what we want to do, but I also think we can measure all these things you described. We can measure, respiratory rate we can measure, heart rate, heart rate variability, skin temperature. There's other things that are probably going to be valuable too: there's something called galvanic stress response, which measures conductance on your skin, and that's probably going to be pretty important for infections. We're just not used to dealing with that kind of data so nobody talks about it, but I think we'll be bringing more and more new kinds of measurements.
I liked your comment that we're you know we're using very very old technology for a 21st century problem. We're using oral thermometers and that's just incredibly Byzantine in today's world. There's so many other things we could be measuring to get a much more better and much more accurate picture of people's health. As a friend of mine likes to say you know the method we're using the most right now is a 14th century method, quarantine, for dealing with the pandemic. That is a good thing to do, don't get me wrong, but you know we're capable of doing so much more with all the advanced technologies that are around these days.
BMC: As you’ve said, wearables can take 250,000 measurements a day, which is just an extraordinary ability to do remote monitoring. To my mind, it's odd that we haven't pursued this more aggressively in healthcare delivery so far. Some of the reasons, we've had about people getting worried about false positives and alerting and overburdening doctors and such feels like more a systems design question that we can quickly overcome. But I just wonder how you think all these things sync together and what have been the barriers to getting them in place so far.
Professor Snyder: Well there's a number of barriers. One is the medical profession itself is very conservative by nature, it doesn't like new technologies. They also worry a lot about people having control over their own health. I think most doctors are scared of these devices that are direct to consumer, because they are worried it will turn them into a hypochondriac. I think it's entirely the wrong way to look at it. These things can simply augment everything you're doing. You teach people - people are eager to learn this stuff and they can be incredibly lifesaving. You've heard it from the Apple Watch and the heart problems.
They come up over and over again. I've had people come up to me and say they've seen their heart rate go up when they've been ill, and it was the first sign. We’ve just got to get used to the idea of people taking control of their own health and educating doctors that these are valuable and not working against them, they're working with them to help manage everybody's health. The fact that they can measure you all the time is just incredible. You can always be following your health state without even trying.
But there is also this old mentality that there's no reason to go to the doctor if you’re healthy. The culture is just set up that way and certainly in the US. It costs money to go to a doctor; there's nobody who pays you to keep pays to keep you healthy, which is the number one problem, the only way people get paid normally is when you're ill, and you go visit the doctor and get some care get some prescriptions and things, but it would be more much more cost beneficial if we actually taught people healthy.
It's a cultural shift. If you think about those very few areas in medicine and life where there's a lot of effort put to keep itself healthy, dentistry is probably the best example of this. You brush your teeth, you actually take measures to keep your teeth healthy. You go to a dentist often before you have cavities and things to get checked up. Medicine should be like dentistry, and that came online about 100 years ago. Medicine’s probably about 100 years behind teeth cleaning.
“Medicine should be like dentistry, and that came online about 100 years ago. Medicine’s probably about 100 years behind teeth cleaning.” Michael Snyder”
BMC: That’s quite terrifying, but you see this across different realms of public policy. For example, in homelessness, the cost of actually keeping people off the streets is less than the cost of then treating them for all the knock-on consequences. It's sort of weird that we don't begin to sort of price that in from a moral perspective, or even on a basic cost benefit analysis. But the question from our mind is how we begin to get there. Tony Blair announced the completion of Human Genome Project in 2001 and you've seen the growth of the industry but it's still taken a reasonably long time. But now there is also a confluence with other elements that are now happening, including Artificial Intelligence and Machine Learning, which just completely, opens our ability to analyse and to understand like biomedicine on a much bigger level.
How we should think about all these little elements, whether it's from the individual or monitoring through to how we can unlock this data for greater understanding and discovery, it's just a much bigger priority and I wonder if Covid-19 has opened up some of those questions in a positive light, despite the tragedy?
Professor Snyder: It is an incredible opportunity. If you think about it, there was no reason we weren't doing telemedicine before all this happened. That was happening, but in a very very small scale. And the whole logic, it makes no sense to go to a doctor and get sick. It's illogical. We should really be doing all this, as much as possible the way you and I are speaking right now. There are times obviously when you need to go to a doctor, especially when you need access to large equipment, things like that, but much of it can be done remotely and I think and it's again very logical keeps people healthier if they do it remotely. But it is a cultural shift, you've got to get people thinking about preventative health. And that just hasn't really stuck and I think a big part of it is something you said earlier: it’s the economics.
You have to show it's cost effective to keep people healthy and this is what you lose by people not being healthy. We need more studies of that sort, to show the savings you can get by keeping people healthy. Diabetes is a good one: I think they estimate that all the satellites of the effects of diabetes, not just the health ones, but also the time not at jobs and things like that it's getting close to a trillion dollars a year. So that's a lot of money. You need to make those kind of economic arguments. Now, suppose we took 10% of that and prevented diabetes, wouldn't that be an incredible savings?
We need to make more compelling arguments. That's upon us, to be honest, the scientists, the policymakers and others. I think one nice thing about the new revolution is some of these devices, you mentioned the information we can now get and the whole AI part. This is now getting very accessible to people. You can get a smartwatch and they're pretty darn cheap. You can get a $100 smartwatch, and quite frankly I think we could put one on everyone in the world if you wanted to. You'd have some level of health management and it's not a bad level to be measuring someone's heart rate and probably a few other things all the time. We'll give some level of health care they currently don't have in many remote parts of our country in the world. There is incredible technology opportunities now and you're right, the Covid-19 has really brought a lot of this to light.
I think that, hey, let's try and fix all these things that are broken with our health care system. Like who pays and the whole preventative medicine side of things. Let's work on keeping people healthy and, again, not react once they're ill.
BMC: I really agree with the idea that states provide some form of monitoring, particularly for vulnerable groups. From a family-based perspective you might also be worried about an elderly parent for which it can provide some alert system.
Professor Snyder: You hit a really good point if I can add to that. It makes a lot of sense, so parents for their kids, or for people who take care of the elderly, to have monitors on these folks. It's a no brainer because they're the ones are making a lot of decisions and checking up on them and it's a very easy thing to do. I live on the West Coast, my Mom who was 91 lived on the East Coast and she would forget the hang up her phone and, so every time - I'm in a family of six kids - we'd all scramble and say nobody can get a hold of Mom, what's going on here. We're pretty sure she probably just left her phone off, but we’d have to call someone up to actually go check on her, to make sure she was okay and she normally was. But if she’d had a simple smartwatch on or, we would have been able to say, yeah her heart rates fine, everything's good, it looks fine.
It’s just very simple things right, that could actually save a lot of people a lot of time and a lot of effort. The way some of this will roll out, it's not the ideal way, but the way some of it will roll out is now people are thinking about using some of these technologies in outpatient clinics, where if you're releasing someone from the hospital and you still need to monitor them. First of all, they'll probably recover better at home, that you can still keep an eye on them with some of these technologies so I think that's one way and you show them that it's cost effective. That hits the current medical system because it's just hard to change the medical system entirely. You have to do a bit at a time. We can show cost effectiveness in certain situations and then I hope it just spreads like wildfire to everyone.
BMC: That's such an interesting point because you can change like your model of care delivery if you get this right. You alluded to this a bit earlier, but the number of infections people sometimes get in hospitals is quite shockingly high, but you can begin to design the systems that can do a lot more care remotely. From the broader public policy perspective this might mean less visits to your doctor's or to hospitals and it can change the whole infrastructure around it in a positive way but at the same time actually making care delivery much closer to the individual. This can remove a lot of friction and costs to the system.
Professor Snyder: I would agree 100%. It's like having a caregiver with you all the time if you've got a monitor on you, at least at some level. That's what you're getting for free with some of these devices and they don't always have to be smartwatches there's other very cool technologies out there as well. We're just at the starting point, but we do have to get people to be willing to adopt it and realise the value of these technologies, and also again realise the value of keeping people healthy.
One other thought related to the pandemic is that when you catch someone early by one of these things it makes tracing a lot easier. If you know when they first got ill, and if it was eight days or nine days before their symptoms show, you have to actually track all those eight and nine days, not just the time when they were symptomatic. So you can actually get much more accurate in tracking infections as well, in principle. It hasn't been done but there's no reason not to. I think we can just get better at both helping the individual but also in watching out in cases of pandemics.
BMC: I guess the utility of this isn't just to help with the immediate crisis but you can monitor for other viral infections in the future as well. I can imagine a life where we quite think quite differently about viral infection and if we see a spike in heart rate we might not go into work, because we now know that we can work remotely and not infect however many x other people, and it can provide that way of thinking about how we manage our own health as well as the people around us in a much more effective way.
Professor Snyder: Yeah, when we first described this, which was about three years ago now everyone's reaction was, Mike, that's nice. So I tried to tell them what you just said, that well maybe you don't send your kids to school and have them infect everyone else in the classroom or you stay home for work and not infect all your colleagues. They said, yeah, that's fine, but I think again what brought the awareness is the Covid-19. You really don't want to be spreading this stuff, people will die if you don't manage this properly. So unfortunately sometimes it takes a wake up call. But I think these will be some of the long-lasting effects. There'll be no reason not to know when people are getting sick at some level, at a much earlier time. I'm not saying it's going to be perfect. I don't think there's any detection system that is and even the tests I use now for Covid-19 have both false positives false negatives. There's some papers out that say even by PCR you miss the first several days of infection. So, no test is perfect and so I think we want as many different types of tests as possible to best catch illness at its earliest possible time.
BMC: That's a really important point because I think often one of the barriers to progress is that people want it to be absolute and failsafe. You see this with some many elements of sort of technology, and it's absolutely right to try and make sure you're trying to make this as universal as possible but there's this sense that, if some people don't have it, so we shouldn't do it at all. Whereas actually you need to sort of build progress incrementally and try have all these pieces fit together. Some people, in particular vulnerable groups, may have the state provide a wearable, whereas people like myself might take on that bit more of that responsibility for themselves. We need to think about that as both the stack of technology and the stack of the system now, and how we like to try and piece all these different elements together. None on their own will provide a failsafe, but it’s about how many different measures as possible work to try and mitigate as much as possible.
“We've got to get them to everyone. They’re going to be $20 in the future.” Michael Snyder”
Professor Snyder: Yeah, there is no perfect medical test in anything. When it comes to cancer, to heart condition, we live with imperfection. We do the best we can. And in the case of the current pandemic and infectious disease you know realistically we have nothing right now. The closest thing is this PCR test which you're not going to do every day on all people. It just isn't realistic. These may be imperfect, but they'll give us some clues in a lot of cases and that can be very, very helpful. Every case counts that you can catch early I would argue, not only to the individual but to all the people they might be spreading it to.
As far as how to distribute the these technologies, I agree. We've got to get them to everyone. You want to see them go out to those who are disadvantaged and have trouble getting access. I do think, even if we've talked about £100 or $100, they're going to be $20 in the future. These are not going to be expensive devices and so they really will get out to everyone. Especially when you show utility, then it moves pretty quickly. Just by analogy the first human genome cost about a billion dollars to sequence. These days you can sequence a genome for human genome for $300. Once you get the first one going, show how to do it and prove the technology, it goes faster.
The first human genome took many many years to sequence. Now you do it in about a day. This is what's going to happen with these technologies. Once you show utility, everybody wants one, there's a lot of smart people out there who think about how to redesign them, get them better faster, cheaper, so I think we're just at the starting point for all this.
BMC: You famously sort of live what you preach, so you have multiple wearables on any given point.
Professor Snyder: So I'm wearing three smartwatches and this rings not a ring, it's a sensor. It's an Oura ring that also measures heart rate and skin temperature. I have a glucose monitor and everything we have goes into smartphones. This is now I think you're going to be the most important medical device. 60% of the world has a smartphone. We're talking Africa, everywhere. That's pretty powerful. It means everybody has the potential to get access to their own health information.
BMC: It's such a powerful point. In the UK there's been a number of barriers to even just getting good electronic health records and to sharing them between different providers. It's kind of crazy that we don't have a lot more autonomy over our own record and what we can do with it. Doctors are phenomenal people across the board generally but they're not infallible, in particular given the sort of time constraints that they are under. You get around nine minutes on average with your GP and it raises questions about whether they are going to be effectively able to diagnose you. From my perspective, I want to get to a level of health where you just have far more autonomy, more understanding of your own personal health, and have a very different relationship with your doctor.
Professor Snyder: Yeah, the number they use in the US is 15 minutes so we get an extra six minutes. But it's the same problem. It's not nearly enough time to really know your health and, like you say, they're doing the best job they can. They're incredible people, but it's just not enough time. No one can know your health like you can know your health, and then having devices and AI and things to help improve that help manage your health better, it's just totally logical. I think that's where it's all gonna go. A computer can store so much more information, can synthesise it, can bring it together. Also when you go to a physician's office, some of the measurements they make are actually much less accurate than what you'll get off of your smartwatches.
A good example is heart rate. My heart rate will vary by 40 beats a minute, depending what's going on: if I biked to the doctor's office versus drove, even if I sit around for a while, it never comes all the way back down. Depending what else is going on, it can just vary hugely. Whereas if I pull my resting heart rate of my watch first thing in the morning, it's always very consistent unless I'm ill, then it'll jump up. Or maybe I'm hyper stressed. My watch will tell me that. The reality is the measurements you can get from these devices can be much much better than what you'll get in the physician's office, so it's a little bit of a fallacy to think that what they're measuring is better than what you to measure at home.
If you think about the world in general: in my family nobody ever goes to stores anymore, they order everything from Amazon. I see no reason why a lot of that won't happen with healthcare as well. Again, not all of it, I think there's still times when you need to see a physician, certain instrumentation certain expertise can be handled pretty well in a doctor's office, but a lot of it can be done offline, can be done in your own home.
BMC: I don't want to keep you for too much longer, but is there any elements you think are important to raise?
Professor Snyder: Yes. It is an amazing time and to go back on something you said if you think about it when you hit these situations where there's gonna be a lot of unemployed people, a lot of people won't have health care. It's a time to shake up the system, how do we do this right, let's work on keeping people healthy while they're healthy and not wait for them to get sick and try and try and fix them. It is a time to try and change things. There's a huge push in the whole healthcare industry now to solve the pandemic to figure out how to better manage people's health. I can't think of a better time to change the world. I hope in the process we don't leave anyone behind. I hope we can do it for every person on the planet, and I know we share goals that way. I’d love to see it happen.