Transcription
Our health care system is broken. Our laboratory in the last 17 years has been trying to improve it. And along the way, I've even learned a lot of new things about my own health.
It's broken because we practice sick care rather than health care. And even when we practice health care, if you think about it, it's a pretty archaic process. You will travel to a physician's office that really hasn't changed much in the last 40 years. When you're there, they'll take lots of blood, and from all that blood in the time you're there, they actually don't make very many measurements. And from those measurements will treat you based on population averages, rather than the individual. We're trying to change all those steps.
And one area in particular is remote health monitoring. What if you could measure your health with sensors and a dashboard, much like your car dashboard? That's what we've been trying to do. And you can now do that thanks to a revolution in wearable devices, smartwatches, rings and [a] host of other devices. These devices will measure things like resting heart rate, heart rate variability, skin temperature and conductance, blood oxygen, even sleep. All of these things we know are very important for our health, and we can now measure them continuously 24/7 as long as you keep them charged.
I'm such a big fan of these devices, I wear eight of them every single day. I'm wearing four smart watches, two rings, even my hearing aids are sensors. They measure physiology like these other devices, but they also measure my socialization, which is important for your health as well. So you can track all of these things in real time. And we've been doing just that.
When they first came out as fitness trackers, we put them on a cohort of people we were studying for health tracking, and what we discovered is that you could tell when people get ill in advance of symptoms from a simple smartwatch, in this case, smartwatch and pulse ox. The first case was when I discovered my Lyme disease of all things, with, in fact, a pulse ox and a smartwatch, where my blood oxygen dropped, my heart rate went up. It was all pre-symptomatically. And then I did visit a physician who actually diagnosed it as Lyme disease. And in the end, we got it treated early and I wound up not getting anything serious. I was cured right away simply because of early detection. And so that turned out to be the first case.
And then we went on to show that you can detect respiratory viral infections with a simple smartwatch, because your heart rate jumped up in advance of symptoms. When the COVID pandemic came along in the spring of 2020, we built a real-time detection system that follows your baseline. It looks for a jump up in resting heart rate and other measures, and then sends a red alert. And it turns out this alerting system worked very, very well for COVID. Because COVID has a long, pre-symptomatic period. It will detect COVID 80 percent of the time, with a median of three days in advance of symptoms. Now it also is very sensitive. It detects asymptomatic cases. I'm showing a case up here where someone was diagnosed with COVID, but they were getting red alerts for two weeks prior to the diagnosis. It’s not specific for COVID, though -- I should warn you, it does pick up other viral infections, which we think is a good thing, but it also picks up other stressors. And the number-one trigger of red alerts is workplace stress. So it actually is both a physical stress detector with viral infections, but also mental health detector. And we're actually exploiting this further to try and tell when people get anxious and have depression from a simple smartwatch.
Infections are not the only thing you detect with these devices. My colleagues at Stanford have actually shown you can pick up a-fibrillation, which is a heart condition that if you catch it early, you can treat it. With machine learning and AI we actually were able to show that smartwatches are pretty sophisticated. They can actually tell your red blood cell count and other measures -- things associated with anemia and hydration, but also even your blood glucose and hemoglobin A1C, as it's called, which is important for diabetes -- you can also pick up with a smartwatch, thanks to the skin conductance measure on these things. So these are very sophisticated devices that can make many measurements as early signs of disease.
They're not the only devices that are out there. Continuous glucose monitors are very, very powerful too. They'll measure your glucose every five minutes. And we have a diabetes epidemic going on that's worse than the COVID pandemic. 11.6 percent of people in this country are diabetic, in the US, and 20 percent of them don't know it. Thirty-eight percent of people are pre-diabetic and 80 percent of those don't know it. And those numbers are going up, by the way. And the prediabetics, most of them will become diabetic. So we think it's very important to catch glucose dysregulation early so that you can prevent diabetes. And it also turns out glucose spikes are associated with cardiovascular disease, which is the number-one killer in the US.
So we actually put these on so-called normal people and prediabetics. And we did discover that some people are normal, they have good glucose control. Some people are moderate glucose spikers, so they spike somewhat. And some people are severe spikers just as bad as diabetics, but they didn't know it. And we can pick this up with the glucose monitor. We then went on to discover that different people spike to different foods. Some will spike to bread, some to pasta, some to bananas. Here's an experiment we did recently where we had 55 people eat seven different carbohydrates, identical amounts. But the carbohydrates were in different forms: beans, berries, grapes, bread, pasta and white rice. It turns out most people are rice spikers like the one in the upper left. In fact, rice is worse than ice cream for most people. Some people, though, are pasta spikers, some are potato spikers, bread, grape spikers. We're all different.
What's going on? We think this has to do with the heterogeneity of type 2 diabetes. It's not just type 1 and type 2, but type 2 has many subtypes. And that's because our glucose regulation is very complicated. We have our pancreas, we have our liver, our muscle, even our brain is a major glucose consumer. And we have many biochemical pathways. Insulin, the GLPs you may have heard of these days, are also important regulators. And there are other hormonal systems as well. And together, these control your glucose. Turns out, some types of diabetes exist that are actually very evident when you actually start looking at this.
And so we took prediabetics and normal folks and typed them for their glucose dysregulation. We typed them for muscle insulin resistance and hepatic insulin resistance, as well as beta-cell defects and incretin defects, the GLPs. And it turns out, once again, everyone is different. So the person at the top arrow turns out [to have a] muscle insulin resistant defect. The one at the bottom arrow has an incretin defect. I myself am a type 2 diabetic. You might not have guessed it. I’m beta-cell defect. We're all different. You can actually tell the subtype of glucose dysregulation you have by drinking a shot of glucose and looking at the shape of the curve. And using machine learning, we can now tell your subtype just from the shape of that glucose curve, whether you’re muscle insulin resistant or have a beta-cell defect. And so what used to be a thousand or several thousand dollars set of tests you can now do for 50 bucks from a local drugstore.
Not only that, you might say, why do I care about my subtype? Well, it turns out your subtype determines the food you'll spike to. So if you’re muscle insulin resistant, you'll spike the potatoes and pasta, but not if you're insulin sensitive. Likewise, if you’re beta-cell defect, you’ll spike the potatoes. You may know that if you actually eat your salad before your French fries, you can generally reduce your spikes, but that's not always the case. We tested this further with regards to subtypes. So we had people eat protein, fiber or fat prior to eating white rice, which spikes them. And it turns out that if you’re muscle insulin resistant or have a beta-cell defect, nothing happens. You'll still spike when you eat your white rice. But if you're insulin-sensitive or normal for beta cells, you actually can suppress those spikes with fat and fiber. So once again, your subtype actually determines not only what foods you'll spike to, but how to mitigate those spikes.
We took this to extreme by actually looking at a group of people who were subtyped. And then we actually looked at their glucose by these continuous glucose monitors, as well as had smartwatches on them. And the idea was to track what they do and when they do it with their effect on glucose and according to their subtypes. So just to get into this a little more detail, we learned some things that you might already know. If you eat your first meal, your biggest meal, first thing in the morning, you'll have lower glucose. If you have it late at night, you have higher glucose. Starchy vegetables, higher glucose. Fruits and less starchy things will give you lower glucose. It turns out most people don't sleep enough. So if you sleep longer, you actually lower your glucose. The same logic can be applied to these subtypes. It's important to know your subtypes so you know what lifestyle things will mitigate that problem. And you can actually ... take this one step further. The medications I use are also very dependent on my subtype. So I respond to certain medications that promote insulin release from my pancreas, but not other kinds of medicines. Actually, I don't respond to metformin, which is the most common medicine. So these glucose monitors are very, very powerful, we think.
What about other biochemical measures? Well it turns out, we've been working very hard to see if we could detect molecules from a single drop of blood. And after spending seven years on this, we set up a method for taking drops of blood from your fingertip and your shoulder, and being able to measure 7,000 different molecules from that drop of blood. Now, I know what that sounds like, but ours actually does work. (Laughter) Then we could do, once again, fun experiments with this. We were able to have 32 people drink the shake, very common shake, and looked at their response. And once again, everyone was different. We had some people [whose] carbohydrates went down, others went way up, others stayed flat. One of the interesting results was inflammation. It turns out when some people drank the shake, their inflammation went down. Others went up. Same shake, pro-inflammatory on some, anti-inflammatory on others. And we think this is important to know if you have digestive issues from the food you eat, which turns out to be about 10 percent of the US population.
We've taken this to the extreme where we basically had one individual take a micro sample every hour for seven straight days during waking hours. And the idea was to try and correlate once again, what they do with its effect on their physiology and biochemistry. And what we discovered is lots of associations -- thousands actually. But one of the interesting was, we discovered that alpha-synuclein, which is involved in dementia and Parkinson's, shows a very interesting pattern. It correlates with certain stresses. So the idea is that if you could actually mitigate alpha-synuclein spikes, what you might do is actually be able to push off people getting dementia or Parkinson's.
Where is all this going? Well I envision a world where all of you, many of you are already, will actually be doing remote monitoring in the form of sensors and microsampling to get yourself measured more frequently and to stay healthy. The nice thing is these technologies are very inexpensive. They could go anywhere around the world. The whole world is capable of wearing these things and using microsampling, and with this, we can actually better manage their health and be able to even hit underserved populations. And then together with other technologies like genome sequencing and yet other sophisticated advances in science, we can better predict disease risk, manage people's health better in terms of early diagnosis, monitoring and treating disease. And the ultimate goal, of course, is to have everyone live long, healthy lives. Thank you. (Applause)