AI has given humans quite a lot in terms of health assessment capability. It wasn’t too long ago that fitness and health wearable trackers were pretty basic: they could measure heart rate and a few other things, but the data itself was not tied to complex analysis.
That’s really no longer the case: as with its use in radiology and elsewhere, AI in wearable health trackers can apply its data to all sorts of amazing in-depth correlations, and insights, making us, in a sense, our own doctors. Just take hypertension, or diabetes … or sleep.
We all have to sleep. And it’s something that impacts our overall health. Poor sleep affects cortisol, blood pressure, and much more. So just being able to measure it well is a profoundly valuable health tool.
Interview
Recently, I sat down with Emily Capodilupo, head of research at Whoop, a firm known for its wearable health tracking devices. She had a lot to say about this frontier, where we’re seeing so much progress in evaluating our health with real-time tracking gear. The Whoop device is a trailblazer for modern health analysis.
In going over her entry into this exciting field, Capodilupo explained that she was looking at the high costs of sleep tests, around $2,000 per person.
“I got really excited about this idea,” she said. “What if we totally flipped the script on its head? Instead of trying to pay people $2,000 a night to come into our sleep lab, could we monitor people in real conditions, rather than under the artificial constraints of the lab? And could we get those people to buy these devices from us, and pay us to collect their sleep data?”
Now, she noted, millions of people wear Whoop’s devices.
“They give us their sleep data, and we’re able to analyze it,” she continued. “We have people who have been wearing Whoop every single night for more than a decade. We have users in every country in the world. There’s a lot you can learn when you look at data like that. It turns out that you really do see things that you can’t see in smaller data sets.”
Researching Pregnancy
Starting with the assertion that pregnancy research in general is “horrifically underfunded,” Capodilupo explained some of the reasons.
“Pregnant women are considered a protected class, and that makes it very difficult to conduct research involving them,” she said. “We respect the fact that unborn fetuses cannot consent, and there are many good reasons for those protections. But the downside is that we often don’t conduct even basic research in this area.”
Capodilupo discussed a project that Whoop did with the University of West Virginia, which she characterized as being situated at the corner of a “maternal health desert,” with the challenges that that entails.
“They’re the only hospital for several hours in every direction, so they serve an incredibly high-risk population,” she said. “Many patients don’t receive enough prenatal care, because if you have to drive four hours for an OB appointment and you feel fine, you’re probably not going to take the day off work to go. As a result, they see unusually high rates of complications, premature births, and other pregnancy-related issues. We wanted to understand what we could learn about pregnancy in this population. What was remarkable is that we also had tens of thousands of Whoop users who happened to be pregnant and were in lower-risk populations. So we partnered with them.”
What they found, she said, involved a lot of transparency in the large data set, in looking at biomarkers, changes in vital signs, and more.
It wasn’t this wild, crazy algorithm,” she said. “It was like, you just look at it, and you’re like, why do the vital signs go up and then go down? What’s happening here? It was just so plain to see. But the thing is, you need it to aggregate over tens of thousands of pregnancies, in order for the noise to go away.”
I wanted to include this longer quote where, again, Capodilupo talks about the challenges involved in pregnancy research, and why more modern data collection helps:
“Traditional academic research has made it really hard, and it’s very expensive to study pregnancy, because to get people to do anything for 9 months is hard,” she said. “And so, pregnancy studies tend to have really high dropout rates, because as you start to not feel well
and you get busy and, you know, all these other things come up, participating in a research study becomes, you know, something on the back burner, and a lower priority. And so, the ability to just track these women and learn things that had never been answered before and then ask new questions that we didn’t even know how to ask has just been really exciting.”
Re-Imagining Care
I went back to the idea of health research in general, and Emily and I pondered the kind of future that is likely to exist when we really integrate these AI capabilities into healthcare.
Capodilupo noted that we as humans are profoundly vulnerable to around 21 chronic diseases that are often the cause of death.
“What if every single night, you could get some kind of readout of: what is your risk, and how has it changed those different things and what could you be doing to offset that risk?” she said. “And so what gets really exciting is not so much that it’s going to be wildly different things. It’s that it’s going to be super frictionless, very inexpensive.”
I asked her if future advances will help us to be able to predict disease before symptoms start to show. Capodilupo believes this is coming, and that we’ll start to see what we can do to live longer.
“I’m really excited for people to stop thinking that things that are behavior modifiable are
random genetic bad luck,” she said. “There are so many things that because we don’t
understand, we just say, “Oh yeah, you know, sometimes that happens” … and the reality is: there are certain things that are actually random bad luck, but so much more is within our control and within the control of medicine, if you catch it earlier. And so I think what we’re going to start to see is people feeling like they have enormously more agency.”
I found all of this to be quite inspiring. We will likely be able to revolutionize health, as Capodilupo imagined, by making what is now cutting-edge available through insurance, and knocking down barriers to better personal health management. Think about how this kind of thing will change our world.
