Does AI know you better than your doctor?

Note: Transcripts are automatically generated from episode audio, and are not fully corrected for spelling, grammar, and formatting.

TOM HALE: If we have a large physiological model that makes predictions about you, and those predictions aren’t just, like, “Hey, you’re getting sick, and in three days you’re going to be on the couch.”

It’s more like, “Hey, in 10 years, if you don’t change your course, you’re going to have a heart attack.” That’s so powerful. It’s powerful from so many different dimensions. We’re going to see more of that healthcare trickling to the digital edge, to the experience you’re having on your device with your wearables, and it’ll be automated.

There’ll be an AI doctor in your pocket or in the cloud who is looking over you and using your large physiological model to understand where you’re headed. And that is a transformation in medicine. I think it’s going to be incredible.

RANA EL KALIOUBY: That’s Tom Hale, CEO of Ōura — the ring that measures and tracks all kinds of health data. You’ll find Ōura rings on the fingers of everyone, from elite athletes to patients managing chronic illness.

Since Tom joined in 2022, Ōura has grown to a reported $11 billion valuation. Their ring has become super popular, and it’s helping people take greater agency over their health.

In this episode, Tom and I talk about how wearables, like the Ōura Ring, give your body a voice so you get the data before your doctor does, and how the trifecta of sensors, data, and machine learning unlocks true personalized, preventative health care.

I’m Rana el Kaliouby, and this is Pioneers of AI, a podcast taking you behind the scenes of the AI revolution.

[THEME MUSIC]

Hi, Tom. Thanks for joining us on Pioneers of AI.

HALE: Hi, Rana. Thanks for having me on the show. I appreciate it.

Copy LinkHow wearables can lead to better sleep and energy

EL KALIOUBY: I want to start with a full disclosure that I wear a Whoop, just because my son—

HALE: Oh, sorry. Wrong podcast. Sorry. We’re on the wrong podcast.

EL KALIOUBY: No.

HALE: No, I’m kidding. You know what? I’m glad you’re wearing a wearable. That’s really important. It’s important for you to have some knowledge of your body and your sleep and all that stuff. It’s powerful.

EL KALIOUBY: I got started on wearables because of my son five years ago. He’s an athlete, and he started tracking his sleep and his physical fitness and whatnot. Now we’re obsessed, and I’ve learned a ton from tracking my sleep, including this key insight: If I eat within two-ish hours of going to bed, my sleep’s messed up.

HALE: You’re right on it, because not as many people know the trick of not eating too close to bedtime. It’s actually weird. Your body almost incentivizes you to eat closer to bed if you stay up past your bedtime. For me, I don’t know if you know the story, but—

EL KALIOUBY: I don’t. I’d love to hear it.

HALE: For me, it was the fall of 2021. COVID was still in effect, the kids were struggling, and I was struggling. It was a tough time. There was a lot of stress, and I—

EL KALIOUBY: Were the kids teens at the time?

HALE: Yeah, teenagers, sort of making the transition into being teens.

As you can imagine, it’s tough anyway. Throw in a dose of COVID, and it’s extra tough. But the point is that I’m a championship sleeper. I can fall asleep at a moment’s notice. I stay asleep for a long time. I’m just really good at sleeping.

But at that moment in my life, I actually lost sleep. For the first time in my life, I was waking up in the middle of the night. I was having trouble falling asleep. I was having trouble staying asleep. I started researching solutions, and I bumped into Ōura.

I got an Ōura Ring, and I’d been a wearables person for quite some time. I’ve always had a wearable or two, and this one was really different. What happened is I changed a couple of my behaviors — the obvious ones. I stopped drinking coffee late in the day, stopped drinking alcohol to take the edge off the coffee, and started trying to go to sleep without staring at the phone before bed.

Don’t watch Netflix. Netflix and sleep compete for your attention. I started sleeping in a cold room, and I started really getting into how you optimize around sleep. Let me tell you something: When I recovered sleep, it was as if I was walking out of a black-and-white television screen of my life experience into a full-color 4K Technicolor movie.

It was amazing. I literally was asking people, “Do you feel like this all the time? Is this how people feel?” It was amazing, and it was such a transformation for me.

EL KALIOUBY: From an energy boost or mental clarity? What was it?

HALE: It was mood, it was energy, it was mental clarity, it was modulation of your energy during the day. It was a really profound transformation for me. And I was like, “Wow, this product is really affecting people’s lives. It affected mine.”

I need to be a part of this company.

EL KALIOUBY: I have to say, it’s incredible to align something you’re deeply passionate about with your everyday work. Was that intentional?

HALE: Well, yeah. That’s always been my intention. I’m sure there are people who are very good at their jobs but don’t love what they’re working on. I’m not one of those people. I have to find something that I love, because if I’m going to put my heart and soul into it, it’s got to be something I’m passionate about.

For everything that I’ve worked on, that’s been true. This is a fascinating domain. It’s like physics and physiology and human behavior and health. There are so many things that are just fascinating about it.

But maybe most importantly, the company itself is mission-driven, with an idea that it can play a role in rethinking the transformation of healthcare from a sick-care system to proactive preventive medicine. An ounce of prevention is worth a pound of cure.

You can get people to change their behaviors, and you get people to understand the kinds of behaviors that will mitigate in favor of long-term health. Some people call it longevity, some people call it health span, or they say we’re shortening the time of sick span toward the end of someone’s life.

There’s obviously a benefit to all people. If you can really get people doing healthy things, they will be healthier and, I think, ultimately happier. If you can do that, you’ll also bend the curve on the cost of healthcare.

In our society, in our country, one out of five dollars is spent on healthcare. And I don’t know about you, but your healthcare experience isn’t great.

HALE: It’s all oriented toward the end of life.

It’s all about those last 10 years. Boy, wouldn’t it be great if we could do a better job? So I think that sense of mission, for me, honestly, was very attractive.

EL KALIOUBY: I want to double-click on that because one of my investment theses is that I like to fish at the intersection of profit and purpose. And I truly believe we invest in early-stage AI companies that unlock human potential. I think it’s not easy to find the intersection of that Venn diagram, but when you do, it’s really powerful.

EL KALIOUBY: So you hit a billion dollars in revenue, and you raised nine hundred million at an eleven billion dollar valuation. Congratulations.

HALE: Thank you.

Copy LinkWhat is driving the wearable health boom

EL KALIOUBY: What’s behind the jump in interest and sales from your perspective? What changed?

HALE: I think it’s a couple of things. It’s never one thing. Part of it is just the flywheel of customers. We have quite a few customers, and they’re out there telling their friends, or their parents, or their kids, or their husband, or their partner, or whatever coworker, that this is a good thing and they should do it.

Many times, it’s a gift of care. I want you to take care of yourself, so I’m giving you this tool. Maybe to address this upfront, I think a lot of people think of wearables as fitness. You should know how fast you were going and what your heart rate was and all that stuff, and Ōura does that as well. But maybe uniquely, Ōura is oriented more around health and, maybe more importantly, not just health in the moment. What’s my heart rate right now as I’m finishing my running exercise?

But what’s my trend over time, and am I improving? Am I not improving? What’s going on?

In some sense, it’s providing you this sort of sense of your physiology over time in a way that you might not perceive, because it’s actually hard for humans to perceive things over a long period of time.

Computers are very good at that. Humans are not. AI is particularly good at it. It’s also sometimes things that you just can’t sense. We don’t know what our blood sugar is because it’s just not something that we can sense. So if you have a glucose biosensor, then you can have a number that tells you what your blood glucose levels are.

So this idea of being a wearable that’s around health, that’s the first thing that I think has been central to our success. It can affect everybody.

Sleep is also a very large TAM. Everybody on the planet, we hope, is going to go to sleep tonight. And depending on how they sleep, they’re going to be at peak performance, or maybe not, tomorrow. So that’s maybe a baseline. I think the other thing to recognize is that, as a product, it has a couple of pretty interesting properties. The first is that measuring from the finger is really important.

EL KALIOUBY: Why the ring form factor?

HALE: Yeah, why the ring form factor? There are lots of reasons, but let’s start with accuracy. It turns out that your heart pumps and sends a wave of blood going down through your arteries, and it stops right here.

This is the first stop on that journey. Whereas if you contrast that with the wrist, it’s actually on its way back to the heart. It’s capillary action. Your veins actually have valves in them. They kind of close so that there’s no backwash in your blood. They close as the blood goes through.

So you’re actually getting both a weaker signal and maybe less information captured in that signal, and you’re also sensing through things like hair and bone and sinew and all these confounding factors. Whereas, of course, here it’s two millimeters of pretty uniform tissue that you’re sensing through.

For illustration, if you’ve ever been to the hospital, where do they put the sensor when they put a PPD sensor? They put it on the tip of your finger for all the same reasons. The second part of it, too, of course, is that at night, you’re looking for comfort. And so this device, which weighs all of four grams and is very low profile and doesn’t have any lights on it —

If you’re not sleeping, the last thing you want to do is look at your clock and be like, “Oh, 3 a.m.” That’s always rough. We kind of adopted this calm-tech strategy of being quiet and sort of receding into the background, and those were really essential elements of differentiation.

And so we learned, when we started introducing the product, that people would wake up in the morning and look at their phone, and they would see a score that sort of said, “How did you sleep?” And then there’d be a prediction about how your day is going to go: what’s your readiness?

That score turns out to be a really powerful score because within one second you’re like, “Hey, maybe I should push it today,” or, “Maybe I should take it easy.”

Copy LinkHow Oura rings calculate readiness scores

EL KALIOUBY: You talked about readiness, and you also have a resilience score. Can you unpack how that is calculated?

HALE: So readiness is about nighttime recovery. If you look at the components of your readiness score — and I’ll pull it up, just so that—

EL KALIOUBY: Readiness score today?

HALE: Yeah.

EL KALIOUBY: Pick a number.

HALE: Pretty good. Not bad. What we look at for readiness is your resting heart rate; your HRV balance, which is over the past 14 days, and HRV is heart rate variability; your body temperature; your recovery index, which is a ratio of how much effort you put out the previous day and how well you recovered; your sleep; your sleep balance, which is over 14 days; and your sleep regularity, which is how close to your appropriate and consistent sleep window you were.

By the way, it turns out the consistency of going to bed and waking up is actually really important because your body has a clock. It’s got a circadian clock, and it’s saying whatever time your chronotype says you should be going to sleep, 9 or 9:30, whatever.

If you miss that window, then it’s off-kilter and you get less efficient sleep because your body has missed that window.

EL KALIOUBY: Chronotypes? I’m very curious about that.

HALE: Yeah. The chronotype is sort of sleep science, but it turns out not everybody’s the same.

Some people are early birds, some people are night owls, some people are larks, some people are wolves, and they have different chronotypes. Chronotype means your optimal circadian rhythm.

I am an early bird. I like to wake up at 5. I like to do things in the morning, and I like to get to sleep by 9, and that’s my chronotype. That cycle dictates what’s optimal for you in terms of eating, exercising, cognitive function, all that stuff.

And once you know your chronotype, and it’s actually one of the things that Oura tells you, you can start to see: Are you matching your chronotype? Are you out of sync with your chronotype? Of course, what is jet lag? Jet lag is when you are out of sync with your chronotype because your body is saying, “I should be asleep right now, but I’m not.

I’m in this meeting in London, and I’m barely able to keep my eyes open.”

So we were just talking about readiness: how much activity have you had in the previous day? You look at all these things, and they’re pulled together. You can decompose them, but it gives you a single score between 0 and 100.

And the thing is, it’s not just a score. It’s also a set of insights, and the insights surface to help nudge your behavior. Maybe it’ll say, “Hey, looks like you ate too close to dinner last night. Your heart rate was elevated in the early part of your sleep, and as a result, you didn’t get great deep sleep.”

So there’s this sort of education that comes along with it. As you’re learning about yourself, you’re seeing these scores and figuring out what’s normal for you, and you’re also being educated about it. But here’s the magic: We measure you at night primarily.

We measure you during the day as well, but measuring at night allows us to set a baseline for you, which is what’s normal for you. And then, as you deviate from that baseline, that’s a signal to the system to help understand what’s going on for you, right? If you deviate in terms of sleep or temperature, or your resting heart rate is elevated, then we can say, “Oh, looks like something’s going on.

Let’s explore that. Let’s explain that.” So this idea of baseline is really important. The other thing that’s really powerful about measuring your body at night is that you more or less go through the same process every night. You fall asleep, you go into deep sleep, your temperature drops, melatonin is released by your brain.

You go into some sleep architecture. The second part of your night is usually more REM sleep than deep sleep. And then when you wake up, your temperature rises a little bit, there’s a little spike of cortisol. All these things happen, but they happen kind of consistently.

Whereas during the day, what happens is very varied, right? You might have a restful day, you might have a tough day, you might have a hard workday, you might have a recovery day. So the day is really important as well, but the signal is not as clear during the day, and it’s more confounded during the day than it is at night.

So those are part of the secret elements of Oura, which, again, the power of Oura is when we get it right, you’re like, “Whoa, how did it know I’m getting sick?” It makes a prediction three days in advance. It says, “You look like you might be getting sick.” And you’re like, “I feel fine. I’m good.” And then three days later, you’re on the couch surrounded by Kleenex, watching Netflix and drinking tea, and you’re like, “How did it know?”

EL KALIOUBY: Coming up after a short break, more on what Oura is looking for when it’s predicting what’s ahead for your health. Stay with us.

[AD BREAK]

Copy LinkHow Oura rings can reveal illness before symptoms appear

I think that is very powerful, how Oura can detect the onset of COVID or IBS flares or if you have a cold or the flu. It’s really fascinating. Is it tapping into, I guess, the difference between a deviation from your baseline sleep architecture, or is it your heart rate variability or your heart rate, or all of the above?

HALE: Again, it depends because each one of those signals might signal something different. If your heart rate lowers late in the night, that’s probably a good sign that you either had alcohol, exercised, or ate too close to bedtime because your heart rate is elevated because it’s doing the work those things stimulated.

I think the other thing that’s really powerful about measuring at night is that you’re not just measuring one metric, you’re measuring multiple metrics. So, for example, if your temperature is just a little bit elevated, and your respiration is just a little bit elevated, and your heart rate is a little bit elevated, and your heart rate variability is just a little bit off, that’s really interesting.

If they happen more or less in the same time frame, your body is probably responding to an infection.

That’s a really interesting set of signals that’s pretty clear. And the way we found this out was during COVID. We had 65,000 people who offered to share their data with Oura as part of a study with UCSF.

We do a lot of work with scientific institutions and research institutions. They shared their data, and they said, “We’ll also share a diagnosis if we get diagnosed with COVID.” And so about 9,500 of them got diagnosed with COVID, and the UCSF research team went back and looked at that data and looked for the signatures in their biometrics.

And lo and behold, two and three-quarters days before they had a positive PCR test or a COVID diagnosis, we saw these signals all in the same time frame. And so now it’s actually a feature of the product called Symptom Radar, which is basically a prediction about the onset of a variety of things, including influenza-like illnesses.

It’s not predicting what illness you have. It’s just saying, “Hey, your body’s biometrics have shifted in a meaningful way.” And that’s just fascinating because the feature is one that lots of people have taken to TikTok or Instagram, and they’ve told really fascinating stories about how Symptom Radar gave them some agency and authority in their health journey.

I’ll give you a couple of examples. One is a fascinating one. It’s a Paralympian, and he came home to stay with his family, and he’s like, “Ah, I’m not feeling great.

“I’m not feeling good.” And they’re like, “Oh, come on, you’re a Paralympian. Buck it up. You gotta train today.” They were giving him kind of the hard line, and he’s like, “I’m not feeling great.” Then he got a Symptom Radar alert, and the next day he got another one. He’s like, “I don’t know.

“Maybe I should listen to this.” Then he ultimately went to the emergency room, and they were like, “It’s a good thing you came because you have appendicitis, and your appendix was about to burst.”

EL KALIOUBY: Does insect?

HALE: He didn’t know that he had appendicitis.

EL KALIOUBY: Of bites

HALE: Yeah. Clearly his body was having a moment, and all that stuff showed up in his biometrics, and it translated into a Symptom Radar alert.

Same thing with lymphoma. There’s a nurse, a pediatric nurse — this was about a year ago. She was feeling terrible, and she wasn’t sure why. She would go to doctors, and they were like, “I don’t know. Maybe you should rest more.” It was not helpful. She kept getting these Symptom Radar alerts and kept pushing and advocating for herself and saying, “I know something’s wrong.

“Look at this. Look at this. Look at this.” And it turned out she had early-stage lymphoma. The thing is, part of it is just her having the agency and knowledge of her own physiology and, call it the independent kind of authority that’s saying, “Yeah, something’s really going on,” gave her that—

EL KALIOUBY: Data, third-party data, right?

HALE: Third-party data. In some sense, your body doesn’t lie. It can’t lie. Whereas doctors know this: patients are remarkably forgetful, or sometimes they can’t translate what they’re experiencing into words, and part of their job is to extract that. It’s really great if you have information that’s objective and continuous and high quality and accurate and maybe contains some information that you might not have access to.

Copy LinkWhy continuous health data matters for managing chronic illness

EL KALIOUBY: For a patient or a user with chronic illness, what kind of insights does the Ōura Ring give that are really fundamentally important to their health journey?

HALE: Yeah, it’s probably two or three things that are the most important and the most common. One, of course, is the signs of an oncoming infection, right? So if you’re immunocompromised, maybe you’re going through chemotherapy, maybe you have ME/CFS, whatever.

For you, a cold is a real problem. It’s not just an annoyance. It’s a real problem. So having that kind of forward visibility turns out to be very useful. Readiness is the next one, because readiness is kind of a proxy for the state of your body’s ability to go through the day.

EL KALIOUBY: Right?

HALE: Yeah, like, are you recovered? We hear this from people all the time, that they use readiness as a barometer of their day.

“I have ME/CFS, and I need to be careful about how much energy I expend.” Then there are a variety of other things. People look at their temperature. They look at their step count. For some people who have heart-related chronic conditions, they’re looking at heart rate or something that we do uniquely, which is your pulse wave velocity.

Pulse wave velocity is a good proxy for arterial stiffness. So for people who are wrestling with advanced arteriosclerosis and stuff like that, it’s a good proxy for whether they’re trending positive or negative. People tend to think of wearables in this kind of acute, “I’m having a heart attack.

“My wearable says I need to go to the hospital.” The thing is, that happens, and it’s a very compelling story, but the reality is that most people are actually managing illness, and that happens 24 hours a day as opposed to the 15 minutes that you’re having some kind of acute crisis.

So that’s another part of it: we’re oriented around that long-term journey, and I think that’s really helpful because people can say, “How am I feeling today relative to how I was feeling a month ago, to how I was feeling two months ago,” etc. That kind of gives them a good baseline.

Copy LinkHow Oura became a breakout tool for women’s health

EL KALIOUBY: I want to shift to women’s health, something I’m super passionate about. Your users, I guess, are predominantly women. But this wasn’t always the case, so what shifted? Why do women love their Ōura Rings?

HALE: It’s funny, it was about six months after I joined the company, and I was sort of trolling around TikTok and Instagram, and I was searching on Ōura, and I was trying to understand the customer. I kept stumbling into these young women who were like, “Girl, you need to get yourself an Ōura Ring, and here’s why.

“Because the patriarchy of medicine wants you to get on hormonal birth control, and you need to stand up for yourself and get off that hormonal birth control, and you should use Natural Cycles, and you should use your Ōura Ring.”

Natural Cycles is a product that people use as a digital contraceptive. It’s an FDA-cleared digital contraceptive. So it’s as efficacious as other forms of contraception. What it uses is your temperature to make a prediction about where you are in your cycle, and it kind of tells you today’s a good day or a bad day.

We had done a partnership with Natural Cycles because our temperature data — which, if you’ve ever had to do this, maybe for fertility reasons or for some other reason — lets you measure your temperature in the morning, and you’re supposed not to move. You’re supposed to wake up and then take your temperature, which, by the way, is hard to do.

If you get it wrong, particularly if you’re trying not to get pregnant using the rhythm method, if you move, you might miss the signal that says, “You probably should not have intercourse today, otherwise you’re gonna get pregnant.” So the fact that the Ōura Ring could measure its temperature overnight very consistently, and provide that very clean, accurate, and crisp signal to Natural Cycles, made it a perfect combination.

Plus, it was a cute ring. So between the two, it was amazing. I remember just starting to look at the numbers, and they were creeping up and to the right, and I’m like, “This is fascinating.” These women were on TikTok talking about it. It was this sort of turning-point moment.

We’d always been focused on women’s health, and we’d always had a really great women’s health team at the company. We had women leading that team, and they were super passionate about what they were doing, and they saw the opportunity. But this combination of these two products was so profound, it just rocketed.

So we have since then really doubled down on women’s health as a use case.

We have probably the biggest data set of women’s health data that’s the most continuous and most accurate in the world today. So there’s lots of latent insights and science in that data that we think will help us be really powerfully predictive, like predicting, I don’t know, when your labor might start or maybe some, you know, other major changes in your cycle.

Older, so none of this technology existed.

EL KALIOUBY: Oura has a proprietary women’s health model. I am curious why you had to build a women-specific model, and what does that entail?

HALE: It applies to you or me, certainly. But I think our vision is pretty simple: We should be a companion device for any woman, starting from the onset of menarche through menopause and beyond.

There are a couple of things. First of all, we have an LLM-based advisor inside the product, and it personalizes to you.

What we were finding was that even with the frontier models, for women’s health, the answers would be either too generic or weirdly unempathic.

They would have a weird tone. If you want to unpack that a bit, all the science and research that’s been done wasn’t until 1992 that women were allowed to be part of clinical trials. So there’s a lot of science out there that basically has no bearing on a woman’s physiology.

In fact, all the LLMs are trained on everything, and in a weird way, they have that as part of their DNA. So that was an observation we had: If we’re going to tune something to women, we think tone is really important.

We know this because our existing advisor product was used primarily by women. They used it twice as long, twice as frequently, with twice as many turns. They wanted to name it. They had almost a relationship with it.

When we asked them why they used it, they said, “Because it’s infinitely empathic and infinitely patient, knows everything about me, and will answer any of my questions,” which I think is a terrible recommendation for my gender. So we steered into the tone a little bit to tune the LLM that way.

Obviously, we have in-house clinicians, including one who’s dedicated to women’s health. We use the standards of practice from the National Institutes of OBGYN and obstetricians. So we built up the corpus of knowledge to be as accurate, scientific, and clinically validated as possible, so that you weren’t going to get advice from an LLM that was trained on Reddit.

And Reddit is amazing. I love Reddit, but sometimes it’s right and sometimes it’s crazy. The last thing that was really important for us, and I’m sure people are sensitive to this, is privacy. Health information is really important.

We wanted to be very clear, particularly around women’s health, that this information was going to be about your relationship with Oura, and you could trust us. We would make sure that information wasn’t going to make its way somewhere else.

I think that notion of trust is central to healthcare, but maybe even more important in women’s health.

So the experience is that if you’re in the advisor and you’re asking questions and it says, “This is a women’s health question,” we’re going to give you the women’s health answer, and we want you to know that it comes from our model and our infrastructure.

The thing is, it’s early days. I think it just went to full production yesterday. But the feedback we’re getting is, “Great, love the tone, love the information, and we love the idea that it’s private.”

EL KALIOUBY: Coming up, more on the technology behind Oura’s sensors and the data they’re collecting. That’s after a short break.

[AD BREAK]

Copy LinkThe future of sensors, data, and AI

One of my investment theses is this idea of the trifecta of sensors, data, and AI, and Oura is squarely in that space. So I want to unpack that. What kind of sensors does Oura Ring use? What does the data look like, especially the multimodal aspect of it? And we touched a little bit on the AI piece of it, but let’s talk sensors, data, AI.

HALE: There are multiple LEDs and photodetectors, and there are LEDs in different wavelengths because the different wavelengths have different properties. Some of them are more resistant to motion artifacts. Some of them see reflections. Basically, a lot of this stuff is about the reflection and absorption of light, and that’s the signaling you’re paying attention to.

So, multiple LEDs and photodetectors, temperature sensing, an accelerometer, which gives you the sense of motion, and then a set of algorithms on the device that are looking across these data sets. Then there are outside data sources. You can get blood tests, and we’ll ingest those blood tests.

A blood sugar monitor, and we ingest that. We use a lot of the sensors on the phone.

EL KALIOUBY: I want to talk about whether there are any negative effects of these sensors. For example, do people get upset? I have a friend, for example, who is type A, data-driven, and really wants to optimize her sleep scores.

As a result, she makes decisions around her social life where she’s like, “You know what? I’m not going to go out tonight because I want to optimize my sleep.” Have you seen examples of that, and how to have a healthy relationship with your wearable?

HALE: For some people, it can be something they need to pay attention to. When someone calls us and says, “I’m feeling this,” we’re like, “Hey, you know what? You probably should just take it off.”

That happens pretty infrequently. We’re here for your health and wellness.

We’re not here to make you obsessed or anything like that.

EL KALIOUBY: What about the expectancy effect? Basically, there was a study where people were given feedback saying that they slept poorly when, in fact, they slept really well. But that affected their perception of how they felt throughout the day, and they acted like they were tired and did worse on cognitive tests.

Do you — because I check my wearable first thing in the morning, and I have to think about it, right?

HALE: I think it actually has to do with whether the wearable gets it right. Because if the wearable does get it right, that’s actually useful information, not bad information. The study you cite is where they’re giving you bad information and seeing what the impact of bad information is on you.

So there’s an element of trust and learned experience. Again, it’s fascinating. I probably shouldn’t say this, but I’m going to say it anyway. There is a delta between people who’ve had a symptom radar alert and people who haven’t in terms of their retention.

When they have a symptom radar alert, they retain way better than someone who hasn’t. The reason is because there’s utility, but there’s also this little delta of trust. They’re like, “Whoa, three days before I got sick, it knew I was sick.”

And they say, “How does it do it? It’s magical technology. It’s incredible.” Those moments of trust are so important because if you’re staking your claim on being a tool for your health, it can’t feel random. It can’t say you slept well when you didn’t. I’m not saying we always get it perfectly right.

There are times when we don’t. But we get it right more often than other people do, and maybe more than other wearables do. That’s a little bit why, when you meet someone who has our ring, they kind of look at you and go, “I see you. What was your readiness score? You and I, we understand.

We get it. We’re part of this. We’re part of this cult.” And I think there’s something really powerful about getting it right. Now, the way that works out in practice is that we won’t ship something that doesn’t have super-high sensitivity and specificity. We won’t ship something that doesn’t match the gold standard of measurement.

We won’t ship something that is inaccurate for, maybe it’s accurate for most people, but inaccurate for some people. We make sure that when we do our sleep studies, we have 1,200 nights across every genotype, phenotype, ethnicity, age, health condition, all the things that would allow us to say, “This is accurate.”

And I think that’s so powerful because it goes to trust. And to your point, if you don’t trust the thing, it’s a toy. It’s not helpful to you.

And I think, again, as we think about the larger ambition of Ōura to be woven into the fabric of health care in some new way — preventative, proactive, personalized, all the things that would make health care different and better — you have to be really accurate and really trusted in order to have that opportunity. And I think that’s maybe a little bit why we’re so excited about Ōura.

Copy LinkHow a large physiological model could power personalized care

EL KALIOUBY: Last question. I told my 17-year-old son I was going to interview you, and I asked him what questions I should ask you. And he said, “What would it take for Ōura to be your tracker for everything in life, not just your physical health?” What are your thoughts on that?

HALE: I think there’s something really important about Ōura, which is that it doesn’t make you do work. In many ways, being a passive tracker is its superpower. So we do a lot of things that are really interesting to help support this idea that everything you do somehow makes its way into Ōura, and we make sense of it.

And I think AI will know all these contextual elements that are not part of the sensor. Today, it’s like we know your location, and we know your biometric signatures, and we know your activity patterns, and we know what you do, and we know your preferences, and all these things.

And so that helps our AI be very intelligent about the things that it says and get it right. And then our biometric algorithms are very science-driven, and they’re very supported by data, so they get it right. To really do that, I think you need a couple of different things to happen.

I think one is you probably actually need a little bit of work. So, for example, we have meal tracking, people taking photos, and it surprises me how many people actually take photos of their meals because we give them this little insight. We give them an understanding of, like, hey, is this high protein?

Is this high carb? Is this high sugar? Is it highly processed? We don’t say, like, “You have 40 calories,” or something, because the truth is that that’s all BS. If anybody’s telling you that they can tell you the calorie load of something from a photo, it is a lie. That is not true. It is very difficult to do because quantities, portions, ingredients, all these things are not apparent.

So you have to convince people to do a little bit of work, and I think that’s part of it. I think the other part of it is that your AI, whichever form it takes, needs to have access to a lot of context. And so for us, that context is your medical records, if you choose to connect them.

It will be the other sensors that you use. It’ll be your cellphone and giving it permission to access your calendar. All these things sort of make the context even richer. In that case, then it can start to be the place where you trust all of your data to be aggregated and be the passive tracker for all these things.

And we have a feature called Automatic Activity Detection, which is really interesting because what it does is look at the signature of your accelerometer, and it makes a guess about what kind of activity you were doing. It sees that your heart rate was elevated. It sees that you were moving in a certain way, and it says, “Hey, looks like you were dancing.”

“Oh yeah, I was dancing,” and then that reinforces the model to say, “Oh yeah, that’s what dancing looks like for you.” I mean, it’s different for everybody, let’s be honest. But that is kind of a mind-blowing moment because we get it right the first time often enough, and people are like, “Whoa, how did you know?”

Now, that’s just a really fascinating thing. And of course, if we knew your calendar and the calendar entry said Sally’s wedding, Saturday night, it’d be really easy. We’d get it right 100 percent of the time because if we saw that signature and you were at Sally’s wedding based on your calendar, it’s a no-brainer that you were dancing.

So I think that kind of multiple sources of data — and the trust, because you have to be willing to share that kind of information, including medical information, blood tests, all the things — then starts to get you to the place where it actually is the tracker for everything.

And I believe that you’re going to have something that’ll be the baseline that tracks your core vitals. That’s probably a ring. You might have a specialty sensor. Maybe, as you said, it’s a hormone monitor. Maybe it’s something else. Maybe there’s something on your face or maybe something in your ears.

And that’s getting other information. That might be more episodic because we don’t wear those things all the time, although I wear my glasses all the time, so maybe I’d wear them all the time. And you’re getting information from there, and then you’re getting information that’s more episodic.

It’s your blood tests. It’s all the data types, your diagnoses, and then it’s your medical records. And all of those things start to be a mosaic of you, and we call this the large physiological model. If we have a large physiological model that makes predictions about you, and those predictions aren’t just like, “Hey, you’re getting sick, and in three days you’re going to be on the couch.”

It’s more like, “Hey, in 10 years, if you don’t change your course, you’re going to have a heart attack.” That’s so powerful. It’s powerful from so many different dimensions: for the individual, for insurance companies, your doctors, for anything that’s part of the way people get treated or taken care of. That physiological model is going to be incredibly powerful.

And I think that’s where this all goes. It’s a powerful vision because if we achieve that, I think health care could be so much better, so much more efficient, so much more effective. I actually think we’re going to see more of that health care trickling to the edge, to the digital edge, to the experience that you’re having on your device with your wearables, and it’ll be automated.

There’ll be an AI doctor in your pocket or in the cloud who is looking over you and using your large physiological model to understand where you’re headed. And that is a transformation in medicine. I think it’s going to be incredible.

EL KALIOUBY: Yeah, I love that. This vision of kind of owning your own health journey, right? And having agency over it. Yeah, very exciting. Well, Tom, thank you so much for joining us on the show. This was fascinating.

Well, I appreciate you allowing me on, and thanks for all the great questions.

There’s such a big shift happening right now at the intersection of AI and health. Since we recorded this interview, Google launched its Fitbit Air. It’s very AI-first — a very different framework than what Ōura and other wearables are doing. There’s this new sleep earbud. Are you getting all these updates too? It’s hard to keep up!