Wearables increasingly look to AI to predict health problems

(Bloomberg) --Haley Billey bought an Oura Ring to track her fertility. It arrived the day after she learned she was pregnant. She slipped the $450 titanium band on anyway.

Processing Content

Months of worrisome readings on measures of energy and stress, levels she initially attributed to pregnancy, persuaded her to seek a professional opinion. The ultimate cause: Hashimoto's disease, an autoimmune disorder.

"The ring can't diagnose you," said Billey, 31, a manager at National Park Friends Alliance in Ann Arbor, Michigan. At least "I can look at the data and take it to my doctor." But she'd like the ring to do more. 

So she's now handing over personal data to help Oura Health Oy, the ring's maker, detect signs of hypertension. It will ultimately feed into a novel artificial intelligence model the company is building to predict events like heart attacks and strokes — years before they happen. 

"The real breakthrough isn't knowing you had a problem," said Oura Chief Executive Officer Tom Hale. "It's knowing before you do, so you can change behavior and prevent it."

Getting there will mean pushing boundaries of data collection and personal-information sharing. Billey is all in. "I'm happy to have them use the data they get from me to build stronger algorithms," she said.

It's a trend that's been building for years and is now entering a new phase around proactive health. Companies worldwide, including Samsung Electronics Co. and Apple Inc., are studying how the technology can predict health events.

It's "the elusive unicorn," said Ramon Llamas, who directs mobile-device research at International Data Corporation. Finding it will require  governments to rewrite regulations regarding what, exactly, constitutes a medical device.

Rings, smartwatches and other such tools are viewed as credible biometric monitors, capturing information about respiratory rates, blood oxygen levels, sleep duration and more. They're increasingly common, with the size of the market estimated last year to be over $90 billion.

Tennis players will this year be allowed to wear the gadgets at grand slam matches. Golfer Rory McIlroy let Whoop Inc. — in which he is an investor — release statistics collected by his wrist band during the Masters Tournament. US Health Secretary Robert F. Kennedy Jr., told Congress he wants every American to wear a wellness tracker.

Gathering and analyzing data is one thing.  "The bar for prediction is much higher," said Joseph Schwab, director of surgical innovation and engineering at Cedars-Sinai Medical Center in Los Angeles.

Inventors keep trying to scale it, as seen at the Consumer Electronic Show in January. One attention-grabbing product was the "longevity mirror" from NuraLogix, which purports to forecast health risks by measuring blood flow from a selfie. Earlier this year, the Death Clock, which uses AI trained on longevity data, went viral for promising to "predict when you'll die."

The wearable industry is focusing on such risk factors as blood sugar, hypertension and pulse that have established ties to heart health. The plan is to connect the dots across the body, from reproduction to cognition, leaning on artificial intelligence to make it happen.

"The same way a large language model predicts the next word, we're building models that can predict the next heartbeat," said Will Ahmed, chief executive officer of Whoop. The closely held Boston-based company was valued at $10.1 billion in March, after raising $575 million in a financing round.

The goal is to warn of heart attacks in as little as 15 minutes before and in some cases years ahead of time — a reality Ahmed said is coming "a lot sooner than people are expecting."

Alphabet Inc.'s Google, which owns Fitbit, just launched a screenless band to rival Whoop. Fitbit also added a feature that integrates a user's medical records and readings from a continuous glucose monitor. It then asks AI to flag conditions and suggest how to address them.

Finnish ring-maker Oura has a women's health chatbot to answer questions about menstrual cycles. In May, it introduced a new feature to monitor birth control and some symptoms of aging, part of a push to foresee ovulation, hormonal shifts and menopause. Premium smartwatch maker Garmin Ltd. partnered with the birth-control app Natural Cycles to help pinpoint ovulation using skin temperature.

At the other end of the life cycle, Samsung Health is working working to detect dementia using indicators such as speech and gait. It plans to roll out an AI "personal health companion" in the next several months to offer advice and "nudges" to Galaxy Watch wearers about risks, like those picked up by its recently launched blood-pressure monitoring feature, said Hon Pak, head of digital health at the Samsung unit.

While no one has been able to show how variations in the vast data sets the companies are building can impact an individual's risk of chronic diseases, "that's what we're aiming to do," said Pak, a physician and former chief medical information officer for the US Army.

The downsides of health-monitoring devices as they now exist are well documented, with heavy use sometimes linked to obsessive tracking and doctor shopping. The Internet is full of accounts of so-called wearable anxiety, where people cancel plans because a sensor spotted possible symptoms of, say, flu or congestion, though they didn't end up falling ill, or report being gaslit by low sleep scores. Frequent alerts can lead to unnecessary testing, said Margaret Lozovotsy, who directs digital health innovations at the American Medical Association.  Too much self-surveillance could shift responsibility for monitoring from experts to individuals with little medical literacy.

"It internalizes this logic of 'I'm only as good as the data that I produce,'" said James Gilmore, author of Bringers of Order: Wearable Technologies and the Manufacturing of Everyday Life.  That mindset might encourage dangerous responses, he said.

The range of concerns about the predictive wearables of the future is broad, beyond the cost barrier for those who can't afford to buy in. Users today tend to skew younger, wealthier and more health-conscious, so inputs used to train the models may not reflect other high-risk populations. Most information gathered may sit outside the protections of the Health Insurance Portability and Accountability Act, or HIPPA, governed instead by terms of service that allow broad secondary uses. Data breaches are always a worry.

"A dystopia I always have is the advent of healthcare spam," said Kevin Fu, an expert in emerging sensor technology at Northeastern University College of Engineering. The former acting director of medical device cybersecurity at the US Food and Drug Administration worries that "I suddenly start getting more advertisements saying, 'Here's some medicine for hypertension' because they're somehow using my watch."

The industry has acknowledged the concerns. Companies have pointed out steps they're taking to strengthen security measures and how they see integrating their products into healthcare systems to, in their view, make them accessible to more people. So far the FDA hasn't loosened regulations barring wearables from diagnosing diseases or confirming medical conditions. The lobbying continues. An example: Oura Health, valued at about $11 billion after raising $875 million last September, is advocating for a new US classification that would allow wearables to alert users to potential health issues without undergoing the lengthy clearance process required for medical devices.

One wearable fan, Thomas Lynch of Florida, said his Oura ring "saved my life" after a major surgery. His ring flagged an elevated heart rate, and ultimately led to his diagnosis of a pulmonary embolism. The credit goes, he said, to what was observed in real time. As for spotting future events?  As a data scientist who works with AI every day, he said he'll be taking that with "a grain of salt."

For Haley Billey,  who is participating in the Oura study, the issues raised don't give her pause. "If a company knows my blood pressure and heart rate and everything, I'm comfortable with it at the moment — and that feels silly to say out loud." She added, "I hope I don't eat my words."

To contact the author of this story:
Miquéla V Thornton in New York at mthornton74@bloomberg.net


Bloomberg News
Wearable technology Health insurance Insurtech Artificial intelligence
MORE FROM DIGITAL INSURANCE
Load More