This Code Breaker Is Using AI to Decode the Heart’s Secret Rhythms

Roeland Decorte grew up in a nursing home in Belgium, where he learned to spot the subtle early signs of mental decline in small changes to how residents walked or talked. When Decorte was 11, his father, who owned and managed the care home, started waking up in the middle of the night with chest pains and an overwhelming sense of impending doom.

He went to two doctors, who briefly listened to his heartbeat through their stethoscopes and diagnosed him with anxiety. But the symptoms persisted, and it was only when he underwent a full set of scans at a private hospital that a third doctor uncovered the source of the problem—a tiny hole between the left and right chambers of his heart. If left unnoticed, it would have killed him—he was 39.

Disaster averted, the young Decorte was able to focus on his studies, and by age 17 he was an undergraduate at the University of Cambridge—the youngest Belgian ever to attend the prestigious college. (This caused some logistical issues: His tutor had to become his legal guardian, and a new payment system had to be put in place at the college bar to prevent him from buying alcohol like his peers.)

He spent the next seven years specializing in ancient codebreaking, and a comfy career in academia (or a more exciting one as an Indiana Jones–style relic hunter) beckoned. But Decorte never stopped thinking about what had happened to his dad and how he could have been diagnosed much sooner if a doctor, any doctor, had spent more than 30 seconds listening to his heart. So in 2019, lacking medical training but armed with the confidence that only an Oxbridge education can provide, the then 27-year-old Decorte founded a company and turned his attention to cracking a different ancient code: the secret rhythm of the heart.

There’s an AI boom in health care, and the only thing slowing it down is a lack of data. Meanwhile, time-pressured doctors can collect information only sporadically. Wearables such as smartwatches might be able to measure pulse, but they’re bad at more specific diagnoses (partly because the wrist is about as far away from the really vital organs as you can get).

Decorte wanted to develop a piece of technology that could monitor the body continuously and precisely, so that people like his father could get the treatment they need more quickly. He began by trying to build sensors into clothes so people could track their vitals without a doctor’s visit. Then he designed an elaborate exoskeleton packed with sensors to measure all kinds of ailments. This attracted some military interest but wouldn’t really have helped someone like Decorte’s father. “I was very naive,” he said when we met recently in the wood-paneled basement of a twee café in Mayfair, London. “There was about two years full-time where I was just working out of the spare room in my house doing nothing else.” But the problem he kept running into was noise: Unless you could build a contraption that pressed each sensor right against the skin, there was too much random interference from people moving around in the world to get a good sense of what was actually happening in the body.

Most PopularGearThe Top New Features Coming to Apple’s iOS 18 and iPadOS 18By Julian ChokkattuCultureConfessions of a Hinge Power UserBy Jason ParhamSecurityWhat You Need to Know About Grok AI and Your PrivacyBy Kate O'FlahertyGearHow Do You Solve a Problem Like Polestar?By Carlton Reid

But perhaps, Decorte thought, noise could also be the solution. During the pandemic, he met PhD student Erika Bondareva, who had published work on diagnosing Covid by analyzing audio data collected by people coughing into an app. Her software checked for patterns common to people with the disease, then looked for those same patterns to try to detect it earlier in others. Together, Bondareva and Decorte worked on expanding that idea to other ailments—starting with heart conditions. Eventually, Decorte said, he found himself replacing every sensor on the exoskeleton he’d designed with an audio sensor. Finally, he realized that the only hardware he needed was a microphone.

Today, his company, Decorte Future Industries, is at the vanguard of an audio-powered revolution in health care. Sophisticated algorithms strip out background noise and focus on interpreting the body’s faint signals. There are smart stethoscopes and apps beginning to hit the market that claim to diagnose Alzheimer’s based on speech patterns, but Decorte wants to go further: He believes the technology he’s developing will be able to diagnose heart problems, stomach cancer, and even blood sugar levels, as well as conditions related to speech and gait. Instead of the mishmash of numerous apps and hardware solutions aimed at different conditions, he sees a single solution: The microphone in your smartphone would always be listening, and once every few weeks you’d get an alert to press it against various parts of your body for more detailed readings.

Decorte has raised millions in funding and is growing a small team in Cambridge. He’s running clinical trials in India—one local doctor thought he was being scammed until Decorte’s colleague played back the recording and the doctor could hear his own voice on the tape from minutes earlier. Decorte’s technology matches up to ECG readings with 99.6 percent accuracy—but with just a microphone, patients can take the readings at home.

It’s been a steep learning curve, but one that draws on his experience of ancient codebreaking as much as the new skills of networking and artificial intelligence. “It’s all pattern recognition,” he says.

This article appears in the September/October 2024 issue of WIRED UK magazine.

About Amit Katwala

Check Also

An Ultrathin Graphene Brain Implant Was Just Tested in a Person

In 2004, Andre Geim and Konstantin Novoselov at the University of Manchester in England achieved …

Leave a Reply