University at Buffalo computer scientist Wenyao Xu got the idea for what he thinks will be the next craze in wearable technology—a choker-style necklace that monitors how much food you eat, by listening to you chew—when he was teaching a class full of students.
“One of them was eating,” he told me in an interview. “I couldn’t find out who, but I said, Don’t eat chips! How did I know he was eating chips? The chewing and swallowing sound.” The idea for AutoDietary, a high-tech calorie-tracking necklace, was born.
Xu, who collaborates with researchers at Northeastern University in China, is building all sorts of wearable devices (another invention is an insole that measures not just how many steps you take, but whether you’re climbing stairs, for example, or hiking). He’s especially enthusiastic about the food tracker, because there’s nothing like it out there right now.
“Activity trackers like Fitbit have been on the market the past few years. But all those devices can only track energy expenditure,” he said. And calorie-counting apps, where users plug in what they’ve eaten, are unreliable. “There’s no device to monitor energy intake” in an accurate and non-invasive way, he said. That’s what AutoDietary aims to do.
The device, which wraps around the neck and has a small, high-fidelity microphone, records sounds produced during chewing and swallowing. (It turns off automatically when you’re done eating, Xu says.) It then transmits data via Bluetooth to a smartphone, where the sounds are run against a library of known foods. In a recent study published in IEEE Sensors Journal, 12 subjects—women and men aged 13 to 49—were given water and six different things to eat: apples, chips, cookies, carrots, peanuts and walnuts. AutoDietary could identify what these people were munching on, 85 per cent of the time.
The AutoDietary device. Photo: University at Buffalo
But it still can’t tell the difference between regular and frosted cornflakes, or what sorts of ingredients are lurking in a bowl of chili. Xu, who is expanding the library of foods it can recognize and improving its algorithms, envisions pairing his food-tracking necklace with a smartwatch that detects insulin levels in the blood. Together, they’d have a better idea whether you’ve downed a Coke or some plain tea. He envisions AutoDietary helping the obese, and people with diabetes. “This is going to save their lives.”
The human body is an unreliable and imperfect machine, yet through wearables, nearly everything we do—the steps we take, the hours we sleep, now even the bites we chew—can be turned into cold, hard, and quantifiable data. Some of our body’s physical processes are probably better left unexamined (at least, at a certain level of detail). It’s true that people with medical conditions could find some use for AutoDietary. The rest of us, though, probably don’t want to think too hard about what we’re chewing all day long.
AutoDietary, might be a great way to lose weight, if not for the reasons envisioned by Xu. In another new study, researchers found that when people could hear themselves chewing—it’s a real turn-off to most of us—they ate less, too.