How Apple Watch’s New Handwashing Feature Works

Apple

The announcement of the Apple Watch OS7 at 2020's WWDC included some intriguing tidbits, but one new feature that had tech observers perking up their ears is its handwashing detection. That's right: in this new pandemic reality, the latest Apple Watch operating system can remind you to wash your hands when you get home, detect whether you're washing your hands, and give you a 20-second timer to help you scrub for the appropriate length of time. It then celebrates when you finish with a "Well Done!" message written in bubbles. It's all very cute and hygienic, but behind those soap bubbles lies some very interesting wearable technology.

The handwashing system combines several different elements. The first is a motion sensor, which is calibrated to determine whether you're performing movements that feel like washing your hands. If the watch detects that you might be rinsing your digits, it uses an internal microphone to check: are there the distinctive sounds of scrubbing in a sink, or are you doing something else with similar motions?

The watch uses machine learning, a kind of artificial intelligence, to gather information and refine its responses on the go. It's been loaded up with an internal database of sounds and movements for activities that resemble hand washing. For instance, kneading bread might have roughly the same movements as cleaning your hands, but it sounds completely different. Brushing your teeth might sound like you're washing your hands, but the motions aren't the same. This database of sounds and movements helps the watch determine whether you're soaping up to your wrists, or making elaborate croissants.

Apple

If the watch does decide you're washing your hands, it starts a timer and monitors how long the motion and sound continue. Anybody who skimps on their routine and stops within 20 seconds will get a gentle reminder to keep going. The watch also sends information about how often you've washed your hands per day, and for how long, to your iPhone's Health App, the same way you'd track your steps.

Machine learning is only as smart as its database, and there are a few sounds that might sound and feel like washing your hands that could result in "false detection." Washing the dishes, for instance, involves lots of hand movements, scrubbing noises, and water. Apple is using a beta community of watch testers to give feedback on how to refine the algorithm and make the handwashing detection as accurate as possible. The sounds in the hand-detection process aren't recorded or saved, in case you were worried somebody might hear your bad rendition of "Mr. Brightside" while soaping away.

Hopefully by the time the operating system is available to download, which is predicted to be around September, it won't be giving chirps and nudges to people who are just trying to peel potatoes.