> I’m interested in how they combine AI with EEG for better sleep tracking.
I'm not super well versed in how this works technically. My understanding is that even with raw EEG, accelerometer, and heart rate data it's not a straightforward problem to figure out what stage of sleep you're in. From the consumer side, I've just noticed a few patches they push down to my headband with "improved sleep tracking" in the changelog.
> have you experienced any discomfort from wearing a headband full of tech all night?
Nope. I was already wearing ear plugs and an eye mask, so the headband is a barely noticeable addition, even when side-sleeping.
It doesn't seem like a huge problem for ML to solve. Have someone wear a consumer EEG headband and an expensive medical device which measures their actual sleep state. Penalize the consumer device when its findings are off of the medical device. If the consumer device is still inaccurate after many iterations, add another input like a wristband with multiple sensors. Use both the EEG band and wristband as inputs to the ML, and train the ML until it's close in accuracy to the expensive medical device. In my mind, there just has to be some combination of easy measurable inputs that, together, correlate very well with sleep state.
I'm not super well versed in how this works technically. My understanding is that even with raw EEG, accelerometer, and heart rate data it's not a straightforward problem to figure out what stage of sleep you're in. From the consumer side, I've just noticed a few patches they push down to my headband with "improved sleep tracking" in the changelog.
> have you experienced any discomfort from wearing a headband full of tech all night?
Nope. I was already wearing ear plugs and an eye mask, so the headband is a barely noticeable addition, even when side-sleeping.