[bryan-vh]

The iPhone requests permission to HealthKit, which comes through the Apple Health app. You will need an Apple Watch to record the sleep stage data during a sleep session, which the ML model then uses to calculate a sleep quality score from 1-100. All of the sleep quality scores are estimates, which is why I made it so that users could review and update their score based on how they felt in the morning, which in turn would tune the ML model to their sleep preferences.

[motohagiography]

That feedback cycle from the healthkit and mapping it to sounds is brilliant. It reminds me a bit of Richard Feynman's lucid dreaming hacks, where he was able to drop into a lucid state using some remembered cues, and invertedly, this is using ML to discover external sound cues to affect sleep state. I wonder what would happen if it worked too well?