I saw a talk on consumer wearables in sleep at the Sleep Trends conference this spring. There was a great presentation comparing consumer sleep trackers, and how their advertised reporting of "accuracy" can be misleading.

If you do a thought experiment: let's say I make a "sleep tracker" that is just a piece of plastic, no sensor, and write an algorithm that always presumes the wearer is in stage N2 sleep. Since we spend around 50% of our night in N2 sleep, this algorithm would start with a fairly high baseline accuracy of 50%.

Sensitivity and specificity are much better measurements of accuracy, because they measure the false positive and false negative rates. A comparison of Oura against a medical polysomnogram showed that while it's very good at detecting sleep, it's very bad at detecting not sleep:

From https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6095823/: "ŌURA ring had a 96% sensitivity to detect sleep, and agreement of 65%, 51%, and 61%, in detecting “light sleep” (N1+N2), “deep sleep” (N3),and REM sleep, respectively. Specificity in detecting wake was 48%"

The current state of the art in sleep measurement really requires some sort of EEG measurement beyond just PPG and HR, what Oura uses. The Dreem headband is a good example (https://dreem.com/), but it's definitely more cumbersome than a ring. It's tailored more like a medical service for insomnia than for the average consumer interested in quantified self.