Hi, i think i dont have answer to that question. In my specific case averaged Fitbit data for a 20+ nights seems to catch the proportions of sleep stages, but i’m not sure if this will be true for everyone… For example averaging for Oura didnt result in consistent results… Averaging can remove some noise, but can’t add more signal…
Also you should be aware that sleep stages defined as patterns of different signals and main one is EEG. If device dont have EEG than it just trying to predict sleep stages. Often we cant get desired signal by catching other signals. Often post-processing cant recover data which wasn’t gathered .
I’ve checked some videos of Muse and found it contorversal for sleep tracking. There is no offline mode for night, it works only with phone as companion.
Hypnodyne Zmax is expensive, but used in research. Software is only for Windows. Autoscoring costs money. And some manual processing is required, it’s not for a casual general consumer.
OpenBCI is for hardcore QS
Dreem 2 is my choise. It’s a consumer device, with app and everything is automated for casual use. Hypnogram can be exported to csv for deep analysis (unfortunately raw signal data isnt avaiable). Accuracy is seems to be fine. Company have stopped selling it to for consumers (but keep working with medical centres and scientists), so the only option is to buy it on ebay. I’ve bought 4 devices at ebay without any issues. All 4 is eu version with deep sleep stims. I’ve bought them for about 300EUR each. To make them last more than 2-3 years they need to be fortified in weak places, especially on the back side electrodes (i’ve used medical tape).
There is also Philips SmartSleep, but it’s bigger and less comfortable than Dreem 2 and accuracy is less studied than Dreem 2.
There is pretty old and discontinued product - zeo sleep. It’s pretty hard to buy on ebay working set. And some magic is required to get everything work.
Also you should know that sleep with eeg is less comfortable and some time is needed to get used to it.