Last Updated on November 24, 2023 by pg@petergamma.org
Validation conference paper from 2023:
«The results showed that the Muse S achieved an accuracy of (DSN: 85.2%, MNet: 86.3%) and Cohen’s Kappa (DSN: 0.77, MNet: 0.79)»
compared to a professional Polysomnography device. We suppose there will be a paper published in a scientific journal soon about the Muse S accuracy.
Finally we got an alternatve to Dreem 2 which also has a validation paper:
Unfortunately Dreem 2 is not available anymore for everyone and for those who can get it it is 1`400 USD. Is this worth the price?
- Both Dreem 2 and Muse S have similar accuracy values of around 80 % compared to a professional polysomnography device. These values are still quiet low.
- A used Dreem 2 had recently a price of about 700 USD on eBay which is a lot of money for a sleep tracker with 4 EEG channel. And this was the version with the closed source app with no sensor RAW data.
- The Dreem headband for researchers is sold for about 1′ 400 USD. With this device we have the option to get all the raw data. But still is this not a lot of money for a device with an accuracy which is very limited?
- We can try to get more accuracy with more EEG channels:
- We can start doing some tests with a Muse headband and the Mind Monitor and the Sleep Python software.
- We can then build a sleep tracker from OpenBCI or PiEEG.
We suppose that with 8 or 16 EEG channels we will beat both Dreem 2 and Muse S headband as far as accuracy is concerned. And who want a closed source Dreem 2 (which is currendly avaliable used on eBay for 700 USD) or a Muse S headband (which is available new for 400 USD from www.techstudio.ch) for sleep tracking if we can have it all open source in Python with Pi EEG or OpenBCI? We suppose that these devices will have a a higher accuracy for a lower price. Authors of validation papers for sleep trackers choose these devices they are most probably more accurate and cost less.