🚨If you (1) sleep; and (2) wear a device that measures sleep/heart rate/ heart rate variability. Here is a thread explaining (briefly) our recent validation of 6 commonly used wearables🧵doi.org/10.3390/s22166…
Caveat: more detail than can be condensed into a thread is required to fully understand such validations. Please use this thread as a primer for digesting the full paper!
In collaboration with @theAIS, we recruited 53 participants to spend 1 night in the @CQUni_Appleton Sleep Lab. Participants spent 9h in bed wearing 6 devices, as well as gold standard polysomnography (PSG; sleep) and electrocardiogram (ECG; heart rate).
The wearable devices were: Apple Watch S6, Garmin F'runner 245 Music, Polar Vantage V, Oura Ring G2, WHOOP 3.0, and Somfit.Records from each device (see paper for specific data extraction) were lined up with PSG-derived and ECG-derived data.
All devices detected >90% of sleep, but Polar, Oura Gen 2, WHOOP 3.0 and Somfit outperformed Apple Watch and Garmin for detecting wake. The devices ranged from 50 to 65% agreement for multi-state sleep when compared to PSG. But what does this mean?
All devices are are valid for field-based assessment of the timing and duration of sleep. While all can improve their assessment of sleep stage, better performing devices may provide valuable information when monitoring for sustained, meaningful changes in sleep stage.
HR : The way in which HR was sampled differed across the devices (again please read paper for context). The devices ranged from moderate relative agreement to almost perfect relative agreement with ECG (Bland Altman plots below).
HRV: Again, the way in which HRV was sampled differed across the devices (please read paper for context). The devices ranged from low relative agreement to almost perfect relative agreement with ECG (Bland Altman plots below).