Skip to main content
. 2024 Oct 10;24(20):6532. doi: 10.3390/s24206532

Table 3.

Precision: mean and standard deviation of epochs assigned to each sleep stage by each wearable and PSG. Precision or positive predictive value was calculated as = true positive/(true positive + false positive).

Oura Assignment PSG Assignment
Wake Light Deep REM
Wake 53.5%
(24.4%)
35.4%
(21.1%)
0.4%
(1.0%)
10.7%
(11.8%)
Light 3.7%
(3.9%)
79.5%
(8.6%)
9.1%
(8.6%)
7.6%
(5.7%)
Deep 0.8%
(0.7%)
21.5%
(17.0%)
77.0%
(17.4%)
0.5%
(0.9%)
REM 5.9%
(8.6%)
14.5%
(11.2%)
0.5%
(1.4%)
79.1%
(13.1%)
Fitbit Assignment Wake Light Deep REM
Wake 51.0%
(21.7%)
33.9%
(18.0%)
3.1%
(3.3%)
12.0%
(10.1%)
Light 5.6%
(7.9%)
72.8%
(13.4%)
12.8%
(9.8%)
8.8%
(6.5%)
Deep 1.3%
(5.9%)
23.3%
(18.6%)
73.2%
(21.5%)
1.8%
(5.1%)
REM 4.5%
(9.5%)
20.3%
(18.3%)
2.1%
(6.5%)
73.1%
(21.9%)
Apple Assignment Wake Light Deep REM
Wake 61.9%
(23.0%)
33.4%
(22.8%)
0.8%
(2.4%)
3.9%
(4.8%)
Light 5.4%
(5.5%)
72.7%
(10.3%)
16.5%
(11.1%)
5.5%
(5.2%)
Deep 0.6%
(0.9%)
10.4%
(17.3%)
87.8%
(19.7%)
0.6%
(2.8%)
REM 3.6%
(5.3%)
17.8%
(10.5%)
0.9%
(2.2%)
77.7%
(12.9%)

Note. Precision for each cell was calculated at the participant level. Mean and standard deviation of participant-level estimates are presented here. Estimates represent 35 participants for Oura, 33 participants for Fitbit, and 29 participants for Apple.