Table 5.
Comparison algorithm with polysomnography in healthy good sleepers (N = 22, Pennsylvania).
| Parameters | Metric | Value | t; DF | P |
|---|---|---|---|---|
| Sleep onset | Difference (min) | −20 (95% CI: −39; −2) | −2.30; 21 | 0.03 |
| MAE (min) | 32.9 | — | — | |
| Sleep wake | Difference (min) | −17 (95% CI: −39; 4) | −1.67; 21 | 0.11 |
| MAE (min) | 21.0 | — | — | |
| SPT-window | Difference in duration (min) | 2 (95% CI: −24; 27) | 0.14; 21 | 0.89 |
| MAE in duration (min) | 37.7 | — | — | |
| c-statistic | 0.83 (IQR: 0.80–0.90) | — | — | |
| c-statistic 24 hour† | 0.95 (IQR: 0.95–0.99) | — | — | |
| Accuracy (%) | 89 (IQR: 86–97) | — | — | |
| Accuracy 24 hour† (%) | 96 (IQR: 95–99) | — | — | |
| Sleep within SPT | Difference in duration (min) | −6 (95% CI: −27; 15) | −0.59; 21 | 0.56 |
| Sensitivity (%) | 93 (IQR: 94–100) | — | — | |
| Sleep efficiency within SPT | Difference (percent point) | −1.74 (95% CI: −4.46; 0.98) | −1.33; 21 | 0.20 |
| MAE (min) | 4.8 | — | — |
*P < 0.005; MAE: mean absolute error; min: minutes; SPT-window: Sleep period time window; CI: Confidence Interval; DF: degrees of freedom; t: t-statistic; IQR: Inter quartile range; †Recording expanded with simulated data of wakefulness to resemble 24 hours.