Table 4.
Measures of event-wise detection performance with the different sleep study setups in comparison to standard PSG with audio and video
| Sleep study setup | Sensitivity (TPR) (%) | Specificity (TNR) (%) | Precision (PPV) (%) | False positive rate (%) |
|---|---|---|---|---|
| 1 channel EMG (all events) | 93.9 | – | 11.0 | – |
| 1 channel EMG (sleep-time events only) | 93.7 | – | 17.5 | – |
| 2 channels EMG (all events) | 75.2 | – | 11.3 | – |
| 2 channels EMG (sleep-time events only) | 73.5 | – | 20.2 | – |
| 3 out of 4 channels EMG (all events) | 88.2 | – | 14.3 | – |
| 3 out of 4 channels EMG (sleep-time events only) | 87.9 | – | 23.6 | – |
| PSG, no audio or video (PSG-N, all events) | 52.0 | 87.2 | 43.0 | 12.8 |
| PSG, no audio or video (PSG-N, sleep-time events only) | 55.3 | 79.5 | 51.4 | 20.5 |
| FES with audio (FES-A, all events) | 78.7 | 91.4 | 61.5 | 8.6 |
| FES with audio (FES-A, sleep-time events only) | 78.7 | 88.8 | 70.3 | 11.3 |
| PSG with audio (PSG-A, all events) | 76.2 | 95.5 | 75.6 | 4.6 |
| PSG with audio (PSG-A, sleep-time events only) | 75.7 | 92.8 | 80.4 | 7.3 |
Specificity and false positive rate were not calculated for EMG setups as true negatives could not be determined
TPR true positive rate, TNR true negative rate, PPV positive predictive value, EMG electromyography, FES forehead electrode set, PSG polysomnography