Skip to main content
. 2023 Apr 6;46(6):zsad101. doi: 10.1093/sleep/zsad101

Figure 9.

Figure 9.

Feature importance identified by SleepEns for each sleep–wake states. Importance ratios were taken for each sleep and arousal state (i.e. AW: Active Wake, QW: Quiet Wake, NR: NREM sleep, and R: REM sleep) from 30 repetitions in the test dataset. Importance ratios are log transformed so that 0 indicates no change in predictive power compared to non-shuffled features, ratio greater than 0 represents feature that improve sleep identification, and negative ratio represents feature that degrade performance of SleepEns. Delta: EEG δ power, Theta: EEG θ power, Sigma: EEG σ power, Beta: EEG β power, Theta/Delta: EEG θ/δ power, Sigma/Delta: EEG σ/δ power, Beta/Delta: EEG β/δ power, Sigma/Theta: EEG σ/θ power, Beta/Theta: EEG β/θ power, Beta/Sigma: EEG β/σ power, EEG RMS: EEG root-mean square.