In 19 adults, agreement of three stored EEG scores reduces total Wake/sleep error but leaves Wake errors, if spectral code 1 means Wake

We tested whether agreement of three forehead EEG channel scores selects Wake/sleep labels closer to a manual reference. We used all 19 records in OpenNeuro ds006695 version 1.0.2. We assumed that spectral code 1 means Wake and codes 2 to 5 mean sleep. The source does not give this numeric spectral dictionary. Each person had equal weight. Agreement retained 91.83% of valid epochs. Mean reference error fell from 12.05% to 9.04%. The paired reduction was 3.01 percentage points. Its 95% person bootstrap interval was 1.83 to 4.86 points. However, 45.82% of retained manual Wake epochs were labeled sleep. This rate was 94.45% in retained epochs from internal manual Wake bouts of 30 to 60 seconds. The overall reduction stayed positive in all planned checks. The channels share electrodes. The manual labels are a reference, and timing is uncertain. The intervals assume independent people and fixed scores. The package includes original score bytes, fixed source versions, code, every epoch, results, and repeat checks.

forehead EEG · sleep staging · Wake detection · channel agreement · reference error · epoch selection · bootstrap interval · OpenNeuro ds006695