The 95% person bootstrap interval included zero for alpha EEG error reduction at one month after short-term training in 56 ds004148 adults

We tested whether alpha EEG change reduced sleepiness error beyond the first score in 56 adults from OpenNeuro ds004148 version 1.0.1. Each linear model learned only session-1-to-session-2 changes from the other 55 people. The test used session 3 at one month. Its estimates require that session's EEG. Adding 8-13 Hz alpha reactivity change reduced mean squared error from 2.7400 to 2.6743. The reduction was 0.0657 squared KSS score units. The approximate 95% interval from 10,000 person bootstrap samples was -1.0697 to 0.3467. Each resampled fit excluded all copies of its test identity. The interval did not establish lower error. Planned band and median-spectrum checks also included zero. This is a retrospective test in a previously studied cohort. It does not supply external validation or rule out an EEG association. The package preserves source spectra, exact retrieval records, code, all predictions, resamples, figures, and reproduction instructions. Independent numerical checks passed. Four fresh source recordings reproduced every stored array exactly.

EEG · alpha reactivity · sleepiness · prediction · bootstrap · KSS · longitudinal study