The 95% person bootstrap interval included zero for offline 8–13 Hz power forecast gain across a four-second gap in 29 ds004148 test adults with closed eyes

We tested whether a recent posterior 8–13 Hz EEG power value reduced forecast error beyond the mean of past values. We fitted one coefficient on 28 adults at session 1. We applied it to 29 different adults at session 3 in OpenNeuro ds004148 version 1.0.1. The target started four seconds after the latest input ended. Each person had equal weight. Mean error fell by 6.247%, but the 95% person bootstrap interval ranged from −10.049% to 17.470%. Only 10 test adults had less error. This test did not establish an added forecast benefit. We rebuilt spectra for 60 adults without an added time filter. The package preserves source identities, spectra, forecasts, code, results, and replay instructions. The source-ID split was not random and had unequal sex counts. The result applies to offline analysis of accepted windows in this cohort.

electroencephalography · EEG · alpha power · forecasting · signal processing · person bootstrap · test-retest · OpenNeuro · offline analysis