Bands set by early 7–13 Hz residual maxima retained a positive Alzheimer-control EEG contrast after age and gender adjustment in 65 processed ds004504 v1.0.9 recordings.

This audit tested a band choice in 36 Alzheimer and 29 control recordings from OpenNeuro ds004504 v1.0.9. Early 100-second posterior EEG samples set a 7–13 Hz residual maximum after a background fit. Late 100-second samples supplied scores. Both four-Hz and five-Hz bands shifted by that maximum minus 10 Hz. The common age and recorded gender model gave an Alzheimer versus control contrast of 0.2572 log10 units with fixed bands and 0.3000 with shifted bands. The main paired change was +0.0428 (95% person bootstrap interval -0.0457 to 0.1285). Its direction was uncertain; both score contrast intervals were above zero. The planned raw-maximum rule gave a change of 0.2606 (95% interval 0.0955 to 0.4129). The main analysis retained all 65 records. A separate comparison preserved the earlier peak-transfer study's exact 55-person sample. All 88 source hashes passed, and the extraction reproduced 167,200 prior band powers exactly. Independent formulas checked scores and 40,000 bootstrap group fits. An offline repeat reproduced all five result files byte for byte. The package preserves spectra for 88 persons, source provenance, code, plans, results, figures, and reproduction commands. The inference is limited to this processed cohort and these band rules. A frequency maximum need not be an alpha oscillator. No disease mechanism or external diagnosis performance is established.

EEG · Alzheimer disease · frequency bands · spectral power · band shift · person bootstrap