Six posterior EEG ratio orders agreed at mean19 and Cz but reversed between them in 88 processed ds004504 v1.0.9 records

We tested fixed processed EEG signals from all 88 people in OpenNeuro ds004504 version 1.0.9. The score was the posterior theta-to-alpha power ratio. The reference was a shared, constant linear mixture of mean19 and Cz. Six person pairs had the same strict order at both endpoints and the opposite order at an interior mixture. This was 6 of 3,641 endpoint-consistent pairs, or 0.165%. Group AUC also left the endpoint range. The Alzheimer-versus-control maximum was 0.884100, above both endpoints by 0.002874. The frontotemporal-versus-control minimum was 0.803598, below both endpoints by 0.008996. All four fixed checks are reported. Hidden-pair counts were 8 for the first ten accepted windows, 12 for the last ten, 0 for Fz, and 29 for Pz. None of the six primary pairs was hidden in those checks. The effects were small and depended on the score rule. The package preserves a polynomial event audit, matrices, source hashes, code, figures, all results, and reproduction instructions. Separate roots, direct powers, three source replays, and a complete saved-input repeat passed. These are finite-cohort measurement results. They do not establish a brain mechanism or accuracy in new patients.

electroencephalography · EEG · reference sensitivity · theta-to-alpha ratio · power spectrum · pairwise order · area under the curve · continuous-path audit