The 95% interval included zero for posterior mean removal in 80-second CLN3 and volunteer EEG samples

We tested posterior mean removal in 32 CLN3 and 63 volunteer EEG records from OpenNeuro ds008896 version 1.0.1. Both rules used P3, P4, O1, O2, and the first 80 retained seconds per person. The baseline subtracted the mean of 19 channels. The local rule subtracted the mean of the four posterior channels. The CLN3 minus volunteer log10 theta-to-alpha contrast rose from 0.19436 to 0.20941. The paired change was 0.01505 log10 units. Its 95% person bootstrap interval was -0.01733 to 0.04959. The interval assumes independent people and fixed source processing. All five fixed sensitivity checks also had change intervals that included zero. The two-band retained-power account showed similar group differences that largely cancelled in the ratio change. This study did not establish an increase under the tested rule. It does not prove a zero effect. We verified all 95 source EEG hashes and all 5,558 stored epochs. A fresh extraction reproduced all 95 compact input files. A second analysis reproduced all seven numeric output files. The package preserves code, sufficient band matrices, source metadata, locked retrieval paths, all scores and bootstrap draws, and reproduction instructions. Missing age and clinical data limit causal and clinical interpretation. The earlier Alzheimer result used a different baseline, region, and adjustment. This is a test in another cohort, not an exact replication.

CLN3 disease · electroencephalography · EEG · theta-to-alpha ratio · posterior mean removal · power spectrum · bootstrap