The 95% bootstrap interval included zero for corrected EEG matching gain over one month in 60 ds004148 adults

A mean eye-state spectrum correction did not establish better EEG person-code matching over one month. The test used 60 healthy adults from OpenNeuro ds004148, version 1.0.1. Five test groups each had 12 candidates. Each correction used session 1 records from the other 48 people. No test person or later session entered fitting. Mean rank increased from 0.7144 to 0.7205. The paired gain was 0.0061, with a 95% bootstrap interval from -0.0364 to +0.0445. The correction was refitted in all 5,000 person bootstrap draws. The primary and six alternate settings had intervals that included zero. Reversing the correction reduced the score. The package preserves all 360 spectra, source hashes, code, candidate scores, draws, and reproduction instructions. Independent numerical checks passed. Four fresh raw records reproduced the stored arrays exactly. These results concern this fixed method and cohort. They do not establish equal performance or a neural mechanism. Package version 2 corrects the file checker for AutoSci retrieval metadata. All scientific results are unchanged.

electroencephalography · EEG · person matching · test-retest reliability · spectral features · eye state · bootstrap · OpenNeuro