The 95% rank-gain interval included zero for a mean eye-state correction in one-month EEG matching of 60 ds004148 adults with a fixed Pearson rule

We tested whether a mean eye-state change could improve posterior EEG person matching one month later. The test used 60 adults from ds004148 version 1.0.1 and a fixed Pearson rule across 1 through 40 Hz. Two fixed folds each had 30 candidates. Each correction used only the other fold in sessions 1 and 2. Mean rank changed from 0.71092 to 0.71523. The gain was 0.00431. Its 95% interval was -0.03144 to +0.03698 from 10000 paired person-bootstrap draws with the correction fitted again. Open-eye queries had lower rank after correction. Closed-eye queries had higher rank. These changes nearly canceled. All four sensitivity-check intervals for average gain included zero. The short-record check used an added quality rule and had 59 eligible people. Independent scalar checks passed. Two fresh complete source records reproduced all spectrum arrays exactly. A complete repeat matched all 11 analysis files byte for byte. The package preserves code, compact inputs, source hashes, figures, plans, and instructions. The result is conditional on this split and source protocol. It does not prove zero benefit or a causal eye-state effect.

electroencephalography · EEG fingerprinting · person matching · eye state · power spectrum · rank statistics · bootstrap · test-retest reliability