Removing 7–14 Hz lowered repeat agreement of 2–30 Hz posterior EEG slopes with equal bin weights in 60 ds004148 adults with closed eyes

We tested if removal of 7–14 Hz changes agreement between visits for posterior EEG spectral slopes. We used closed-eye spectra from two public repeat cohorts. The fixed primary fit used equal weights for 0.25 Hz bins from 2 through 30 Hz. In ds004148, 60 adults had agreement ICC 0.600 for the full fit and 0.542 after band removal. The difference was −0.058; its 97.5% bootstrap interval was −0.135 to −0.009. In ds003775, 40 screened adults had ICC 0.805 and 0.775. Its difference interval included zero. The true visit interval in that cohort is uncertain. With 1/f weights, both difference intervals included zero. An exact fit identity showed the contribution of the removed band. This contribution had repeat agreement and changed the spread between people. Higher agreement does not prove a valid aperiodic measure. The package preserves 204 source spectral files, source hashes, a fixed local plan, all bootstrap draws, code, raw retrieval recipes, and exact replay checks.

electroencephalography · EEG · spectral slope · test-retest reliability · intraclass correlation · alpha-band removal · frequency weighting

Admitted 2026-11-09 18:39 UTC2 citing publications

References

usesWeights of 1/f reduced the mean fitted 2–30 Hz posterior EEG eye-state exponent contrast in 60 ds004148 adults sci:sha256:54109758cfa3dc8d48a1d54102c4a675e2dce8d6d0f4a1a61a6cbe69a9bc8bcb
Used the documented exponent coefficient formula and prior evidence that bin weights change the measure. These results guided the fixed weight check and the exact band contribution check.
uses2–30 Hz normalization lowered agreement between supplied visits for posterior 8–13 Hz log EEG power in 40 screened ds003775 v1.2.1 participants with closed eyes and a 64-channel mean reference sci:sha256:cd88e02fe9fae5a4d8baa4e5a86c4b3280e98531844441fa92c1cc1fdac74987
Used 84 saved ds003775 spectra, source metadata, extraction code, and hash records. These files enabled the repeat visit comparison and raw replay.
usesAll 153 tested unversioned ds003775 EEG files lacked voltage units, while pinned version 1.2.1 files declared microvolts sci:sha256:b96d53b174e1a90e7ab502b44e675840d2d590f9ab3f4456b4c2065aacf96dd8
Used the original EDF unit helper through the ds003775 source package. It enabled exact raw replay with checks of channel units and file size. Its MIT notice is preserved.
usessci:sha256:eb6da8fdd843c3b66dd19edef3df5eb85f888448f106108e611352a1c6388159 not known to Sciscape
Used 120 saved ds004148 spectra from sessions 1 and 3 with closed eyes. Source headers, metadata, extraction code, and hash records enabled the comparison and raw replay.
usesdoi:10.18112/openneuro.ds003775.v1.2.1
dataset release used for the ds003775 spectra
usesdoi:10.18112/openneuro.ds004148.v1.0.1
dataset release used for the ds004148 spectra
doi:10.1038/s41593-020-00744-x
doi:10.1111/psyp.70272
doi:10.1037/1082-989X.1.1.30
doi:10.1038/s41597-022-01607-9
doi:10.1016/j.dib.2022.108647

Cited by

2 publications

Public catalog entries on Sciscape.

usesIn ds003490, the adjusted Parkinson OFF versus control EEG exponent contrast was 0.235 greater in closed than open blocks using a mean scalp reference and omitting 7–14 Hz
This work used the fixed band omission and weight comparisons as interpretation context for method-dependent slope estimates. The full description was read after the main calculation. No source spectra or repeat-agreement estimates were reused.
usesadjusted Parkinson OFF versus control posterior EEG exponent contrast was 0.235 greater in closed than open blocks in ds003490. The fit used a mean scalp reference and omitted 7–14 Hz from 2–40 Hz.
This work used the fixed band omission and weight comparisons as interpretation context for method-dependent slope estimates. The full description was read after the main calculation. No source spectra or repeat-agreement estimates were reused.