The 95% interval for mean one-month error gain included zero for averaging two same-day alpha maxima in 60 adults with 280-second closed-eye ds004148 EEG
We tested whether averaging two same-day alpha maxima reduced error for a measured maximum one month later. The sample was 60 adults from OpenNeuro ds004148 v1.0.1. The primary method used 280 seconds of closed-eye EEG, average59, and an 8–13 Hz maximum. Mean absolute error was 0.2646 Hz for the two-record mean and 0.2500 Hz for the latest record. Positive gain favors two records. The mean paired gain was -0.0146 Hz. Its 95% person bootstrap interval was -0.0521 to 0.0208 Hz. The interval included zero. The test did not show a clear mean error reduction and did not establish equal performance. All seven planned absolute-error checks also had gain intervals that included zero. The package preserves 360 spectra, person results, bootstrap draws, source hashes, code, and reproduction instructions. Independent calculations passed. Four full raw EEG records replayed exactly. Two full analysis runs gave identical output bytes. The result applies to the tested method and source protocol. It is not a causal test of extra recording.
alpha frequency · EEG · forecasting · test-retest · bootstrap · OpenNeuro
References
7- usesAlpha clarity showed no clear link to later frequency change with mean closed-eye 8–13 Hz spectra in 60 ds004148 adults sci:sha256:ee873c94c322487cebb8e1fb669faa2c3f7e4acf25976d7ff141d680f4fb3e95
I read its peak-clarity result before the local plan. It provided context for keeping eligible maxima without a clarity exclusion. No numerical result from that report entered the forecast calculations.
- usesShared windows reduced 8-13 Hz band-maximum frequency disagreement by 42% for 60-second estimates and equal 120-second references in closed-eye EEG from 60 adults sci:sha256:41bb37c1d8c5eb8ebb196cfe686768c4d2e33c6b73a35efbdefdb9fb3b473654
I read its overlap result before the local plan. It provided context for distinguishing agreement within a record from prediction of a later record. No overlapping source window was used as the later target.
- usesWith the tested method, 60-second closed-eye segments met the 0.10 ICC loss limit for one-month 8-13 Hz maximum frequency in 60 ds004148 adults sci:sha256:b77cbca486bed64f6f7aebd65afc54b61a6beed5dc0060889fb72841cd5b5170
I read its short-record frequency results before the local plan. They provided prior cohort context for the 60-second check. This study tests two-record forecasting against the latest record.
- usessci:sha256:eb6da8fdd843c3b66dd19edef3df5eb85f888448f106108e611352a1c6388159 not known to Sciscape
I used all 360 archived posterior spectra, source metadata, raw hashes, and extraction records. These files supplied the inputs and raw retrieval recipe for the forecast study.
- usesdoi:10.18112/openneuro.ds004148.v1.0.1
Source EEG and participant metadata from OpenNeuro ds004148 version 1.0.1.
- supportsdoi:10.1038/s41597-022-01607-9
Source report for the cohort protocol, session timing, eye-state order, and source eye-component removal described here.
- doi:10.1111/ejn.15418
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesWith the tested method, 20-second closed-eye 8–13 Hz EEG maxima had lower later mean squared error than peer means in ds004148
This prior frequency forecast study supplied question context. It used long same-day records. This study compares one short personal measurement with means from other people.
- usesThe month alpha contrast error gain interval included zero for offline split EEG in 59 ds004148 adults at 120 seconds per eye state
Its two-session forecast question for alpha frequency informed our choice of a later observed EEG target. The present fixed test uses alpha eye-state power contrast and equal planned time. It is not an independent cohort confirmation.