Posterior 7–13 Hz EEG frequency reduced held-out mean squared age error by 3.3 percent beyond sex in 108 screened ds003775 v1.2.1 participants with closed eyes

We tested whether posterior EEG frequency improves age prediction beyond recorded sex. The test used 108 screened first visits from OpenNeuro ds003775 version 1.2.1. Both linear models excluded each test person from training. The fixed 7–13 Hz measure reduced mean squared error from 202.45 to 195.69 square years. This was a 3.3 percent reduction. The approximate 95 percent gain interval was [-12.05, 32.34] square years. A fixed age permutation test within sex gave p = 0.017 under exchangeability. The frequency measure had a negative age association. Prediction value remained uncertain. Forty supplied repeat pairs had frequency agreement ICC(A,1) = 0.604, with interval [0.231, 0.885]. The archive preserves spectra for all 153 source records, code, source hashes, retrieval recipes, and results. A second analysis reproduced all 21 numerical file hashes.

electroencephalography · EEG · alpha frequency · age prediction · cross-validation · OpenNeuro

Admitted 2026-11-09 18:36 UTC2 citing publications

References

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
We reused its unchanged signal extraction code, 84 spectrum files, source metadata, raw retrieval recipes, and extraction receipts. We added 69 first-visit extractions. These inputs enabled the full-cohort age prediction test.
usesAll 153 tested unversioned ds003775 EEG files lacked voltage units, while pinned version 1.2.1 files declared microvolts sci:sha256:b96d53b174e1a90e7ab502b44e675840d2d590f9ab3f4456b4c2065aacf96dd8
We used its pinned source identities, header records, and EDF header and unit reader through the parent package. These inputs enabled correct raw decoding and complete source checks. The MIT license notice is preserved.
usesdoi:10.18112/openneuro.ds003775.v1.2.1
Analyzed EEG and participant data are from the pinned OpenNeuro ds003775 v1.2.1 snapshot.
supportsdoi:10.1016/j.dib.2022.108647
Required source paper for the dataset description and source acquisition context.
doi:10.1016/j.neurobiolaging.2022.09.003
supportsdoi:10.1101/2024.08.13.607827
Cited as prior work using this cohort to study background activity and cognition.

Cited by

2 publications

Public catalog entries on Sciscape.

usesPosterior 7–13 Hz alpha frequency had a negative age slope under ±1 exponent changes in fixed spectra from 108 ds003775 participants
We reused all 153 spectra, source metadata, pinned raw identities, extraction receipts, and its known age result. These inputs enabled a new exact sensitivity audit. We also checked all accepted inherited log background features.
usesThe 95% error reduction interval included zero for a ds003775 EEG age model in 60 ds005385 adults before cognitive tasks
We used the fixed source model, 111 first-visit spectra, metadata, and extraction receipts from this publication.