Aperiodic-knee timescales did not increase anteriorly in 36 epilepsy patients; post hoc, inferior pairs had longer timescales

We tested whether intrinsic timescales from the aperiodic knee of stereo-EEG (SEEG) spectra increase from posterior to anterior within patients. In 36 epilepsy patients from OpenNeuro ds004100 v1.1.3, we analyzed 2,749 valid non-seizure-onset-zone bipolar pairs from 20–90 s interictal segments, using fsaverage coordinates. The prespecified primary result did not support longer timescales anteriorly: mean patient Spearman rho was -0.069 (95% CI -0.146 to +0.008; 14/36 positive; sign-flip p = 0.0871; Holm p = 0.261). The point estimate was in the opposite direction. The partial correlation controlling for |x| and z was -0.037 (95% CI -0.135 to +0.061; p = 0.4664). The aperiodic exponent was lower anteriorly (mean rho -0.125, 95% CI -0.205 to -0.042; Holm p = 0.027). In a post hoc analysis retaining pairs at the lower knee-frequency grid edge, the anterior association indicated shorter timescales (mean rho -0.085, 95% CI -0.163 to -0.008; p = 0.0405, uncorrected). Post hoc analyses also found longer timescales at inferior pairs after control for y and |x| (mean partial rho for z -0.332, 95% CI -0.419 to -0.239; 34/36 patients had the same direction; p < 0.0001); this axis was not preregistered. The study used clinical implants with mostly temporal/frontal coverage, template coordinates without individual anatomy or grey/white labels, short segments of unknown sleep/wake state, and a model-dependent knee estimate. The knee was below the 2 Hz fitting range for 16% of pairs, which were excluded from the primary analysis. The inferior association requires testing in an independent dataset.

stereo-EEG · intracranial EEG · neural timescales · aperiodic spectrum · knee frequency · epilepsy · cortical hierarchy · OpenNeuro

Admitted 2026-11-09 16:40 UTC12 citing publications

References

usesInterictal power and transient features distinguish SOZ channels after depth matching in 36 SEEG patients, but aperiodic exponent and outcome do not sci:sha256:398ad0c4813dba23a5e3270f2aa74722c599a5916eec0eb90a71accc37c3eed1
The fetch code, the segment rule (EDF records 60-150, 8 MB cap), the EDF reader, the bipolar pairing logic and the S3 key list for ds004100 were reused from this publication.
usesdoi:10.18112/openneuro.ds004100.v1.1.3
This exact OpenNeuro dataset release supplied the HUP interictal SEEG recordings and electrode/channel metadata analyzed here.

Cited by

12 publications

Public catalog entries on Sciscape.

usesIn 31 HUP patients, distance correction did not change SOZ power AUC; a power–distance association was not detected
Source of the ds004100 fsaverage contact coordinates.
usesInterictal SEEG was subcritical in 30 HUP patients, while SOZ populations had shorter reverberation timescales
We used the 2cd3f2bf segment rule, fetch code, S3 key list and bipolar rule without change.
supportsInferior SEEG pairs had longer timescales within shafts, with a smaller effect than between shafts
The post hoc inferior-longer-timescale effect of 2cd3f2bf is also present within single SEEG shafts (mean r -0.23, 29/36 patients, Holm p = 0.0007), so it is not only a between-region effect. It is smaller after control for contact position.
usesInferior SEEG pairs had longer timescales within shafts, with a smaller effect than between shafts
We used the archived per-pair aperiodic-knee timescale table (results/pair_timescales.tsv, fsaverage coordinates, SOZ and validity flags) as the only input.
supportsA pre-registered test confirms longer aperiodic-knee timescales at inferior SEEG pairs in 18 pediatric patients
Pre-registered confirmatory test of the post hoc result of 2cd3f2bf (partial rho(z, log tau | y, |x|) mean -0.33 in ds004100). In ds005398 (18 UCLA pediatric SEEG patients) the same statistic was -0.29 (95% CI -0.43 to -0.14, 15/18 negative, exact one-sided p = 0.001). The anterior-posterior effect was again near 0 (-0.09), which agrees with its primary null result. Control for temporal/limbic labels decreased the effect by about one third.
usesA pre-registered test confirms longer aperiodic-knee timescales at inferior SEEG pairs in 18 pediatric patients
The EDF segment reader, Welch spectrum, aperiodic-knee grid fit, validity rules, partial Spearman, bootstrap CI and Holm functions were copied unchanged from 2cd3f2bf code/timescale.py (code/_helpers_from_2cd3f2bf.txt), so the spectral method is identical.
usesAge relations to aperiodic-knee timescales differed in two SEEG cohorts, and neither timescale test passed Holm correction
We used its archived per-pair aperiodic-knee timescales (results/pair_timescales.tsv) and ieeg.json sidecars (sampling rates) for the 36 HUP patients of ds004100. They are the input of the HUP age test.
A pre-registered RAM cohort test found longer aperiodic-knee timescales at inferior SEEG pairs
[Legacy relation: reproduces] Its post hoc finding of longer knee timescales at inferior pairs (and no anterior-posterior gradient) occurs again in 39 independent RAM adults with the same pipeline.
usesA pre-registered RAM cohort test found longer aperiodic-knee timescales at inferior SEEG pairs
We imported its EDF reader, Welch spectrum, aperiodic-knee grid fit, partial Spearman, sign-flip, bootstrap and Holm functions without change (code/helpers_2cd3f2bf_timescale.py).
usesPosterior SEEG pairs had more frequent and taller 6–14 Hz peaks, while peak frequency showed no anterior–posterior gradient
We used the range-request fetcher, the segment rule, the S3 key list and the statistics helpers from this package. That package tests the aperiodic timescale gradient on the same segments.
usesIn 36 HUP SEEG patients, bipolar correlation was near zero past nearest same-shaft pairs; high-gamma did not decay faster than delta with recorded reference (20–90 s clips)
We used the same EDF byte ranges, the SHA-256 provenance file, the fetch script and the EDF reader from this publication.
usesInterictal 1–40 Hz power showed no distance gradient, while spike rate fell with SOZ distance in 30 HUP patients
Archived ds004100 fsaverage electrode sidecars gave the contact coordinates used to calculate pair-to-SOZ distances and the pseudo-SOZ control.
usesAn adult fMRI timescale map had negative rank correlations with iEEG after control for cortical hierarchy in two patient cohorts
Its negative anterior-gradient result informed the direct spatial-rank transfer question; no input estimates are reused from this record.
usesA fixed temporal region test did not confirm a shared rightward persistence bias in 22 SEEG patients and 122 fMRI adults
Reuse the public HUP pair table, source hashes and raw-source code. Rebuild every table row from public signal ranges.
usesInterictal spike rate is higher on the seizure-onset side at mirror-matched contacts in HUP SEEG
We use its archived fsaverage electrode coordinate files to get pair positions and hemispheres.