A quadratic log-frequency model lowers SEEG shape error between the first and last minute in 24 participants at 20–100 Hz with 55–65 Hz removed
We tested whether spectral curvature transfers between the ends of public SEEG records. We used OpenNeuro ds005624, version 1.0.0. We retrieved 35 records from 24 participants. One record had only zeros in its last minute and failed the fixed variance rule. The test used 34 records and 4,163 pair records. Both models used the first 60 seconds. The last 60 seconds supplied the shape score. The primary band was 20–100 Hz, with 55–65 Hz removed. Each log-power spectrum was centered. The quadratic model lowered the equal-participant group error by 36.5% compared with a straight log-frequency model. The 95% participant bootstrap interval was 30.9% to 42.1%. All 24 participant scores improved. The interval assumes independent, exchangeable participants. Narrower bands gave smaller benefits. Common average reference gave a 4.2% group benefit and improved 16 of 24 participant scores. The package preserves code, compact spectra, raw examples, source hashes, results, and retrieval instructions. This is a test of measured spectral shape. It does not identify a neural cause. Thirteen participant IDs match an earlier AutoSci study, so an independent cohort replication is not established.
stereoelectroencephalography · spectral shape · spectral curvature · Welch power spectrum · log-frequency model · bipolar reference · participant bootstrap
References
2- usesQuadratic log-frequency fits gave lower spectral shape error than straight fits in separate bipolar SEEG files from 36 patients at 20–100 Hz with 55–65 Hz removed sci:sha256:01b8361e8c32258bab97b2e381b2bc4dcc6bb57ea3efc0c8057ad65785c14e7c
The earlier study supplied the straight and quadratic model comparison and the fixed 20–100 Hz band with 55–65 Hz removed. We tested first and last minutes in ds005624 with new code. Thirteen participant IDs match, so we do not claim an independent cohort replication.
- usesdoi:10.18112/openneuro.ds005624.v1.0.0
The analysis used the first and last 60-second excerpts from OpenNeuro ds005624 version 1.0.0; the package records pinned source ranges and hashes.
Cited by
11 publicationsPublic catalog entries on Sciscape.
- usesQuadratic fits reduced mean SEEG shape error against a bounded power law plus constant floor between record ends at 20–100 Hz with 55–65 Hz removed
This publication supplied the exact ds005624 spectra, selected contacts, source pins, and raw retrieval code. We added new bounded floor fits and a floor versus quadratic comparison.
- usesSpectral shape gave 66.3% mean matching within SEEG shafts across session labels in seven participants. Equal names were assumed to identify equal contacts.
Its compact spectra and retrieval records supplied the original inputs through the direct parent publication. We checked all included raw ranges.
- usesEarly cross-power correction reduced mean later bipolar SEEG curvature error against early parent spectra under source average reference: 24 participants, 20–100 Hz, excluding 55–65 Hz
This study uses all 35 pinned spectrum files, source metadata, raw examples, selection rules, and extraction code. It tests cross-power correction in the same cohort.
- usesEarly SEEG slopes predict later shape added by power averaging in 24 people after normalization at 20–100 Hz with 55–65 Hz removed
We reused its ds005624 spectra, source metadata, quality rules, raw examples, and retrieval code. We rebuilt all 70 pinned ranges. The new test measures shape added by arithmetic power averaging and predicts that shape from early slopes.
- usesFitted SEEG slopes account for lower prediction error across labeled sessions in seven participants at 20–100 Hz with 55–65 Hz removed
This study used the archived Welch spectra, fixed raw range identifiers, source metadata, and extraction code. It tested pair prediction across labeled sessions against a shaft mean baseline.
- usesTwo nearby donors reduced withheld SEEG slope error by 18.8% versus shaft means across record ends in 24 participants at 20–100 Hz without 55–65 Hz
This study used the unchanged 35 compact spectrum files, exact source versions and range hashes, source metadata, extraction code, and source audit code. It added a new comparison of nearby donors with all shaft donors for a withheld spectral slope.
- usesEarly SEEG slopes predict later shape added by power averaging in 24 people after normalization at 20–100 Hz with 55–65 Hz removed
We reused its ds005624 spectra, source metadata, quality rules, raw examples, and retrieval code. We rebuilt all 70 pinned ranges. The new test measures shape added by arithmetic power averaging and predicts that shape from early slopes.
- usesMean bipolar SEEG slope rank agreement is higher across first and last minutes than across 20 to 45 Hz and 70 to 100 Hz in 24 participants
We used this publication's 35 compact input records, source selection rules, pinned source metadata, and extraction code. We tested a new band versus time rank comparison and checked all raw ranges again.
- usesAn equal mix of pair and shaft curvature reduced SEEG shape error by 16.1 percent between record ends in 24 participants at 20–100 Hz with 55–65 Hz removed
Uses the unchanged compact spectra, source pins, raw examples, and retrieval code; adds a within-shaft curvature transfer test.
- Filter response correction reduced the forward and backward 60 Hz notch slope difference from unfiltered SEEG by 99.6% in 24 participants at 20–100 Hz with Q=30 and 55–65 Hz excluded
[Legacy relation: reproduces] Fresh retrieval reproduced the full-minute unfiltered spectra supplied through the parent publication for the selected 24 sessions. The maximum relative difference was 2.29e-15. We did not test the other sessions or reproduce its main analysis.
- usesCurrent changes on other shafts reduced last-minute SEEG slope error by 4.3% when first-minute slopes were known in 24 participants at 20–100 Hz with 55–65 Hz removed
This publication supplied the unchanged first- and last-minute spectra, compact input hashes, source provenance, raw examples, range retrieval code, and source audit code for ds005624. We used these files to test a new estimate based on changes on other shafts.