Earlier sustained 6 dB fast-to-slow power rises gave higher mean SOZ label AUC than rise occurrence within 20 seconds in two seizures from each of 26 SEEG patients
We tested whether event timing adds separation of clinical seizure onset zone (SOZ) labels beyond event occurrence. We used two seizures from each of 26 stereoelectroencephalography (SEEG) patients in OpenNeuro ds004100 version 1.1.3. The fixed marker required a 6 dB fast-to-slow power ratio rise in two consecutive one-second blocks. Fast power was 20–50 Hz. Slow power was 2–12 Hz. The test window was 20 seconds after source onset. We compared local bipolar pairs within 56 shafts. Positive pairs had SOZ labels on both contacts. Negative pairs had no SOZ contact. Mean area under the curve (AUC) was 0.724 for earlier rise time and 0.643 for rise occurrence. The equal-patient difference was 0.082. The 95% patient bootstrap interval was 0.034 to 0.128. The primary sign-reversal p value was 0.0030 under independent patients and a sign-symmetric null. Timing had higher AUC in 18 patients, lower AUC in 5, and equal AUC in 3. An exact identity showed that only order among two detected rises adds AUC. All planned checks had positive mean differences. The three-block sustain check had an interval that included zero. Fresh feature and result calculations matched the archived arrays and tables. The package preserves source versions and hashes, 2,742 pair feature sets, one raw example, all patient results, code, proof, and retrieval instructions. The result concerns a fixed marker and source clinical labels in one selected cohort. It does not establish causal tissue recruitment or surgical benefit.
stereoelectroencephalography · seizure onset zone · power ratio · event timing · AUC · epilepsy
References
4- usesWith a common average reference, early gamma change gave lower mean SOZ label AUC than pre-onset power within contact name groups in two seizures from each of 30 SEEG patients sci:sha256:929844fa4f36e3ba985c7214a9ce1b9a99d9120defae189e5e09618636b3c65c
The prior comparison of gamma change with pre-onset power informed the fixed relative-power question and the focus on what event timing adds to occurrence. This study retrieved source signals anew and did not reuse prior code or feature files.
- usesWith a common average reference, a shared baseline increased early seizure gamma rank agreement in 31 SEEG patients sci:sha256:54e4a0a2a07493893e8898f4d1fddee90b37464b6dfe38ec1d48f085bc2e33b6
The prior shared-baseline result informed the use of a separate baseline for each seizure and the explicit reporting of constant-score shafts in cross-seizure rank agreement. This study did not reuse prior code or feature files.
- usesdoi:10.18112/openneuro.ds004100.v1.1.3
The study uses SEEG recordings and clinical SOZ labels from OpenNeuro ds004100 version 1.1.3, whose saved dataset metadata declares CC0.
- doi:10.1093/brain%2Fawn111
Cited by
11 publicationsPublic catalog entries on Sciscape.
- usesMean 6 dB rise detection increased after marked seizure onset in paired 10 s periods from 26 selected SEEG patients
We used its fixed cohort, source metadata, headers, pinned range records, and MIT helper code. We added the pre-onset blocks and tested a new endpoint.
- usesA fixed contact number rule gave higher mean label concordance than 6 dB rise timing at 61 boundaries in 26 SEEG patients with two seizures each
This study supplied the 26 patient cohort, signal retrieval code, EDF decoder, feature code, quality rules, metadata, headers, pinned range receipts, saved powers, and raw example. We checked all powers against new source downloads and added the local boundary analysis.
- usesMedian offsets from other shaft groups reduced mean SEEG power ratio prediction error before seizure onset in 26 selected patients
This work supplied the selected patients, band powers, metadata, source extraction code, receipts, and raw example.
- usesUnder patient independence, the slow power denominator gave no clear SOZ timing AUC gain at 6 dB in 26 SEEG patients with two seizures each
We reused the fixed 26 patient cohort, 52 seizure windows, band powers, extraction code, and pinned source provenance. The new primary test compares ratio timing with fast power timing alone.
- usesA correction from other shaft groups reduced mean transferred spectral baseline error in 26 selected SEEG patients
Its cohort selection, spectral feature code, source metadata, range receipts, and power arrays were carried through the immediate parent. We rebuilt all features from new source downloads. This is the same public cohort, not an independent cohort test.
- usesFast power did not exceed baseline during 29 of 446 first sustained 6 dB ratio rises in 26 SEEG patients
We used the unchanged power features, fixed patient and pair selection, source metadata, pinned range receipts, and extraction code. They enabled the new event component calculation and fast power requirement tests.
- usesThe mean 6 dB spectral order score was 0.12 within label classes across two seizures in 18 SEEG patients
We reused its 26-patient feature package, pinned source metadata, and retrieval code. Its prior timing correlation informed the new within-label order question. We rebuilt all features from source bytes. The new comparison rule requires two detected rises, the same label, one shaft, and no common contact. The parent cohort is not an independent replication.
- usesTransferred baselines increased mean pre-onset alerts for a 6 dB rule when both directions were averaged in 26 selected SEEG patients
We reused its 26-patient feature package, cohort, pinned source metadata, and retrieval code. Its timing result informed the calibration question. All 52 fresh public source ranges matched their saved hashes, and every feature array rebuilt exactly. The new test uses all valid pairs and separate pre-onset calibration and evaluation windows. This is not an independent cohort test.
- usesA 3 dB fast power rule delayed or removed 131 of 446 ratio events in two seizures from each of 26 SEEG patients
We used its fixed 26 patient cohort, feature arrays, source metadata selection, retrieval code, and power extraction code. We added a same block fast power rule and a paired AUC test. Fresh source ranges rebuilt all features, and the reference AUC matched the cited result.
- usesA 6 dB spectral rise rule gave a mean order score of 0.12 within label classes across two seizures in 18 SEEG patients
We reused its 26-patient feature package, pinned source metadata, and retrieval code. Its prior timing correlation informed the new within-label order question. We rebuilt all features from source bytes. The new comparison rule requires two detected rises, the same label, one shaft, and no common contact. The parent cohort is not an independent replication.
- usesUnder patient independence, the mean 6 dB rise time order score interval included zero within SOZ classes in 19 SEEG patients with two seizures each
We used its 26-patient feature arrays, public source pins, metadata, headers, and extraction code. These files enabled the new within-label time order test. Fresh public signal sections reproduced its archived power values exactly.