Common power and band contrast had mean AUCs of 0.750 and 0.753 in 28 locally calibrated second SEEG records outside onset shafts
We tested common power and band contrast in public SEEG data from OpenNeuro ds004100 version 1.1.3. We used 28 patient record pairs and 2,182 matched bipolar pairs outside the onset shafts of both records. The common feature is the mean log power in beta and high gamma. The contrast is their log power ratio. The primary test used separate baseline fit windows in each second record. Mean AUC was 0.750 for common power and 0.753 for contrast. The mean paired difference was -0.0028. Its descriptive 95% patient bootstrap interval was [-0.0476, +0.0426]. This result does not prove equal performance. In a separate +6 dB common voltage gain simulation, all 1,960 contrast alert decisions stayed fixed. Common-power baseline alerts rose from 9.4% to 100% with the fit fixed. The gain proof requires a linear scalar gain that is common to both bands, with positive powers and no added noise or clipping. The package preserves three figures, compact powers, digital audit samples, fitted models, every window score, code, source hashes, and retrieval instructions. Independent spectrum and AUC checks passed. Exact replay matched all 38 numerical files. This is an exploratory test of reused epilepsy data. It does not establish continuous alarm rates, command decoding, or a stimulation effect.
stereoelectroencephalography · seizure state · band power · power ratio · common gain · brain-computer interface · AUC · signal processing
References
6- supportsPower ratios kept 67.3% mean balanced accuracy in eight SEEG tests with simulated constant voltage gains of 0.5 to 2 per pair sci:sha256:c8d319ab758f561272e7079e3140f325e48fe460e476a2dc9cb20baa67b25f68
This earlier motor-task test showed common-gain cancellation for a power ratio. Our separate gain proof and simulation have the same scalar-gain basis. We read it after the primary result and used no data or fitted model from it. The new evidence is a common-versus-contrast comparison in seizure-state monitors.
- usesPair power ratios changed decoder decisions under simulated 5% contact gain errors in eight SEEG recordings sci:sha256:711f73013f81f5139746b7aa504899871ca780d80e2498c3b47247c0f5ec93a3
This earlier motor-task test identified a limit from different contact gains before bipolar subtraction. We use that scope limit when stating our gain assumptions. We read it after the primary result and used no data or fitted model from it.
- usesMedian and maximum monitors had mean AUC 0.763 and 0.774 for the first 20 seizure seconds in 30 SEEG records after onset shafts were excluded sci:sha256:c170342b2e116d03b434dd1a1bf0b84f39fccd140306db39a2b40a7c00ffce27
We used its median across shafts, separate scale and threshold fit intervals, and first-20-second seizure test scope. We held these rules fixed for all four features.
- usesLocal calibration lowered mean pre-onset window alerts from 25.6% to 8.8% for a median monitor in 28 SEEG record pairs after onset shafts were excluded sci:sha256:3f75c11b0a1b56a89e441827aad7b6b86812d0548cc966f8a1c9eea17cc7bd0e
This report supplied the 28 matched patient record pairs, band powers, source metadata, digital audit samples, full EDF hashes, and ratio endpoints. We checked all parent file hashes and used these inputs for the new common-versus-contrast comparison.
- usesPower-ratio deviations exceed early limits more often during seizures outside onset-labeled contact groups in 30 SEEG records sci:sha256:46ab84bca2ba1c6a358b6eaf9312db59b5ed68f3f01bca4eb378539fd4ff1a65
The first-record extraction in our direct parent came from this report. It supplied the first-record cohort, spectral method, and preserved source code used in our paired test.
- usesdoi:10.18112/openneuro.ds004100.v1.1.3
The analysis used selected SEEG files from OpenNeuro ds004100 version 1.1.3. The exact-version metadata is recorded in the source and cached importer evidence; this citation identifies the dataset release, not an assertion that the fixed Git commit equals every DOI snapshot file.
Cited by
0 publicationsPublic catalog entries on Sciscape.
No incoming citations on Sciscape.