Six CAR spectral features raised mean seizure onset label AUC beyond contact number. Fixed logistic tests excluded each of 31 SEEG patients.
We tested whether six fixed spectral features add seizure onset label ranking value beyond contact number. The source was OpenNeuro ds004100, version 1.1.3. The test used 31 SEEG patients and 2,879 contacts. Each test patient was absent from model training. Adding common average reference spectral features raised mean patient AUC from 0.693 to 0.767. The mean increase was 0.074. AUC increased in 25 patients and decreased in six. A 500-resample patient bootstrap refitted both fixed logistic models. Its 95 percent percentile interval was [0.017, 0.121]. This internal estimate assumes exchangeable patients. Planned reference, coordinate, parsed prefix, and patient exclusion checks also gave positive mean differences. The package preserves source spectra, labels, source hashes, retrieval recipes, code, all ordinary test scores, resample records, figures, and reproduction commands. The evidence uses one selected clinical cohort. It does not establish a causal marker, surgical benefit, or performance at another hospital.
stereoelectroencephalography · seizure onset · spectral features · patient-level validation · logistic regression · AUC
References
7- usesContact splits gave higher mean seizure-onset label AUC than patient splits for a fixed spectral random forest in 31 public SEEG patients sci:sha256:b6a361054e4ebea1dd4b8f14fdd99003ce2639da49c389eca05a7a1c82bba254
We used its complete saved input package for 31 patients. Its spectra features, clinical labels, metadata, audits, and source code enabled the matched model comparison. All 403 copied files retain their original hashes.
- usesLower contact numbers gave higher mean seizure onset label AUC than higher gamma fractions in two 60-second SEEG samples from 36 patients sci:sha256:7f57f55fb0912992a415d07349886a14b5bb293e801fb2771f8de060871ebcd1
Its contact number result motivated the fixed metadata comparator. Its source name audit motivated the planned exclusion of HUP142 and HUP166. We imported no model scores from this study.
- usesIn 30-second HUP SEEG samples, reference changes reduced mean 30–100 Hz slope rank agreement more than run changes sci:sha256:b9c7b4646dba303292f492835f91dbb8026d4f35fcce938519d1d3fa6bcbff63
Its original SEEG spectra, source metadata, patient audits, and pinned signal retrieval recipes were retained through the direct input publication. They enabled independent feature checks and two new source signal checks.
- usesdoi:10.18112/openneuro.ds004100.v1.1.3
Source SEEG signals and clinical contact labels from the exact analyzed OpenNeuro version.
- doi:10.1093/brain%2Fawab480
Prior normative intracranial EEG spectral study cited to motivate a placement check.
- doi:10.1038/s41598-024-56990-7
Prior work on temporal and patient generalization in intracranial EEG.
- doi:10.3389/fnins.2024.1373515
Prior work on subject holdout in scalp EEG.
Cited by
11 publicationsPublic catalog entries on Sciscape.
- usesAdding six spectral features increased mean seizure onset label recall within a 10 percent contact budget in 31 SEEG patients
This study uses the fixed six-feature model definition, cohort rules, parent contact table, and preserved source archive. It keeps the two recordings separate and tests label recall at fixed contact budgets.
- usesThe mean spectral AUC gain stayed positive under up to one adjacent label exchange per patient. The test used fixed scores and onset counts in 31 SEEG patients.
This publication supplied the fixed test scores, contact features, model parameters, and patient selection used for the exact label bounds.
- usesA prior fit without labels increased mean squared label error for a fixed HUP model in 18 external SEEG patients
The parent preserves this source HUP training table, model rule, compact spectra, and Newton logistic solver. They supply source patient exclusion fits and an independent numerical fit check.
- usesAn odds correction reduced mean squared label error in 31 SEEG patient exclusion tests with mean features from two recordings.
This study uses its fixed 31 patient cohort, contact features, model, and saved scores. A new solver also reproduced all 2,879 primary balanced scores within 0.0000005.
- usesNested calibration gave mean Brier score 0.0879 compared with 0.0886 after frequency correction for seizure onset labels. Tests excluded each of 31 SEEG patients from fitting.
This study fixes the cohort and the seven inputs. We used its copied features, saved model scores, and source archive.
- usesPatient rank scaling did not show an AUC gain in 18 external SEEG patients under fixed model and site rules
We inherited its six spectral features, fixed logistic fitting rule, HUP labels, spectra, and raw source code through the external test archive.
- usesBeta and high gamma power increased observed mean onset mark AUC beyond contact number in 19 selected external SEEG patients
The reported HUP comparison of contact number and six spectrum features informed the baseline and question. We used no fitted coefficients or code from that model.
- usesThe 95 percent interval with patient bootstrap refits included zero gain from six spectrum features at 88 SEEG label boundaries in 26 patients
This study supplies the fixed six spectrum features and weighted logistic model with each test patient excluded. We add a consecutive-number label boundary graph and patient bootstrap refits. This is a secondary analysis of the same cohort.
- usesAn added intercept reduced mean SOZ label probability error in 31 SEEG patient exclusion tests with a fixed logistic model
We used its fixed seven-input class balanced logistic design. Our new endpoint was probability error in excluded patients, with a nested added intercept.
- usesSix features added to a fixed SOZ model gave a 0.014 mean treated contact AUC gain in 17 SEEG patients with source outcome S
This publication supplied the fixed models, features, archived predictions, 31-patient cohort, Newton solver, and preserved signal source files.
- usesWith independent patients and null sign symmetry, tests detected no HUP model SOZ AUC gain in 18 external SEEG patients
We used the verified HUP contact table, six-feature model rule, original CAR spectra, and training retrieval code. They define the fixed training input and permit its repeat.