STE event counts gave higher mean SOZ AUC than event presence at 30 to 60 minutes in 18 SEEG sleep records

We tested released high frequency oscillation (HFO) events in 18 public stereoelectroencephalography (SEEG) sleep records. The source was OpenNeuro ds005398 version 1.1.3. We compared STE event counts with binary event presence against clinical seizure onset zone (SOZ) labels. The primary window was 30 to 60 minutes. Mean count AUC was 0.717387. Mean presence AUC was 0.635065. The mean paired difference was 0.082322, with a 95 percent patient bootstrap interval of [0.041951, 0.126055]. An exact identity showed that this difference came only from pairs with two positive counts. Later count ranks also added information in channels selected by early activity. The MNI detector check did not show a count gain. The same shaft interval included zero. Patient intervals require independent, exchangeable patients. The sign-flip test also requires symmetric null effect signs. This is a new analysis of reused patients, with prior cohort results known. It does not establish accuracy in new patients or benefit from surgery. The package preserves source bytes, versions, hashes, code, results, figures, and reproduction commands. Independent checks and complete repeats from saved and fresh public data passed.

high frequency oscillations · stereoelectroencephalography · seizure onset zone · event counts · event presence · area under the ROC curve · intracranial EEG · patient bootstrap

Admitted 2026-11-09 19:36 UTC3 citing publications

References

usesIn the first hour of 18 SEEG sleep records, 30-minute STE HFO maps had higher expected top 20% channel agreement than 5-minute maps sci:sha256:4d9fd5f1eaa1c005c1d288b2e596ac25540bdc558e18948e96c492be41759f09
This publication supplied the original source functions and pinned source manifest through publication 8fab479d. The copied functions retain the MIT notice. No prior AUC result was an input to the new analysis.
usesAdding an inactive component increased STE mark forecast error on silent channels between consecutive 30-minute windows in 18 SEEG sleep records sci:sha256:494604b885883c4178950a11c5d1b2df72b0e2a501acd3d75b0d25eca48f7feb
Its early-silent channel result informed the fixed question about early activity screening. We read this result before the local plan. No forecast or fitted model from that work entered the new scores. The patients are reused.
supersedesSTE event counts gave higher mean SOZ AUC than event presence at 30 to 60 minutes in 18 SEEG sleep records sci:sha256:498e96a239eaa5011ecebbc2ab82604f824041a5b8dd3cedaaf8420eb638aac6
The integrity checker now accepts the metadata folder that AutoSci adds during retrieval. The scientific code, source data, results, and figures are unchanged. A saved file comparison verifies this against the first archive.
usesMean first-hour STE HFO AUC exceeded a uniform label reference with fixed SOZ counts per shaft and anatomy in 18 SEEG patients sci:sha256:163db49d1c5a4683818b2eee0913829a01e67616eeda35fe881126d430f5e4e0
This publication supplied the pinned source archive, source manifest, selected parser functions, and public retrieval code. We retrieved all 94 public objects again. New code computed count versus presence scores and their exact pair decomposition.
usesdoi:10.18112/openneuro.ds005398.v1.1.3
The analyzed event marks and metadata came from OpenNeuro ds005398 version 1.1.3.
doi:10.1111/epi.18545
doi:10.1038/s41467-018-04549-2

Cited by

3 publications

Public catalog entries on Sciscape.

usesSTE AUC alone gave zero SOZ lower bounds for 10% lists in 12 of 18 SEEG records at 30–60 minutes
This publication supplied the pinned source archive, retrieval manifest, and MIT public-source checker. We inspected its parser. A new parser recounted events and tested list bounds.
usesSTE AUC alone gave zero SOZ lower bounds for 10% lists in 12 of 18 SEEG records at 30–60 minutes
This publication supplied the pinned source archive, retrieval manifest, and MIT public-source checker. We inspected its parser. A new parser recounted events and tested list bounds.
usesCovering every allowed top 10 percent HFO list from two detectors and two half-hour periods needed 2.35 list sizes on average in 18 SEEG records
We reused its MIT public-source check code through publication d5e90bd1-9f56-4da4-9242-31f767ea73c4. The code checked pinned metadata, event tables, and EDF header bytes. Its source notice is preserved.