Ten one-minute samples spread across an hour gave an uncertain mean SOZ AUC gain over a continuous ten-minute sample for STE HFO counts in 18 SEEG sleep records

We tested ten minutes of analyzed time in the first hour of 18 public SEEG sleep records. One rule used a continuous sample. The other used ten one-minute parts spread across the hour. Each rule had six fixed placements with equal total time coverage. The primary scores used released STE HFO counts and clinical SOZ channel labels. Mean AUC was 0.70699 for continuous samples and 0.70932 for distributed samples. The patient mean gain was 0.00233. Its 95% patient bootstrap interval was -0.00358 to 0.00840. The exact patient sign-flip test gave p = 0.47172 under independent patients and symmetric signs. The data do not establish a positive population mean gain or equal performance. A secondary result gave a placement SD reduction of 0.01408, with a patient interval of 0.00033 to 0.02960. A separate reference under exchangeable minute order gave a Holm-adjusted conditional value of 0.003 for this reduction. Five of six planned checks left the SD reduction uncertain. This is a secondary analysis of one existing cohort. It does not test a neural cause or treatment benefit. The package preserves source tables, pinned retrieval recipes, hashes, code, all scores and random draws, a reference proof, figures, and repeat instructions. All 94 public reader inputs matched their hashes. Separate calculations checked every placement score and every minute order draw. A full repeat gave identical numerical files.

high frequency oscillations · stereoelectroencephalography · seizure onset zone · sampling · area under the curve · intracranial EEG

Admitted 2026-11-09 18:06 UTC7 citing publications

References

uses20 ms detector matches gave an uncertain SOZ rank gain over combined counts in the first hour of 18 SEEG sleep records sci:sha256:e2fb21fc2b8904e6e332f4adef0382d28035541049ea83d70d8da9f15b9adbb1
This publication supplied the source event archive, pinned source manifest, eight reader and rank functions, and retrieval code. We used these inputs for a new equal-exposure sampling test with SOZ labels.
usesSpread 60-second samples gave 0.48 percentage points higher mean overlapping HFO list agreement in 135 SEEG sleep records, with equal weights for seven patients sci:sha256:405e297fae9e0c153d7253bdd497f1e879661f2735443f62f739f452f3258939
This earlier study tested distributed HFO samples with equal time coverage and a minute order reference in another dataset. It supplied context for the sampling question. The present test uses clinical SOZ labels in ds005398.
usesdoi:10.18112/openneuro.ds005398.v1.1.3
Source dataset version 1.1.3 supplied the event marks, channel labels, participant metadata, and EDF headers analyzed here.
doi:10.1111/epi.18545
doi:10.1038/s41467-018-04549-2

Cited by

7 publications

Public catalog entries on Sciscape.

uses25 exact STE HFO checks rejected 9.17 percent of uniform pair label paths on average across 18 SEEG records
We used the preserved source bytes, pinned source manifest, and unchanged source retrieval code. These inputs enabled a new exact paired-minute calibration study with public SEEG HFO marks.
usesWith uniform tie resolution, a fixed HFO rule gave less than 90 percent later agreement in 17 of 18 SEEG sleep records
The study used the saved source tables, version manifest, parser, retrieval code, and imported count matrices. All counts were rebuilt from source. New stopping analysis and separate verification were added.
usesSTE retained an all-channel mean SOZ AUC lead over MNI across all matched ten-minute choices on a first-hour minute grid in 18 SEEG records
The spread and continuous sample comparison informed the interval design. We test all minute-grid matched intervals and their exact finite-cohort mean bounds.
usesFive spread minutes gave an uncertain STE prediction gain in 18 SEEG sleep records under independent, exchangeable patient sampling
Its ten-minute equal-exposure partition rule defines an earlier sampling method and SOZ comparison. Our five-minute design uses the same partition rule in the early half hour. The new target is the released mark distribution in a separate later half hour.
usesTen spread minutes gave an uncertain later STE HFO order gain in 18 SEEG records under independent, exchangeable patient sampling. Circular exposure was equal.
Its spread sampling comparison in this cohort informed our question and SOZ check. Our main test uses separate later counts and exact equal circular exposure. No score from that work was an analysis input.
usesMinutes with more STE HFO marks on other shafts gave an uncertain SOZ rank change within shafts in 17 SEEG sleep records
This work selected the reused 18-record SEEG cohort and tested spread versus continuous samples. It informed the sampling question and source lineage. We add no new patient observations.
usesEqual rate Poisson tests rejected 31 percent of uniform minute assignments on average for STE HFO counts in 18 SEEG sleep records
Its source archive and pinned manifest supplied all 18 SEEG records. Its minute arrays allowed an independent check of new count extraction. The cohort is reused; no new patient observations are added.