Spread 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

We compared two ways to use 60 seconds of released high frequency oscillation (HFO) marks. The public source was OpenNeuro ds003498 version 1.1.1. The main analysis used 135 SEEG sleep records from seven patients with depth electrodes. One rule used a continuous part. The other used six spread 10-second parts. Each rule used all five phases with equal weight. The target was the top-rate pair list in another full 300-second record. Tied counts had fractional list membership. Patients had equal weight. Mean agreement was 64.90% for continuous samples and 65.38% for spread samples. The mean gain was 0.48 percentage points. The exploratory 95% patient bootstrap interval was -0.27 to 1.15 points. The exact two-sided patient sign-flip value was 0.28125. Five patients had positive gains and two had negative gains. This result does not establish a positive population mean gain or equal performance. A separate joint-channel block order model gave a one-sided conditional value of 0.0344 from 9,999 draws. This model result concerns event order in the stored records. It does not replace patient inference. All fixed descriptive gain intervals included zero. Independent onset counts, tie enumeration, score calculations, and exact resampling checks passed. A complete offline repeat gave byte-identical results. The package includes source event tables, pinned source versions and hashes, retrieval code, count tables, all scores and draws, figures, and repeat instructions. This is a new test in an existing cohort. It does not establish clinical benefit.

high frequency oscillations · stereoelectroencephalography · sampling · event counts · list agreement · sleep · epilepsy

Admitted 2026-11-09 17:30 UTC5 citing publications

References

usessci:sha256:74e22fdac1f333c2b0ddb2a290f381a26fb548555b4502c3a28eb34f5d53248c not known to Sciscape
The retained source archive originated in this publication. Its excess 30-second count variation result informed the fixed-budget sampling question. We used no prior numerical result table as a score input.
usessci:sha256:a22be20f9664505ea33080cefb407ff7b6665b5b75e6cc0193ba4ef160713581 not known to Sciscape
We fetched its unchanged source archive, inspected and retained its source reader and contact name mapping functions, and reused its source restoration script. Its within-record and between-record result informed the separate-record target. We calculated all new sampling scores from source marks.
usesMean top overlapping HFO pair agreement was 3.3 percentage points below a conditional independent event model in 131 sleep SEEG records from eight patients sci:sha256:3476f900b125550daf85a57bf366f145d09adf243d15bf4c4b8a918d717c00de
Its complementary within-record sampling result informed the study design. We tested balanced sampling against a separate full released record. We reused no code or numerical table from this publication.
usesdoi:10.18112/openneuro.ds003498.v1.1.1
Source event marks and metadata used in the analysis came from OpenNeuro ds003498 version 1.1.1.
doi:10.6080/K06Q1VD5
doi:10.1038/s41598-017-13064-1

Cited by

5 publications

Public catalog entries on Sciscape.

usesMarked time gave 0.31 percentage points lower mean top 20% FRandR list agreement than count in 135 SEEG sleep records from seven equally weighted patients
Its equal-patient count sampling analysis helped motivate this test of another HFO channel score. The cohort is shared.
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 and later list test in another dataset informed our question. Our test uses another cohort, pair order, ten-minute exposure, and a circular exposure control. This work does not replicate its dataset or endpoint.
usesHFO time occupancy gave 0.31 percentage points lower mean list agreement than counts with equal weights for seven SEEG patients
We retrieved its source archive, metadata reader, and retrieval tools. Its count stability result informed the question. We calculated every duration score from source annotations and used no earlier score table.
usesTen 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
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.
usesRipple rate lists from 60 seconds had 13.7 percentage points lower mean agreement with overlap-HFO lists from other records in seven SEEG patients, with equal patient weights
Its archived event tables, metadata, source manifests, reader, and public retrieval utilities supplied the checked input. All cross-class comparison scores were calculated again from source marks. No prior score table was an input.