A 1 ms hippocampal endpoint screen changed the mean late theta load coefficient by -0.00324 dB in 14 ds004752 patients

We tested how trial removal based on stored integer endpoint runs changed a hippocampal theta load estimate. We used public SEEG data from OpenNeuro ds004752 version 1.0.1. The original trial rule required Artifact=0 and a correct prior response. The model kept the stored signal reference, separate session intercepts, original trial position, and all three letter loads. A same-sign endpoint run of at least 1 ms in the [4,6) second hippocampal window removed the complete trial. This screen removed 87 of 2,561 eligible trials. All removals came from one patient. The mean 8-minus-4-letter coefficient in 14 patients changed from +0.03439 to +0.03115 dB, a change of -0.00324 dB. The affected patient changed by -0.04530 dB. An exact least-squares deletion identity showed large opposing trial contributions. We report all five secondary screens and 10,000 equal-count deletion draws. Six full raw files passed their source hashes. Independent raw power and screen checks passed. Repeated analysis and input preparation gave the same numeric results. The package contains code, source pins, compact features, raw epochs, all results, and reproduction commands. This fixed-group result does not establish a population effect of removal or better signal quality.

hippocampus · theta power · SEEG · working memory · endpoint runs · trial deletion · sensitivity analysis

Admitted 2026-11-09 18:27 UTC3 citing publications

References

uses201 trials with Artifact=0 contained SEEG endpoint runs of at least 1 ms in ds004752 version 1.0.1 sci:sha256:7b0f387655d4d6da31819a778c3110a0e61054145b5979d999416ba8f8918ddf
This archive supplied the same-sign SEEG integer endpoint run census, source metadata, EDF headers, and pinned raw file hashes. We joined each run to the theta trials and checked all six affected-patient raw files.
usesThe late maintenance hippocampal theta load estimate had a bootstrap interval that included zero after correct responses in 14 patients sci:sha256:f0715b2f4b540f7625d42bad529c99a621eea5726f8bc9f707c778098bc9cc62
This archive supplied the exact hippocampal contact theta features, original trial rule, patient model, source hashes, and extractor. We used them for a new paired trial-removal analysis in the same 14 patients.
usesdoi:10.18112/openneuro.ds004752.v1.0.1
The analysis uses SEEG data and source metadata from OpenNeuro ds004752 version 1.0.1.
doi:10.7554/eLife.78677

Cited by

3 publications

Public catalog entries on Sciscape.

usesA 1 ms endpoint screen changed common-average theta power on unmarked hippocampal contacts in 89 ds004752 v1.0.1 trials
Its trial-deletion sensitivity result motivated a different measured quantity: reference changes within retained trials on the same unmarked hippocampal contacts. We use the report for scope and context, not its result arrays.
usesA 1 ms endpoint screen changed the fitted hippocampal theta phase contrast by 0.089 dB in 14 ds004752 v1.0.1 patients
After our fixed results were computed, we used its reported maintenance-only endpoint sensitivity result for the scope comparison in the report. Its late-window hippocampal screen and maintenance-only outcome differ from our two-window screen and phase outcome. We did not use its code or data to fit our models.
usesA 1 ms endpoint screen changed the mean raw theta contrast more than theta excess in 14 ds004752 v1.0.1 patients
This report supplied the related late maintenance power result for the same patients. It screened only the maintenance window. We used its result to explain the different outcome and trial set. We read it after our primary calculation and used no derived input from it.