Two early measures reduced loss for five-pulse SEEG error estimates at 20–500 ms in six of eight patients with epilepsy

We tested estimates of later SEEG waveform error from five early stimulation pulses. The public sample had eight patients with epilepsy, 86 sites, and 886 recording instances. Each test patient was excluded from model fitting. The model used early mean waveform energy and a ratio of variance before and after the pulse. At 20–500 ms, mean squared log error was 0.2995, compared with 0.3275 for one scale fitted from other patients. The reduction was 8.56%. Six patients had lower loss. The mean change was -0.02804. A 95% bootstrap range that repeated model fitting was [-0.06260, 0.01682]. This range includes zero change. The before-pulse control also improved. The result does not establish a stimulation-specific gain or a gain in new patients. The package preserves exact waveform products, all models and scores, two figures, code, source metadata, retrieval instructions, and an early-input forecast tool. Fresh public-source checks covered three patients. A second run reproduced all numerical result files exactly.

stereoelectroencephalography · SEEG · waveform error · cross-patient prediction · epilepsy

Admitted 2026-11-09 18:17 UTC2 citing publications

References

usesFive later pulses did not show higher SEEG size correlation across shafts at 20–500 ms than before stimulation in eight sessions with readouts selected from five early pulses sci:sha256:3e568cd9997feedb9652f1ade1733ba15c832a7040976086282aa09650536328
The inherited compact trial Gram representation and same-block reference cancellation support exact calculation of the tested waveform metrics. The immediate input package supplies the corrected chronological pulse selection.
usesA five-pulse check selected more repeatable bipolar SEEG pairs within eight epilepsy patient sessions sci:sha256:b5b7963746f5c6d2a2a8f7844ec590732a4eb152bc7a70979b3bc95c3e27820f
We inherited the public-source extraction recipe through the immediate input package. We executed its current version to rebuild three patients from pinned source hashes.
usesA fixed 1.5 factor reduced relative error estimate differences for five-pulse remote SEEG means at 20–500 ms in eight patients sci:sha256:ea314f0cf0b1d18b61f04d53605a2b63e560f1628fda25199c0cc18d30ae31a2
The fixed sample-mean error correction and its score-dependent results informed our original C baseline and the comparison with a fitted error estimate. We extend the scalar correction question with donor fits that exclude each test patient.
usesMean scaled SEEG waveform error at 20–500 ms was 0.032 below a five-pulse estimate in eight patient sessions sci:sha256:6546d7b768dee897061cfc0f38d7ba2a1bf9b4733bb01186ca978a0b5855224e
We used all first-five and final-five centered bipolar Gram products, site and recording-pair tables, source metadata, and the event chronology correction from this package. These inputs define the same fixed eight-patient sample.
usesdoi:10.18112/openneuro.ds004696.v1.0.1
The analysis used the exact trial products, source metadata, and fixed eight-patient sample derived from OpenNeuro ds004696 version 1.0.1.
doi:10.1523/JNEUROSCI.2201-22.2023

Cited by

2 publications

Public catalog entries on Sciscape.

usesShared removal gave mixed 20–500 ms SEEG forecast effects and raised error on shafts with simulated gain 4 faults after five calibration pulses in eight people
This publication supplied the compact pulse inner products, recording-pair tables, site tables, and source metadata. We kept these inputs unchanged and applied a shared calibration pulse decision.
usesA before-pulse energy model gave 1.1 percent more SEEG error forecast loss after five fit pulses at 85 selected sites in eight people
Its five-pulse mean error study informed our choice to test individual pulse error as a separate target. This result uses related data and is not independent replication.