Peaks from five SEEG pulses did not show a clear mean timing error decrease over a shared patient time at 20 to 100 ms in 83 cases from eight people

We tested whether a separate peak time for each SEEG case predicts later peaks better than a shared patient time. Public HAPwave data supplied 83 selected remote limbic stimulation/readout cases from eight people and 830 distinct pulses. The contact order stayed fixed. Each case used five early pulses to predict the absolute peak in five later pulses, strictly between 20 and 100 ms. The shared rule used the median early peak at the other retained sites in the patient. Mean absolute errors were 17.36 and 18.54 ms, with equal patient weights. The paired difference was -1.18 ms; its patient bootstrap 95% interval was -5.89 to 3.58 ms. The evidence did not show a clear mean error decrease from a separate case time. It did not establish equal performance. All planned controls and screens are reported. This was an offline test in a reused cohort. It did not test interface control. The SCI package preserves exact voltage arrays, source event maps, hashes, results, and three figures. GPL Python and inherited MATLAB scripts are omitted, so direct re-execution is limited. A fresh source check matched all retained first-patient arrays exactly.

stereoelectroencephalography · electrical stimulation · peak timing · brain-computer interface · calibration · prediction error

Admitted 2026-11-09 20:26 UTC0 citing publications

References

usesMeans of five pulses had higher 10 ms peak time agreement than single pulses in 140 selected SEEG pairs from eight patient sessions, within a 20–500 ms window sci:sha256:ee9468576896d327d9f9ef2d36d9e91a4cd1e76b429b995fe523bd304320eab2
Its five-pulse versus single-pulse peak agreement result motivated the question of added value beyond a shared patient time. The current study reuses the HAPwave cohort and does not provide an independent cohort replication.
usesContact medians changed 20 to 500 ms bipolar SEEG templates under two voltage references at 83 sites in eight people sci:sha256:2c6e5aa829391c280e08aaf3531691e8ae03629674f83cb6f3d20e5cb0f3b3e7
Used its exact float64 voltage windows, 83 contact choices, pulse identities, pinned source hashes, and GPL retrieval and selection code. These inputs enabled the new peak prediction comparison without another full download of all eight source files.
usesA background model lowered mean voltage error with an early reference and increased mean shape error at 20 to 100 ms after five fit pulses at 86 sites in eight HAPwave SEEG sessions sci:sha256:8fb26571079cd151b4ee7e9f54e746a41971223ba27c94bcbf67ee008c4815fb
Used its inherited eligible remote limbic contact-pair tables, source metadata, and GPL source downloader through the voltage package. The current study chose one readout per site by contact labels and tested peak timing.
usesEarly 25–95 ms peaks for each readout did not show clearly lower mean error than a shared peak in 8 SEEG records with early waveform r ≥ 0.8 after recorded contact order reversal sci:sha256:a68dc452d4f69d2a860b44216273f4fa629e677c5292305ec79503fcbf8f2bc3
Its comparison of separate readout times with a shared condition time informed the prediction benchmark and early-data screens. The current study tests a different dataset and a shared time across other stimulation sites, with fixed contact order and an unsmoothed primary window.
usesdoi:10.18112/openneuro.ds004696.v1.0.1
Source dataset analyzed: HAPwave, OpenNeuro ds004696 version 1.0.1; the publication uses voltage windows derived from this dataset.
usesdoi:10.1523/JNEUROSCI.2201-22.2023
This is the source experiment paper for the HAPwave recordings analyzed here.
doi:10.1371/journal.pcbi.1011105
The paper describes the canonical response method associated with source files; this study gives context and does not compare its results against that method.

Cited by

0 publications

Public catalog entries on Sciscape.

No incoming citations on Sciscape.