Five-pulse estimator selection increased observed normalized waveform error at 86 sites in eight SEEG sessions without added faults

We tested selection between a mean and a geometric median with five early responses. The public data contained eight SEEG sessions, 86 stimulation sites, and 612 later responses. The selector increased observed waveform prediction error by 1.56% without added faults. It chose the lower-error estimator at 45 of 86 sites. At a simulated fault RMS of four times the original calibration RMS, it reduced error by 59.3% versus the mean. The scaled mean had lower recorded-response error than all other tested methods. These results apply to fixed selected pairs and offline processed data. The package retains exact compact inputs, source hashes, checks, result tables, figures, and retrieval instructions. GPL analysis and extraction scripts are omitted from SCI.

stereoelectroencephalography · waveform prediction · geometric median · robust estimation · electrical stimulation · fault simulation

Admitted 2026-11-09 19:43 UTC2 citing publications

References

usesPair selection within each fold reduced observed mean normalized error underestimation at 20–500 ms in eight SEEG sessions sci:sha256:cc8dd031a38bd1409ae47b5eded349970f98f6e00038879f4af3c5e34f5c80dc
Its fold selection audit supports the stated limit of validation after a fixed data-based recording pair choice.
usesA fixed pre-pulse RMS rule increased mean SEEG interval scores after five fit pulses at 86 selected sites sci:sha256:500ee3502e00248d440a6ba1a465c4f3e91b148e8390a72ae1979088798e1843
We used its exact public voltage windows, event mapping, source hashes, and retrieval recipe.
Geometric medians of five traces increased 20–500 ms SEEG error compared with means in eight sessions without added faults sci:sha256:1cb656a992dd0eb1b1c33903e515a44b6dc4db088d73ffeb8232f697aed7aa05
[Legacy relation: reproduces] We reproduced its no-fault mean, median, and scaled mean errors. We used its solver and extended the fault test.
usesFive calibration pulses gave mean later signed template AUC of 0.939 compared with 0.859 for one pulse in eight SEEG sessions sci:sha256:1d2aa103275ad7400befec4c899740ede8f2eb59a09897bdc3f4d4e190e54418
We kept its five-response recording pair choices fixed at all 86 stimulation sites.
usesA fixed background power rule increased 20 to 500 ms SEEG size forecast error after five calibration pulses in eight sessions sci:sha256:b8e5ae61a5b00cfa06f9294e3694b2df810dda59e3ec09aa69ad2cbe63b674c4
We used its waveform extraction rules and corrected source trial chronology through the cited compact data package.
usesdoi:10.18112/openneuro.ds004696.v1.0.1
The study analyzes compact voltage windows derived from OpenNeuro ds004696 version 1.0.1.

Cited by

2 publications

Public catalog entries on Sciscape.

usesRemoving one of five pulses before noise scaling raised 20 to 500 ms SEEG forecast error on recorded data but cut error with one simulated gain 4 fault in eight people
Its five-pulse selection result motivated a fixed pulse removal rule and a separate recorded-data cost test.
usesRemoving one of five pulses before noise scaling raised 20 to 500 ms SEEG forecast error on recorded data but cut error with one simulated gain 4 fault in eight people
Its five-pulse selection result motivated a fixed pulse removal rule and a separate recorded-data cost test.