A linear input from voltage size before a pulse, fit in other patients, increased SEEG forecast error at 86 readouts in eight epilepsy sessions
We tested whether voltage size before an electrical pulse improved forecasts of voltage size after it. We used public SEEG data from OpenNeuro ds004696 version 1.0.1. The test used 612 later pulses at 86 fixed readouts in eight epilepsy sessions. Each readout was selected with the first five good pulses only. The reference forecast used the mean log size of all available earlier pulses at the same site. The added input used a linear slope fit with early pulses from the seven other patients. Both forecasts had the same local history. With equal person weights, squared log error rose from 0.053507 to 0.056043, an increase of 4.74 percent. The primary error reduction was -0.002536. Its descriptive 95 percent person interval was [-0.003979, -0.001201], with fits fixed. Error increased in seven of eight patients and in each planned history and window check. The package contains exact voltage windows, pulse forecasts, three figures, source hashes, retrieval code, and reproduction steps. We checked two complete public sources and verified the timing and equations. This is a result for one linear input on a reused clinical cohort. It does not show that all brain state inputs fail or test a stimulation control policy.
stereoelectroencephalography · electrical brain stimulation · voltage size · forecasting · epilepsy · brain-computer interfaces · signal processing
References
7- usesPast pulse updates reduced normalized 20–500 ms SEEG forecast error versus fixed five-pulse templates at 86 selected sites in eight epilepsy sessions sci:sha256:fd07147e53abc31f50ad87cee1d777a9d91b45b3d2a172703485899005f00530
Uses its exact 1,042 pulse windows, event joins, source metadata, and retrieval code at 86 fixed readouts. The new test forecasts log voltage size with an input from before the current pulse. It does not compare full waveform losses with size losses.
- 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
Uses its first-five readout choices, candidate contact tables, and selected comparison windows through the immediate input archive. Later forecast error did not select the readout.
- usesRecorded pulse gaps from 2 to 8 seconds had no clear mean SEEG gain relation in eight sessions with site and order controls sci:sha256:77ca2cd94f79de36a61ae64814637757c4181c890e0e63769f25e428fe358ac5
Uses its complete good pulse extraction and source retrieval recipe through the immediate input archive. The exact windows and source row mapping support the new before-pulse size input test.
- usesdoi:10.18112/openneuro.ds004696.v1.0.1
Source data are the processed SEEG voltage arrays in OpenNeuro ds004696 version 1.0.1; the dataset declares CC0.
- usesdoi:10.1523/JNEUROSCI.2201-22.2023
Primary publication describing the OpenNeuro dataset and its human limbic stimulation recordings.
- usesdoi:10.1371/journal.pcbi.1011105
Source CRP method credited for preprocessing the dataset derivatives used here; the fitted CRP response is not the target.
- doi:10.1111/epi.17528
Separate-cohort background on trial-by-trial response variation and phase dependence; it supplies context, not data or a direct replication target.
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesAn updated local RMS model did not establish lower SEEG forecast error at 86 selected readouts in eight sessions
Its earlier-input RMS result and matched response-history mean motivated the present test of local slope updates. We kept the reference tied to the same completed target history. The new model, input window, and response window differ.
- usesWith five local calibration pulses, a STReEF slope did not establish lower 20–500 ms SEEG size forecast error in eight HAPwave patients
The earlier HAPwave test of slopes from other patients supplied context. It used a different input window, one selected readout per site, and updated local means. The present study tests transfer between datasets with all retained readouts and fixed local means.