All earlier pulses gave lower mean SEEG forecast error than four recent pulses in five recordings at 20–500 ms
We tested forecasts of recorded voltage after an electrical pulse at a known site. The source was OpenNeuro ds004457, version 1.0.2. The sample had five patients, 171 sites, and 1,124 scored pulses. Each forecast used only earlier complete responses at the same site and current. The test window was 20 to 500 ms after the pulse. A mean of four recent pulses had 2.50 percent more error than a mean of all earlier pulses. This is the mean of five patient error ratios. The descriptive 95 percent interval for error reduction was minus 6.00 to plus 2.11 percent. Both updating rules had less mean error than a forecast fixed after four pulses. Observed pulse order gave a smaller recent-only error penalty than shuffled order. That conditional test does not show benefit across new patients or identify a neural cause. The package includes three figures, voltage features, all primary forecasts, input event IDs, source hashes, code, and reproduction commands. A fresh raw extraction matched the packaged data, and a second analysis gave identical result files. This exploratory result is limited to accepted-pulse voltage forecasts in the five recordings. It does not establish live BCI performance.
stereoelectroencephalography · electrical stimulation · voltage forecasting · response variability · brain-computer interface · pulse order · forecast error
References
4- usesPooled site sets included the correct site for 93.6 percent of later SEEG pulses within five sessions at 20 to 100 ms sci:sha256:16a0821e6ea5302515dea9c46cffb63ec910cd0b438ee578fb5d0ecad4563dc7
Used its native 5 ms voltage bins, source event and contact metadata, readout choices, and raw retrieval and extraction code. A fresh raw rebuild matched the copied arrays and event files.
- usesdoi:10.18112/openneuro.ds004457.v1.0.2
The analysis used the recorded voltage data in OpenNeuro ds004457, version 1.0.2.
- doi:10.1523/JNEUROSCI.1325-22.2023
- doi:10.1111/epi.17528
Cited by
9 publicationsPublic catalog entries on Sciscape.
- usesA matrix forecast with training on other shafts raised mean relative error after one accepted pulse in five SEEG recordings at 20–500 ms
We used its 5 ms SEEG voltage bins, source metadata, raw file manifest, and MIT extraction and retrieval code. We adapted the loader. These inputs enabled the new time-separated matrix-versus-scalar forecast test.
- usesRetained pulse order showed no clear extra gain for mean SEEG forecasts at 20–500 ms with variance scaling in eight sessions
Used prior growing-versus-recent and pulse-order results in another SEEG dataset to motivate the exact finite-set averaging reference.
- usesPre-pulse ridge updates had lower mean relative forecast error than early updates in five SEEG replays with earlier fits from other shafts
Its earlier-pulse history result motivated the mean of all earlier local responses as the reference forecast. The present test adds other-shaft readouts and a 100–500 ms target.
- usesResponses 1 and 4 gave 3.4% higher mean SEEG forecast error than responses 2 and 3 in five sessions at 20–500 ms
Uses the earlier count-versus-recency comparison to define a separate fixed-count question. Both new estimates use two responses. This citation supplies study context and no new data or score.
- usesA last-pulse correction increased SEEG forecast error over the past mean in three of five sessions at 20–100 ms
Its comparison of all earlier responses with four recent responses defined the main forecast reference. We test a fitted last-response residual instead of discarding older responses. The patients overlap. Our final-four target set and eligibility threshold differ, so this is not an exact repetition.
- usesMeans of four pulses reduced normalized 20–100 ms voltage error by 33.3% relative to zero across blocks at five sites in three SEEG sessions with equal site and patient weights
This prior work tested fixed and updated mean response forecasts with zero and other-site controls over a pulse sequence. We use that existing mean-forecast approach. We supply a narrower source-block test at five supported repeated sites, with four later pulses per site and a primary 20–100 ms window. We report raw losses, normalized losses, weighting sensitivity, and exact archived vectors. We reused no code or arrays from this publication.
- usesFour signed pre-pulse modes per pair raised mean relative 20–100 ms SEEG forecast error by 19.2% at 94 later sites in five sessions with ridge strength 0.1
Used its 5 ms voltage bins, public contact and event metadata, fixed recording targets, and pinned source retrieval recipe. Fresh native extraction matched the compact inputs exactly.
- usesA SEEG library from four earlier pulses per known site identified three of six first return pulses at 20–100 ms in three sessions
Its earlier-pulse forecast test informed the strict rule that a response must end before a later query pulse begins. Our test identifies a known site among the available candidates.
- usesBackground voltage prediction reduced mean centered SEEG error at 20–100 ms by 3.89 percent compared with four-pulse templates at 94 later sites in five sessions
Used its 5 ms voltage benchmark, fixed recording contacts, and pinned source subset through the immediate source. These inputs define the tested recording pairs. All five raw rebuilds matched the preserved voltage arrays and event tables.