Three consecutive scores gave no clear pre-response detection gain in eight SEEG streams at an early target of one command per eligible minute
We tested one score and three consecutive scores in eight fixed public SEEG streams from OpenNeuro ds004473 version 1.0.2. Each rule had its own threshold. Both thresholds met an early target of one outside-guard command per eligible minute. Later evaluation covered 55.93 sampled minutes and 425 recorded responses. Mean pre-response detection was 8.37% and 8.78%. The paired change was 0.41 percentage points, with a descriptive 95% patient interval from -3.67 to 4.40. Later rates were 0.60 and 0.80 commands per eligible minute. Four three-score replays exceeded the early target. A wider response interval gave higher detection with three scores, but allowed later commands. The package preserves a finite calibration proof, complete compact inputs, all commands and hits, source recipes, code, and figures. These reused task data do not establish intended command detection or live BCI control.
brain-computer interface · stereoelectroencephalography · command detection · rate calibration · consecutive scores · offline replay
References
5- usesA fixed SEEG decoder separated windows before keypress from rest on later trials in eight patients sci:sha256:bd6c59023a95ca7dc23d6549123fb8e0b3724d75412db49c1e004ce64fc1dd57
This test uses its event features, fixed early decoder design, and event range extraction code through the source publication chain.
- usesA higher score threshold changed exact command times in eight fixed SEEG streams with a one-second waiting period sci:sha256:f3d4cb75d8fa2ee27205d550a7dff8e125e5fafcc430399c663e4072ed9a50fc
This test uses its preserved feature arrays, fixed models, source scores, and provenance. It also uses the proof that an earliest-candidate schedule has maximum count under a fixed wait.
- usesAt a zero threshold, eight-pair SEEG decoders gave a mean of 36.7 outside-guard triggers per eligible minute in eight later task scans sci:sha256:22495a59283fcfea65785cbf2a7a410c3a81dc75e721320a08a5ed114e14bd43
This test uses the continuous SEEG features and exact source range recipes from this study through the direct input publication.
- usesdoi:10.18112/openneuro.ds004473.v1.0.2
The analysis uses the eight patients’ SEEG records and derived inputs from OpenNeuro ds004473 version 1.0.2.
- doi:10.1088/1741-2552%2Facae0a
Cited by
5 publicationsPublic catalog entries on Sciscape.
- usesTwo-second interval thresholds reduced background output and detection before keypresses in eight later SEEG records
Defines prior work on threshold fitting for commands per eligible minute and consecutive scores. The new comparison tests any candidate output per monitored interval and its detection cost.
- usesThree consecutive SEEG scores gave no clear mean gain in retrospective detection area at maximum budgets of 0 to 6 background commands per eligible minute in eight patients
This publication supplied all eight SEEG input records, features, models, scores, response labels, and time regions. The new study compares complete threshold frontiers under the same maximum later background budgets. It also reproduces the prior early maximum-count thresholds and 48 later command schedules.
- usesFixed SEEG commands showed no clear 500 ms timing gain over local delay swaps in eight recordings
Used its complete continuous features, original frozen models and scores, source records, and retrieval code. The new analysis tests local response delays. It does not use its later dwell thresholds.
- usesTask background training increased mean detections and background commands in eight later SEEG streams at an early target of one command per eligible minute
We used its continuous command evaluation and maximum eligible-count calibration design. We applied it to two new fitted models, a separate fit/calibration/test split, and a common 500-sample feature history.
- usesA fixed SEEG score correction increased mean test loss for 100 ms key forecasts in eight cued task recordings
We used its continuous power arrays, fixed source scores, source model training inputs, and retrieval code. These files supplied the neural input and its raw source records.