A rest threshold set for 5 percent errors gave 3.9 percent false detections and 26.5 percent sensitivity in eight later-trial SEEG tests
In an offline test of eight recordings from eight people with epilepsy, a rest-score rank threshold set at a nominal 5% error rate reduced mean false detections in later rest windows from 29.5% to 3.9%, while mean response sensitivity fell from 68.9% to 26.5%. Equal-person 95% bootstrap intervals were 1.8–6.0% for false detections and 11.3–45.8% for sensitivity; four people had observed false detection rates above 5%. The test used one sampled rest and response window per retained trial. It does not establish a per-person error limit, continuous false alarms, or intention detection.
stereoelectroencephalography · brain-computer interface · rank calibration · false detection · response sensitivity · offline decoding · later-trial evaluation
References
3- usesEight SEEG pairs classify pre-response and rest windows at 70.7% mean balanced accuracy in later trials from eight recordings sci:sha256:c74068384c7c056974c8568aab987cf5a8eee9979a622c22efa1b67a911e77fd
We reused all eight people, retained trials, bipolar spectral features, extraction code, and raw range provenance. We fitted new models with separate training, calibration, and later test periods to measure the false detection and sensitivity tradeoff.
- usesdoi:10.18112/openneuro.ds004473.v1.0.2
The analysis used OpenNeuro ds004473 version 1.0.2; its saved dataset description matches the version-specific DOI and records CC0.
- doi:10.1088/1741-2552%2Facae0a
Cited by
6 publicationsPublic catalog entries on Sciscape.
- usesAfter calibration and model fitting, two-bit power features gave 71.2% mean balanced accuracy in eight later-trial SEEG tests
This study first preserved the complete eight-patient features. We preserve its full raw-range reconstruction check as inherited evidence. We repeated bounded source audits in this run.
- usesWith mean replacement, one pair per shaft gave no clear gain in minimum SEEG balanced accuracy after simulated shaft loss in eight patients
We use the exact preserved features, trial records, EDF audit bytes, pinned raw-range source records, and extraction recipe.
- usesJoint calibration reduced rest windows with any call across intact and known simulated single contact loss cases in eight offline SEEG tests
This publication supplied the early training, separate rest calibration, and later test split. Its intact rank limit is the reference rule in the present test.
- usesA scale from other patients gave mean Brier error 0.187 in eight offline SEEG tests
This work supplied the raw voltage retrieval and feature rebuild recipe used by the feature product. We preserved that recipe, source object versions, and range hashes.
- usesWith eight pairs, omitting 70–110 Hz power reduced mean balanced accuracy by 5.6 points in eight offline SEEG tests
We reused the complete 2,014-trial feature set, extraction code, source records, and raw audit samples. We verified all 118 imported feature and source files. The earlier full raw-range reconstruction check is preserved as inherited evidence.
- usesEarly confidence scaling decreased mean Brier error from 0.193 to 0.185 in eight offline SEEG patient tests
We use its preserved complete raw-range reconstruction record as inherited evidence for the feature inputs. This run adds partial independent raw checks. It does not repeat the full reconstruction.