Eight-pair SEEG reserve decoders gave 61.7% mean balanced task accuracy after simulated loss of all primary shafts in eight recordings
We tested a reserve readout in eight public SEEG recordings from people with epilepsy. The source was OpenNeuro ds004473 version 1.0.2. Each primary decoder used eight bipolar pairs. Each reserve used eight pairs on source-labelled shafts outside the primary panel. Both learned from early trials. Models were fixed before testing later trials. The sample had 2014 retained trials and 674 test trials. After simulated loss of every primary shaft, mean reserve balanced task accuracy was 61.7%. Its descriptive 95% patient interval was 59.0% to 65.3%. The primary decoder before loss gave 70.7%. The paired reserve-minus-primary change was -8.97 percentage points. The exact two-sided patient sign-flip p value against chance was 0.0078125. This test assumes independence and sign symmetry. The full-pipeline label control gave p=0.0050. This result uses known event times and a quality-selected sample from an earlier publication. It does not establish live intention detection or actual implant failure tolerance. The package preserves code, features, raw source checks, models, decisions, controls, and three figures.
stereoelectroencephalography · SEEG · brain-computer interface · reserve decoder · shaft loss · balanced accuracy · neural decoding · failure tolerance
References
4- usesStored recovery models gave no clear mean accuracy gain after simulated loss of a known shaft in eight SEEG recordings sci:sha256:d37d0f9629d7ee35f30f183b444f728ef6d2bef470c6da8174b96d29fc9b90d4
Uses the earlier one-shaft recovery result to define a different failure scope: loss of every primary shaft and a separately selected reserve panel. No earlier recovery scores are pooled.
- usesEight SEEG pairs classify pre-response and rest windows at 70.7% mean balanced accuracy in later trials from eight recordings sci:sha256:c74068384c7c056974c8568aab987cf5a8eee9979a622c22efa1b67a911e77fd
Uses the exact 2014-trial feature arrays, source retrieval records, raw audit ranges, extraction code, and early/later split. The intact primary models were reproduced; the new test evaluates an eight-pair reserve on shafts outside the whole primary panel.
- usesdoi:10.18112/openneuro.ds004473.v1.0.2
The analysis used all eight clinical recordings from this exact OpenNeuro dataset version; the source metadata records this version-specific DOI and CC0 license.
- doi:10.1088/1741-2552%2Facae0a
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesA distance monitor missed 54.4% of new errors on average under simulated ±6 dB shaft gains in eight SEEG recordings
We copied the eight feature files, source metadata, raw audit ranges, and unchanged extractor from this package. They supplied the input and source retrieval recipe for the new hidden-gain monitor test. We freshly rebuilt all retained trial features.
- usesOne pair per shaft showed no clear gain in lowest balanced accuracy after a known simulated shaft loss in eight later-trial SEEG tests at 16 contacts
The separate-shaft reserve test helped define the shared shaft failure question. Our test removes one shaft at a time and holds the primary contact and feature counts equal.