Endpoint tests missed seven decision changes between simulated contact gains 0.8 and 1.2 in eight SEEG records
We tested whether endpoint gain checks can show stable SEEG decoder decisions throughout a gain interval. We used public data from eight people and fixed pair power ratio models. The test had 1348 later windows. One recorded contact had a constant simulated gain before digital subtraction. At gains 0.8 and 1.2, endpoint checks accepted 978 windows. Seven of these windows changed decision at a gain between the endpoints. Fresh source bytes confirmed all seven. A conservative bound with 256 gain segments retained 70.06% of windows as an equal person mean. It retained 50.35% of all windows as correct decisions. This was 12.40 percentage points above a one-segment bound. The descriptive 95% person bootstrap interval was 9.60 to 15.85 points. The gate also retained wrong decisions. It bounds stability under the specified gain model, not correctness. The package preserves code, fixed models, data, source hashes, voltage checks, window bounds, failure witnesses, and three figures. This offline test concerns event-defined task states in one session per person. It does not establish online intent or stimulation performance.
stereoelectroencephalography · brain-computer interfaces · power ratio decoder · contact gain · decision stability · score bounds · signal processing
References
6- usesPair power ratios changed decoder decisions under simulated 5% contact gain errors in eight SEEG recordings sci:sha256:711f73013f81f5139746b7aa504899871ca780d80e2498c3b47247c0f5ec93a3
We use its exact contact spectral moments, fixed pair power ratio models, chronological split, candidate features, raw audits, and source hashes. We extend selected gain tests to continuous gain bounds and interior failure witnesses.
- usesBand power ratios retain 67.0% balanced accuracy under simulated constant voltage gains per shaft in eight SEEG recordings sci:sha256:fa4b2be456049aafc97572ea0a4d44239abb9b5dc00341d8bcb7a0fd368bf201
Its pair power ratio transform defines the inherited decoder input. We test its stability under unequal gain at one contact before digital subtraction.
- 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 use its original OpenNeuro feature extraction, retained trial rules, event windows, source metadata, and pinned raw range records, as preserved in the contact gain package.
- usesTraining for ±6 dB gain on one shaft raised correct classes with a guarantee in eight SEEG records under a linear gain model sci:sha256:5b0d343335c2da00c225c25a050de434c99e47b5020cc51baebb2b35ef408f94
We use its shaft gain bound as prior proof scope. This study gives a separate bound for nonlinear band power changes at one recorded contact.
- usesdoi:10.18112/openneuro.ds004473.v1.0.2
This is the exact OpenNeuro dataset release used in the analysis; its saved metadata records CC0.
- doi:10.1088/1741-2552%2Facae0a
Cited by
1 publicationPublic catalog entries on Sciscape.
- usesEndpoint tests missed the largest accepted error in 19 of 35 shaft cases for eight fixed SEEG readers under ideal constant gains within ±6 dB
This prior study showed that endpoint tests can miss decisions under unequal contact gains. We use it as intellectual context. Our new audit concerns the largest accepted-error ratio under uniform affine shaft gain; it does not reuse its code or claim an independent discovery of endpoint failure.