Quadratic bounds certified 39.4 percent of later SEEG decisions for simultaneous constant contact gain changes of up to five percent in eight recordings
We developed a bound for a fixed SEEG reader when all contact gains can change at the same time. Each gain stays constant within a 500 ms window and lies between 0.95 and 1.05. The bound uses bipolar power, common power, and real cross power. It finds exact power extrema for each pair and band, then forms a conservative linear score interval. We reused public OpenNeuro ds004473 v1.0.2 data and early-trained models from eight people with epilepsy. The primary equal-person mean was 39.4 percent certified later decisions; its descriptive 95 percent person bootstrap interval was 28.3 to 52.0 percent. The simpler norm bound certified 4.1 percent. Pooled counts were 550 certified windows out of 1348. A search that used no task labels found allowed gain vectors that changed 513 decisions. No certified decision changed. A 60-digit outward-rounded interval check confirmed all primary flags for the archived numeric inputs. All retained spectral inputs were rebuilt from pinned public byte ranges and matched exactly. Stable decisions can still be wrong. The study tested stored digital gains and offline task labels; it did not test actual amplifier faults, live brain control, or stimulation. The package preserves the proof, code, input data, source hashes, model values, per-window bounds, gain vectors, and figures.
stereoelectroencephalography · brain-computer interface · signal processing · decision stability · contact gain · quadratic bounds
References
6- usesPair power ratios changed decoder decisions under simulated 5% contact gain errors in eight SEEG recordings sci:sha256:711f73013f81f5139746b7aa504899871ca780d80e2498c3b47247c0f5ec93a3
We reused the compact bipolar/common/cross spectral moments, all candidate features, fixed split, source recipes, and parent models. We derived and tested bounds for simultaneous changes at all contacts.
- usesJoint score bounds gave 96.3% mean trial coverage and 34.5% mean output yield in eight later SEEG tests with known simulated loss of one contact sci:sha256:4a1c02b9cd875b5cc70d2dbb57771096a5247dbab564b70b51b8fea2e6e55e83
Its prediction sets under known contact loss informed the distinction between label coverage and decision stability. No fitted output was used.
- 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 its eight-pair pre-response reader design and chronological split through the direct parent. Our refit preserved its recorded later decisions and accuracy.
- 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
Its common shaft gain certificates informed the distinction between additive log-power uncertainty and unequal contact gains before bipolar subtraction.
- usesdoi:10.18112/openneuro.ds004473.v1.0.2
The analyzed data are from OpenNeuro ds004473 v1.0.2; the publication records the pinned source version and its URL.
- doi:10.1088/1741-2552%2Facae0a
Cited by
3 publicationsPublic catalog entries on Sciscape.
- usesSeparate allowed 12-sample errors at the window center changed all 1,348 tested SEEG decisions within a 100 µV RMS limit per pair in eight people
This work supplied method context for quadratic power bounds and fixed log power reader intervals under gain errors. We applied a score interval to short additive sample errors. We did not use its gain-error results as data.
- usesSimulated joint contact gain errors within five percent changed 172 SEEG decisions that passed separate contact tests in five offline sessions
This study established prior work on simultaneous constant contact gains before bipolar subtraction. It uses nonlinear power features and conservative bounds on ds004473. Our fixed affine site reader on ds004457 permits exact separate and joint gain radii.
- usesShared contact cells certified 58.6 percent of later SEEG decisions for constant contact gains within five percent in eight recordings
Uses its exact spectral moments, candidate features, frozen effective model values, chronological splits, whole-box certificates, exact allowed change witnesses, and public source range recipes. The new bounds enforce shared contact cells.