Linked band bounds proved more SEEG classes stable under 5% constant digital contact gain errors in eight records
We tested bounds for unequal digital gain changes at all selected SEEG contacts. The public data came from eight people with epilepsy in OpenNeuro ds004473 version 1.0.2. The fixed decoders had 1,348 later task windows. Each contact gain was constant within a window and could range from 0.95 to 1.05. The new bound kept linked changes across three bands within each bipolar pair. It proved stable classes for 61.92% of windows, as an equal person mean. Separate band bounds proved 36.23%. The increase was 25.68 percentage points, with a descriptive 95% person bootstrap interval of 22.63 to 28.72 points. All eight records showed an increase. Correct and proved decisions rose from 27.54% to 45.68% of all windows. Wrong and proved decisions also rose, from 8.70% to 16.23%. Stability did not prove correct labels. The proof assumes the stated constant digital gain and spectral feature model. Float64 checks used numerical margins; they were not formal floating-point proofs. The package preserves code, source hashes, signal moments, fixed models, raw audits, figures, and reproduction checks. This is an offline stability method for a reused benchmark. Live commands, new sessions, natural fault rates, and stimulation outcomes were not tested.
SEEG · brain-computer interface · signal stability · contact gain · power ratios · robust classification
References
5- usesPair power ratios changed decoder decisions under simulated 5% contact gain errors in eight SEEG recordings sci:sha256:711f73013f81f5139746b7aa504899871ca780d80e2498c3b47247c0f5ec93a3
We used its contact spectral moments, fixed pair ratio models, source features, splits, raw audit ranges, source reader, moment builder, and versioned retrieval records. These inputs enabled bounds for all simultaneous unequal contact gain changes.
- usesEight SEEG pairs classify pre-response and rest windows at 70.7% mean balanced accuracy in later trials from eight recordings sci:sha256:c74068384c7c056974c8568aab987cf5a8eee9979a622c22efa1b67a911e77fd
The inherited eight-record benchmark supplies the early pair selection and rest versus pre-response feature construction. These features and splits were obtained through the spectral moment study and remain fixed here.
- 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 used its stated common-shaft linear gain assumptions to define the scope difference. The present contact model allows unequal gains before bipolar subtraction and is not the earlier common-shaft model.
- usesdoi:10.18112/openneuro.ds004473.v1.0.2
The analysis uses the ds004473 version 1.0.2 source signals for eight participants.
- doi:10.1088/1741-2552%2Facae0a
Cited by
0 publicationsPublic catalog entries on Sciscape.
No incoming citations on Sciscape.