Four-pulse fitting reduced error but did not show smaller joint envelopes for 20–500 ms SEEG in eight sessions with five early pulses
We tested a five-pulse calibration budget for forecasts of stimulation responses. The public data were SEEG from eight epilepsy patients in OpenNeuro ds004696, version 1.0.1. We used 86 sites, 886 stimulation-to-readout pairs, and 430 later test pulses. Pulse 1 fixed the noise scales. The split rule used pulse 1 as its forecast. The full rank rule fitted pulses 2 to 5. Both rules had an 80 percent marginal target under conditional exchangeability. We proved the smallest fixed ball about the four-pulse mean that contains the full rank set. Its radius is 5/3 times the largest early residual in the fixed joint norm. At 20 to 500 ms, its geometric mean radius changed by -0.3 [-11.0, 10.2] percent relative to the split rule. Brackets give 95 percent patient bootstrap intervals. This result did not establish equal radii. Split-ball complete-pulse coverage was 77.3 [69.8, 83.5] percent. Exact full rank coverage was 79.8 [71.8, 86.0] percent. Full enclosing-ball coverage was 89.0 [84.1, 92.9] percent. Four-pulse fitting reduced the scaled mean squared forecast error from 8.64 to 4.43. The exact set and its ball must be assessed separately. Time-ordered pulses do not prove exchangeability. The study did not test a live BCI or a new session. The package contains exact trial products, source hashes, code, proof, figures, and reproduction instructions.
stereoelectroencephalography · brain stimulation · conformal prediction · joint coverage · calibration · forecasting · uncertainty quantification · neural responses
References
6- usesFive-pulse noise scaling reduced later single-pulse error for 20–500 ms centered bipolar SEEG in eight patient sessions sci:sha256:fbc06ab469d48eac7a6f54dd97cd24ccd9c7c5ca4b0db4d64340536b3e81428a
This publication supplied exact waveform products, corrected chronological event maps, source hashes, and retrieval code. Fresh source rebuilds matched these inputs.
- usesSeparate pair bounds covered 71.5 percent of individual SEEG responses and 17.0 percent of whole pulses at 20 to 500 ms in eight sessions sci:sha256:1a0ecb0d254415256c85ec04c2b9bce25a0f6893277ff47091e5d6c673436869
This publication supplied the complete-pulse split-rank construction. We changed its fitting and calibration allocation to compare five early pulses at an 80 percent target.
- usesA five-pulse check selected more repeatable bipolar SEEG pairs within eight epilepsy patient sessions sci:sha256:b5b7963746f5c6d2a2a8f7844ec590732a4eb152bc7a70979b3bc95c3e27820f
This publication defined the bipolar readout rules and the original compact extraction. We kept those rules for the fixed-budget test.
- usesA joint rank check accepted 83.2 percent of later SEEG responses at 20 to 500 ms in eight patients after five calibration pulses sci:sha256:293b34e71ea8d2f99a69e38b3c7c5227f6549f67906ebb9986851394508c542b
This publication supplied the joint augmented-mean rank construction. We fixed the first-pulse scales and derived its smallest enclosing radius about the four-pulse mean.
- usesdoi:10.18112/openneuro.ds004696.v1.0.1
The study uses source SEEG derivatives from OpenNeuro ds004696 version 1.0.1.
- doi:10.1523/JNEUROSCI.2201-22.2023
Cited by
1 publicationPublic catalog entries on Sciscape.
- usesFitting on 20 of 29 stored pulses reduced mean relative SEEG band width by 20.6 percent in six people
This prior five-pulse study provides method context for the tradeoff between fitting a mean and forming a joint response set. It uses another dataset, full rank sets, and an 80 percent target. The new test uses two disjoint splits at a 90 percent target in ds006065.