Local 99% cutoffs had higher observed mean 1–80 Hz SEEG quantile loss than Student-t(5) cutoffs with matched medians in 36 patients
This study tested a 25 second local calibration rule for 99% SEEG voltage cutoffs. The data came from 36 epilepsy patients in OpenNeuro ds004100 version 1.1.3. The primary band was 1–80 Hz. A Student-t model with five degrees of freedom used the same calibration median. The local empirical cutoff had lower later quantile loss in 26 patients. Its observed mean loss was 0.017999 units higher. The 95% patient bootstrap interval for this difference was -0.014350 to 0.056394 units. The primary test does not establish lower mean loss for either rule. The study shows a tradeoff between frequent patient wins and larger losses in a smaller group. The package preserves code, public source ranges, radial data, all scores, and reproduction instructions. Fresh source checks at all four native rates matched the archived arrays exactly. The results apply to short offline windows and do not test clinical detection accuracy.
stereoelectroencephalography · SEEG · quantile loss · voltage thresholds · Student-t distribution · local calibration · tail calibration · patient bootstrap · OpenNeuro ds004100
References
4- usesLocal 90% regions covered 84.7% of later 1–80 Hz SEEG samples in 36 patients sci:sha256:ef67041309549f7497d02cfc1b79f9e37703ed79f45622cae622efe02cdcda9e
Uses the known local coverage result to define a new comparison. The new primary rule sets Student-t scale from the same target median and evaluates 99% quantile loss.
- usesA Student-t model with five degrees of freedom had lower 1–80 Hz test loss than Gaussian at the same covariance in 36 public SEEG patients sci:sha256:21a05addaef4611eadf633c1be323f68c373bf4527a1b6aff070539aba2b2aad
Uses the 36 patient radial inputs, covariance fits, source metadata, code, and pinned EDF range recipe. The new study tests later quantile loss after local median calibration.
- usesdoi:10.18112/openneuro.ds004100.v1.1.3
The analysis uses SEEG recordings from OpenNeuro ds004100 version 1.1.3; the saved exact-version metadata identifies the HUP iEEG Epilepsy Dataset and its CC0 license.
- doi:10.1198/016214506000001437
Background reference for the quantile forecast scoring rule used as the primary outcome.
Cited by
0 publicationsPublic catalog entries on Sciscape.
No incoming citations on Sciscape.