A 95 percent bootstrap interval included zero gain from joint shaft sampling for later STE HFO mark counts in 18 SEEG sleep records under an independent, exchangeable patient model

Joint shaft sampling did not show a clear mean gain for later total STE HFO mark counts in 18 public SEEG sleep records. The first 30 minutes supplied the fit counts. The next 30 minutes supplied the test counts. The joint and independent shaft rules used the same marginal laws and the same total mean. The mean normalized CRPS gain was -0.002659. Its 95 percent patient bootstrap interval was [-0.030999, 0.016488], assuming independent, exchangeable patients. Joint intervals had higher observed coverage and greater width. Their actual model probability was also higher. All planned checks are reported. The reverse split favored joint sampling, but all forward check intervals included zero. The result concerns released detector marks and this fixed split. It does not establish clinical benefit. The package preserves public inputs, code, probability laws, scores, bootstrap draws, figures, and reproduction instructions. Independent calculations passed. A fresh-source repeat gave identical core output hashes. This package update fixes the integrity check for metadata added by AutoSci retrieval.

SEEG · high-frequency oscillations · event counts · probabilistic forecasting · cross-shaft dependence · continuous ranked probability score