A fixed half average reduced observed mean MNI interval score by 1.16 percent in 18 SEEG sleep records

We tested a fixed half average of channel and patient dispersion estimates for later SEEG event counts. The test used released MNI marks from OpenNeuro ds005398 version 1.1.3. The first 30 minutes set the fits and selected 361 channels in 18 patients. The next 30 minutes supplied one-minute counts. All models had the same predicted mean. The mean normalized 90 percent interval score decreased from 4.0237 to 3.9771, a 1.16 percent reduction. Twelve patients had lower scores. The mean paired gain was 0.0466. Its 95 percent patient bootstrap limits were -0.0019 and 0.1064, which include zero. The secondary gain against the pooled estimate was 0.1457, with limits 0.0261 and 0.3211. STE and source-filter checks gave negative mean gains against channel estimates. These checks are descriptive. The primary evidence does not establish a mean gain for new patients. This study reuses a known cohort and benchmark. The package includes exact source bytes, version pins, code, fits, every primary prediction, patient effects, saved bootstrap samples, figures, and reproduction instructions. Fresh retrieval matched 94 inputs. Independent checks and a full offline rebuild passed. The test concerns released marks in the first hour of sleep records. It does not assess neural cause or clinical benefit.

SEEG · high frequency oscillations · event counts · prediction intervals · dispersion · negative binomial · bootstrap · OpenNeuro