Subtraction of overlap expected from uniform circular shifts lowered mean channel rank agreement for five-minute sleep SEEG maps in eight patients

We tested whether subtraction of expected overlap improves agreement between repeated maps of fast ripples that overlap ripples. We used public sleep SEEG event marks from OpenNeuro ds003498 version 1.0.1. The primary analysis used 152 five-minute records from eight patients with temporal lobe epilepsy. The score was the observed overlap count minus its exact mean under uniform circular timing shifts. Negative scores were retained. Each patient received equal weight. Mean Spearman channel rank agreement fell from 0.7571 to 0.6432. The mean patient difference was -0.1139, with a 95 percent patient bootstrap interval of [-0.1782, -0.0374]. Seven of eight patients had lower agreement. Fixed timing and duration checks gave the same direction. The mean change in agreement of the highest-scoring channel sets was smaller. A diagnostic added after the primary result found higher agreement when negative scores were replaced with zero. The package preserves 655 source files, source hashes, code, plans, all results, figures, and reproduction instructions. Independent checks confirmed the calculations. A second run from a new source retrieval reproduced all main result file hashes. The study reuses a selected cohort. It tests rank repeatability and does not test clinical accuracy.

high frequency oscillations · fast ripples · ripple overlap · intracranial EEG · rank agreement · circular shifts · temporal lobe epilepsy