A one second count cap gave no mean list agreement gain for overlapping HFO marks in seven SEEG patients

We tested a limit of one event mark per bipolar pair per second. The test used public SEEG annotations from OpenNeuro ds003498 version 1.1.1. There were 135 distinct 300 second sleep records from seven patients. The main marks were ripples that occur with fast ripples. Each source record gave a usual count list and a capped count list. Both lists were compared with the same usual count list from a different record. Lists had three of 15 pairs or five of 24 pairs. Cutoff ties had equal selection probabilities. Each patient had the same weight. Mean agreement was 70.59% with usual counts and 70.46% with capped counts. The gain was -0.13 percentage point. The descriptive 95% patient bootstrap interval was -0.76 to +0.34 point. The two sided patient sign flip probability was 0.8125. The cap reduced counted marks by 10.12% on average across patients. Three patients had a gain and four had a loss. The result shows no mean gain in this sample. It does not show equivalence or clinical accuracy. The package includes all fixed sensitivity tests, source files, source hashes, code, output tables, figures, and offline reproduction steps. All ten output files were byte identical in a new offline run. All 201 fresh small source downloads matched saved hashes. One full waveform passed a separate checksum and format check.

high frequency oscillations · stereoelectroencephalography · event count cap · list agreement · epilepsy