A 3 dB alpha peak rule gave patient mean reference and time disagreement of 9.33 and 12.24 percent in 19 SEEG records

We tested whether reference change gave more alpha peak label disagreement than change to the next minute. We used all 19 patients and 2,070 contacts from a pinned public SEEG sample. The fixed rule required a local 8 to 13 Hz maximum at least 3 dB above a linear background. Patient mean disagreement was 9.33 percent across references and 12.24 percent across minutes. Mean reference minus time disagreement was -2.91 percentage points. The patient bootstrap 95 percent interval was -6.93 to -0.075 points. The planned exact sign flip test gave p = 0.0858 and did not reject zero at the 5 percent level. That test requires independent patient contrasts and sign symmetry under the null. A fixed quadratic background check reversed the mean direction. An added influence audit showed a large contribution from one patient. The evidence does not establish a stable general order of the two changes. The result concerns the stated label rule and short interictal records from patients with epilepsy. It does not identify the best reference or confirm neural oscillations. This is a new analysis of an existing cohort. We rebuilt all 19 spectrum files from pinned public signal ranges. All 1,912,680 density values matched exactly. Fresh processes also gave exact main result file matches. The package preserves the code, spectra, source receipts, metadata, results, figures, local plan, and repeat instructions.

stereo-electroencephalography · SEEG · alpha peak · spectral analysis · reference choice · time stability · binary labels · epilepsy