In one resting SEEG patient, same-shaft bipolar band-power correlation fell below the time-shift null at 15–20 mm; bipolar PR was 74–103 versus 5–39 for monopolar contacts

We measured redundancy in resting-state SEEG band-power features for one patient with epilepsy, using 175 s from OpenNeuro ds005670 sub-01. The recorded EDF object is identified by its source URL and ETag; the exact immutable dataset release is uncertain. From 226 retained contacts we formed 203 adjacent-contact bipolar channels. The archived method reports 350 non-overlapping 0.5-s windows; the machine-readable results record 349. Across five bands, median correlation for non-contact-sharing, same-shaft bipolar pairs was 0.13–0.23 at 5–10 mm and 0.05–0.09 at 15–20 mm, below the time-shift null 95th-percentile absolute correlation of about 0.11. Different-shaft pairs at 5–10 mm had median correlations of 0.04–0.12. Bipolar participation ratios were 74–103, compared with 5–39 for 226 monopolar contacts; monopolar values were lowest in theta (5.3), alpha (6.6), and high gamma (9.2). Bipolar participation ratios were 58–81% of the time-shift null values. The analysis was not pre-registered, epileptiform events were not rejected, and the median correlations conceal high-correlation pairs. The one-patient, rest-only study had no behavioral or decoding outcome, so it describes feature redundancy and does not establish decoding value. Participation ratio depends on window count and is interpreted relative to the time-shift null and reference condition.

SEEG · intracranial EEG · band power · feature redundancy · bipolar reference · monopolar reference · participation ratio · brain-computer interface · resting state · electrophysiology