NREM spindles rarely co-occur across SEEG pairs that share no contact

Question: when a detector finds a sleep spindle at one SEEG bipolar pair (site A), how often does another pair (site B) also have a spindle? This is important for spindle-locked closed-loop stimulation that detects at one site and stimulates at another. Data: OpenNeuro ds003498, 9 patients, 5 min of NREM sleep each, 462 adjacent bipolar pairs on 8-contact depth electrodes. Method: the offline 11-16 Hz detector of publication 307cb98b; for each ordered pair, the fraction of A spindles that overlap a B spindle; a null from 200 circular time shifts of the B events; exact sign-flip tests on per-patient medians. Results: the median co-occurrence was 0.36 for pairs that share a contact, 0.22 at 2 steps, 0.17 at 3 steps, 0.11 at 4-6 steps on the same shaft, 0.17 for another shaft in the same hemisphere and 0.09 for the other hemisphere. The shift null was about 0.06. All categories were above the null (p 0.004-0.035). Co-occurrence decreased with separation in 9/9 patients. On average, each spindle reached 16% of the pairs on other shafts (median of patient means; chance 8%); only 0.3% of spindles reached half of them. With a stricter detector (peak > 5x median), the excess at 3 or more steps and between shafts was approximately zero. Limits: no electrode coordinates (contact steps, not mm), 5 min per patient, epilepsy patients, offline detection, and the result depends on the detection threshold. Package: code, input SHA-256 values, per-pair and per-spindle tables, figure, sensitivity run, and a byte-identical re-run check.

sleep spindle · NREM sleep · SEEG · intracranial EEG · co-occurrence · epilepsy · closed-loop stimulation