Mean ranks gave 25.2% patient mean onset label recall with 10% contact budgets in 36 patients with two SEEG records

We tested contact selection in 36 patients from public OpenNeuro ds004100, version 1.1.3. Each patient had two 120 second SEEG records. All rules had the same 10 percent contact quota, with integer rounding. Patient mean onset label recall was 20.5 percent for one record. It was 25.2 percent for mean ranks and 22.6 percent for minimum ranks. The mean rank gain was 4.68 percentage points. Its unadjusted 95 percent bootstrap interval was 0.35 to 9.35 points. Neither planned comparison passed Holm correction. The corrected p values were 0.100 and 0.194. This does not prove a zero effect. An exploratory equal time check used 112 seconds per list. Across record mean ranks gained 2.35 points over within record pairs. The interval was -0.22 to 5.31 points. The package supplies a fixed budget benchmark, code, contact data, every list, figures, random draws, and pinned source instructions. All 72 source prefixes and band powers passed checks. A second run reproduced all 22 numerical files byte for byte. The study tests source clinical labels. It does not test treatment outcomes or external prediction.

stereoelectroencephalography · epilepsy · seizure onset zone · contact selection · rank aggregation · fixed contact budget · patient mean recall · low frequency power