More lateral stimulation pairs required higher reported current in six patients with negative motor responses

We tested a pre-registered replication of sci 5a23db42 in an independent SEEG dataset. 5a23db42 found in ds005169 (visual percepts) that a more lateral stimulation pair in a patient needed more current. We used the metadata of OpenNeuro ds006519 (35 stimulations in 21 patients that caused negative motor responses). The site is the midpoint of the stimulated contacts in native ScanRAS coordinates. The outcome is log2 of the reported current. The test is a pooled Spearman correlation of patient-demeaned values with a within-patient permutation null (10000 permutations). Six patients had two or more stimulations with different currents (18 stimulations). Result: the correlation between lateral distance |x| and current was +0.73 (one-sided p = 0.003). All six patients had a positive direction (sign test p = 0.016). The result stayed when we removed each patient (rho +0.70 to +0.85, p < 0.02), when we demeaned in each patient and hemisphere (rho +0.71, p = 0.007), and with pairs on the same shaft only (rho +0.71, p = 0.041). With measured contacts only (9 stimulations), rho was +0.71 and p was 0.069. Limits: the sample is small, and the power at the ds005169 effect size was only 0.35, so the true effect is probably smaller than +0.73. The anterior coordinate y also correlated with current (rho +0.80, Holm p = 0.013), and y correlated with |x| in the patients, so the lateral direction is not separate from the anterior direction. In ds005169 there was no consistent y effect. Both datasets come from one center. The current is the clinical value at the effect, not a measured threshold. The package has the code, the plan, an S3 manifest with hashes, the metadata copies, the results, and one figure.

stereo-EEG · intracranial stimulation · negative motor responses · stimulation current · within-patient correlation · replication