Selection for retained template energy gave no clear 100 to 500 ms AUC gain after known simulated shaft loss in seven SEEG sessions at eight contacts

We tested whether a rule that selects recording pairs for retained template energy improved later SEEG response reading after one known simulated shaft loss. The source was public OpenNeuro ds004696 version 1.0.1. Seven epilepsy sessions supplied 79 stimulation sites and 395 separate test pulses. Each panel used four pairs and eight distinct contacts. Five early pulses set pair selection, a mean template, and diagonal noise weights. Five later pulses tested each intact and shaft loss case. The primary window was strict 100 < t < 500 ms. With equal patient weight, mean lowest AUC was 0.7825 for total-energy selection and 0.7835 for retained-energy selection. The difference was +0.0010; the descriptive 95 percent patient bootstrap interval was -0.0014 to +0.0035. The exact two-sided patient sign-flip value was 0.50. These data show no clear gain and do not prove equal performance. The release contains all selected panels, scores, loss cases, figures, a buffered reader, and a source retrieval recipe. A fresh source replay checked one patient. The loss was simulated. The AUC separates pulse-aligned voltage windows and does not prove neural origin or live BCI benefit.

SEEG · electrical stimulation · brain-computer interface · shaft loss · matched-template reader · AUC · OpenNeuro