Common average and bipolar scores selected different contacts in 15 of 16 SEEG conditions from eight people

We tested whether a recording reference changes contact selection for an interface that reads adjacent-contact differences. We used public SEEG from eight STReEF participants. The fixed sample had two stimulation conditions per person and 973 condition-anchor records. Each calibration used five pulses in a 20 to 500 ms window. Common average and bipolar scores selected different sets of four anchors in 15 of 16 conditions. Mean overlap was 48.4%. All later forecasts used bipolar voltage at the selected anchors. Mean skill was 0.098 after common average selection and 0.081 after bipolar selection. The paired difference was -0.017, with an exploratory 95% interval from -0.236 to 0.158. These data show that contact choice depends on the calibration reference. They do not establish a better forecast method. The package preserves compact signal inputs, code, source version pins, hashes, three figures, and complete reproduction instructions.

stereoelectroencephalography · SEEG · common average reference · bipolar reference · contact selection · forecast skill · electrical stimulation