Gains from other sites changed mean normalized 2 mA SEEG forecast error by less than 1 percent in three people at 20–500 ms

We tested whether responses at other stimulation sites can reduce calibration at a target site. The public STReEF sample had three people, 31 physical sites, and 54 ordered conditions. Five 1 mA pulses formed each target template. A fixed scale used products of distinct pulses to limit pulse noise. Three 2 mA fit pulses at other physical sites set a shared gain within each person. Both contact orders at the target site stayed out of that fit. Two separate 2 mA pulses scored the forecast. The score used centered SEEG voltage at 20 to 500 ms. Each physical site had equal weight within a person. Against the low-only scale, error changes were +0.39%, -0.80%, and -0.01%. The mean person change was -0.14%. These small, mixed changes do not establish a useful gain or equal performance. The fixed stable-response subset had +4.57% mean error change. Three local 2 mA fit pulses gave lower error in all three people. Only 6 of 54 ordered conditions had all donor fit pulses before their test. Current changed with time. Earlier studies used the same test pulses. This is an offline test of recorded voltage, not a causal current law or a live BCI test. The package has code, exact inputs, fit states, all scores, source pins, figures, and repeat steps. All 540 fresh source ranges and all 270 saved arrays matched.

stereoelectroencephalography · electrical stimulation · forecasting · pulse variability · calibration transfer · brain-computer interfaces