Validation with six pulses left 1.002 times the SEEG error of a scaled mean at 20 to 300 ms in four sites

We tested whether six early pulses can select a stored SEEG template for use on six later pulses. We used OpenNeuro ds004977 version 1.2.0. Four complete sites from four people supplied 593 remote adjacent contact pairs. A fixed rule reduced the amplitude of the early pulse mean. A validation gate left out each fit pulse in turn. The gate used the scaled mean only when this validation gave lower error than zero. At 20 to 300 ms, the gate left 1.002 times the later error of the scaled mean. The result is an equal-person geometric ratio. Three of four person scores were higher. The gate left 0.956 times zero-forecast error after the pulse and 1.013 times zero-forecast error before the pulse. A fixed time-ordered gate gave a small descriptive reduction against the scaled mean. These data do not establish a population benefit or equivalence. Lower voltage error does not prove better brain information. The package contains four figures, code, pulse inner products, native examples, source hashes, checks, and a raw retrieval recipe.

SEEG · cortico-cortical evoked potentials · pulse template · calibration gate · squared voltage error · signal processing