Ridge forecasts from prior micro-SEEG voltage reduced template error but did not beat zero at 20 to 150 ms in one pig

We tested whether voltage before a pulse could improve a forecast of later micro-SEEG voltage. We used OpenNeuro ds004819 version 1.0.0 from one pig under spinal stimulation. The test used 36 fixed contact differences, eight current blocks, 80 calibration pulses, and 79 later test pulses. For the 20 to 150 ms target, a ridge map from prior voltage reduced mean normalized template error from 1.0608 to 1.0160. It reduced template error in all eight blocks. A zero forecast scored 1.0000. The map did not beat zero on mean error. A timing control supported dependence on pulse-specific recorded state. Baseline subtraction changed the comparator and pair weights even when forecast residuals stayed the same. The package gives exact epochs, code, model coefficients, four figures, source hashes, and retrieval instructions. A fresh raw check and independent score checks passed. The data do not isolate a neural cause or establish a benefit in a human BCI. Calibration pools all blocks; this is an offline test at known currents.

micro-SEEG · intracranial EEG · voltage forecast · ridge regression · electrical stimulation · pig · baseline subtraction · OpenNeuro