A 7 Hz phase input increased mean forecast error for later log response size after five calibration pulses at 86 SEEG readouts in eight patients

The added phase input increased mean later squared log RMS error by 6.68 percent in this test. The models used five early pulses at each of 86 fixed readouts in eight patients. They then forecast response size for five later pulses. The target was natural log RMS voltage from 20 to 500 ms after each pulse. A fixed 7 Hz sinusoid fitted from -500 to -20 ms supplied the added phase input. The ridge penalty was 5. Mean test error was 0.065345 for past power and 0.069712 for past power plus phase. The paired gain was -0.004368; its descriptive 95 percent person bootstrap interval was [-0.008256, -0.000802]. The exact two-sided sign-flip p value was 0.054688. The descriptive interval excludes zero for this sample and this comparison. The exact p value is above 0.05. This test does not support added forecast value for the primary phase input. The package adds verified past voltage, fitted models, event maps, predictions, and three figures. The result applies to this small linear model and this fixed clinical sample. It does not establish neural causation or live interface performance.

SEEG · phase estimation · evoked response · forecasting · ridge regression · brain-computer interface · electrical stimulation · OpenNeuro