Five second phase references reduced mean absolute log error for SEEG maxima in nine selected adults

Five second phase references had less mean error than 50 second phase references in the tested SEEG windows. The test used 36 windows from nine selected adults in OpenNeuro ds003688 version 1.0.7. The signals used adjacent contact differences and a 0.5 to 20 Hz band. Mean absolute log error fell from 0.609 to 0.402, a 33.9 percent reduction. The mean patient gain was 0.206, with a descriptive 95 percent patient bootstrap interval of 0.102 to 0.307. Eight adults had positive gains. The exact two-sided patient sign-flip p value was 0.01172. This test assumes independent patient effects and joint sign symmetry. The local reference also failed to reproduce many large maxima. This conditional comparison does not identify a neural cause or test prediction on new data. The package includes all compact inputs, source records, code, reference draws, and reproduction instructions.

stereoelectroencephalography · SEEG · phase surrogates · voltage maxima · signal processing · patient bootstrap