An early amplitude gate increased prediction skill for the future part of a centered theta label at 50 ms in seven SEEG rest records

We tested three causal amplitude gates in seven public SEEG rest records. A fixed model predicted the future part of a centered theta voltage label at a 50 ms delay. Thresholds used 122 to 238 s. Results used 242 to 578 s. The main gate kept times above the early 75th percentile of combined predicted amplitude. It kept 25.79 percent of later times. Mean future squared-error skill rose from 20.58 to 30.15 percent relative to zero at the same selected times. All seven person means increased. Phase error fell from 56.50 to 38.19 degrees. Mean absolute future MSE rose from 571.37 to 763.38 squared microvolts as target energy rose. The forecast amplitude gate gave 35.71 percent future skill and 47.31 degrees of phase error. Thus, the gates favored different outcomes. In an independent Gaussian voltage control, a known-amplitude gate reduced phase error from 63.12 to 45.57 degrees with zero future prediction skill. Phase agreement alone cannot establish future prediction under this reference. This exploratory result uses one existing rest record per person. It does not test stimulation effects. The package preserves selected public voltage, source provenance, fixed fits, complex outputs, thresholds, tables, three figures, code, a proof, and reproduction instructions. Fresh source hashes and complete bipolar extraction were verified.

stereoelectroencephalography · theta phase · causal prediction · amplitude gating · future prediction · brain interfaces