Earlier pair corrections reduce later SEEG slope conversion error in 24 participants at 20–100 Hz without 55–65 Hz under fixed reference pools

We tested whether an earlier contact pair correction reduces later SEEG spectral slope conversion error. We used public OpenNeuro ds005624 version 1.0.0. The analysis used 24 participants, 34 records, and 4091 eligible pairs. The first minute supplied corrections. The last minute supplied test data. Each record kept the same common average reference pool. We fit negative log-log slopes at 20 to 100 Hz and omitted 55 to 65 Hz. The comparison used earlier corrections from at least two other pairs on the same named shaft. These donor pairs shared no contact with the target. Mean squared error fell from 0.5077 to 0.2430, a reduction of 52.1%. The participant bootstrap interval for this reduction was 37.5% to 69.9%. All 24 participants had lower error. Both methods used the later mean contact slope. The result tests reference conversion within records. It does not test future absolute spectra, task effects, or clinical benefit. The package contains the plan, powers, source pins, raw examples, code, tables, participant bootstrap draws, and checks of all 70 fresh raw ranges.

stereoelectroencephalography · SEEG · spectral slope · bipolar recording · reference conversion · power spectrum · Welch method