A direction rule changed mean key Brier error by -0.00036 for eight SEEG pairs after 64 calibration trials in eight people

We tested pair selection in eight public SEEG recordings from OpenNeuro ds004473 version 1.0.2. Each reader used eight pairs, 64 calibration trials, and logistic C = 0.1. We required band power class differences to have the same direction in two 32-trial blocks. The test used 979 later trials in the next condition. Mean key Brier error was 0.27904 with this rule and 0.27940 with pooled selection. The paired difference was -0.00036. Its descriptive 95 percent bootstrap interval was -0.00792 to 0.00668. A constant probability of 0.5 had error 0.25. The direction rule did not show a later key prediction benefit. This is an exploratory offline test in a reused cohort. The package preserves features, voltages, source records, code, models, predictions, four figures, and replay checks.

stereoelectroencephalography · brain-computer interface · pair selection · Brier error · calibration · logistic regression · OpenNeuro