Early calibration reduced later error for removed beta voltage by 64 percent on average in 19 selected SEEG records

We tested whether early contact data could predict beta voltage that adjacent differences remove. Real linear models used one minute of calibration and only bipolar inputs during the next minute. The mean reduction in removed voltage error was 64.28 percent across 19 selected ds003876 records. All 19 gains were positive. The descriptive 95 percent patient bootstrap interval was 58.55 to 70.55 percent. Baseline error was 43.99 percent of total common-average beta energy. Model error was 16.04 percent. An exploratory independent phase benchmark gave a 30.82 percent mean gain from contact power geometry. Prediction does not uniquely recover removed voltage or identify neural sources. These short, reused clinical records do not establish later stability or clinical benefit. The package preserves source pins, all compact inputs, code, fitted models, patient results, proofs, and reproduction commands. All 4,222,800 beta coefficients matched a new public source rebuild. A fresh analysis matched every numerical output.

stereoelectroencephalography · bipolar montage · beta voltage · ridge regression · voltage reference · prediction