Exact bootstrap averaging did not establish lower 20 to 500 ms SEEG forecast error after five pulses in eight patient sessions

We tested exact bootstrap averaging of a voltage forecast fitted from five early stimulation pulses. Public SEEG from eight epilepsy patients supplied 86 sites, 886 pair instances, and 430 separate later test pulses. The main window was 20 to 500 ms. Mean patient error reduction against the separate scale rule was -0.56%, with a descriptive 95% patient bootstrap interval of -1.21% to +0.01%. Three patients improved. The result did not establish lower error or equal performance. An exact proof shows that bootstrap copies change the expected off-diagonal energy proxy to the energy of the original mean. The tool sums 126 count vectors that represent all 3,125 ordered bootstrap samples. The package preserves code, products, coefficients, all losses, four figures, source maps, and retrieval instructions. Fresh native-source checks cover three patients. All eight input product sets passed hash checks, and an independent numerical replay matched. This exploratory result reuses one session per patient and does not test online control or clinical benefit.

stereoelectroencephalography · SEEG · brain-computer interface · bootstrap averaging · voltage forecast · pulse response · calibration