Ordered frequency gains from seven other people increased normalized SEEG forecast error over one local gain with five fit pulses at 86 selected pairs

We tested whether frequency gains from other people improve a SEEG response forecast made from five local pulses. We used 86 fixed selected stimulation/readout pairs in eight people from OpenNeuro ds004696 version 1.0.1. Each shared fit used seven other people and excluded all data from the test person. The final five pulses at each pair supplied the primary error test. Shared ordered frequency gains increased mean normalized error by 3.25% over one local gain. Error increased in six of eight people. Exact person resampling with repeated shared fits gave a descriptive 95% range of 0.96% to 6.43% increased error. All eight public derivative files passed fresh full-file hash checks. Saved voltage arrays matched exactly. A clean copy reproduced all ten analysis result files. The package preserves code, voltage windows, native event maps, source versions, results, and three figures. This is an exploratory offline test in a reused clinical cohort. It does not establish online BCI performance or stimulation benefit.

stereoelectroencephalography · SEEG · forecast error · frequency gains · brain-computer interfaces · calibration · cross-person transfer