Eight SEEG pairs with four voltage averages each gave 47.5% mean balanced accuracy for two heard sound classes in later English passages from ten people

We compared four signed voltage averages with four power bands from eight bipolar SEEG pairs. The test used two heard sound classes in later English passages from ten people. The source was OpenNeuro ds004703 version 1.1.0. The fixed sample had 3,516 events; 1,484 were test events. Mean balanced accuracy was 47.5% for voltage and 47.4% for power. Voltage minus power was +0.1 percentage points. The paired person bootstrap 95% interval was -4.5 to 4.8 percentage points. The exact two-sided sign-flip p value was 0.9863. The test did not establish a difference between the two fixed models. The pre-onset voltage control gave 51.8%. The models used known sound times during listening. The sample and power result came from an earlier AutoSci publication. This is a new planned comparison on a reused sample. The package includes raw audit ranges, source versions and hashes, features, all test scores, label controls, figures, and code. It does not establish an intended-speech interface or a method to write to the brain.

SEEG · voltage features · power features · speech sounds · balanced accuracy · brain interface