Bilateral ridge forecasts gave no clear 50 ms error reduction across later sessions in seven people at equal input count

We tested whether the other hippocampal channel reduces 50 ms voltage forecast error across later sessions. The source was 250 public chronic depth SEEG records from seven people with epilepsy. Both main ridge readers used 80 input samples. The reference used 80 samples from the target channel. The bilateral reader used 40 samples from each channel. Earlier sessions selected and fitted each model. The test used 69 later sessions before marked stimulation. The equal person mean relative error reduction was -0.69%. Its descriptive 95% person bootstrap interval was -1.48% to +0.20%. The exact sign-flip reference gave p = 0.1875 under independent symmetric person contrasts. Two people had lower bilateral error. The bilateral reader gave a 0.89% mean reduction versus a shorter reference with fewer inputs. This shows why the input budget matters. The test does not establish equal performance, a causal brain connection, or clinical benefit. The package preserves all source files, hashes, code, weights, results, three figures, and repeat instructions. Fresh raw source checks, an independent EDF reader, independent score checks, and a complete replay passed.

hippocampus · brain-computer interface · voltage forecasting · ridge regression · session transfer · intracranial EEG · signal processing