A 5 ms delayed companion gave no clear mean error decrease for ridge recovery of simulated 50 ms hippocampal SEEG loss in seven people

We tested whether an intact companion depth channel reduced voltage error during a known, simulated 50 ms loss of one channel. The test used 222 public records from seven people with chronic hippocampal depth leads in OpenNeuro ds007095 v1.0.0. A ridge model used companion samples at least 5 ms old. A matched reference froze companion history at the start of each loss. Equal-person mean normalized squared error was 0.70434 with the available companion and 0.70940 with frozen history. The mean decrease was 0.00506, or 0.71%. The descriptive 95% whole-person bootstrap interval was -0.00013 to 0.01116. The exact two-sided sign-flip reference gave p = 0.21875. Five people had lower error and two had higher error. Longer companion delays reduced the mean gain. These data did not establish a reliable mean gain or equal performance. The reused-cohort analysis is exploratory. It precedes marked stimulation and assumes one fully intact companion. It does not test real stimulation artifacts or task decoding. The package includes all selected raw data, code, source hashes, models, predictions, checks, and three figures. A separate full replay matched all fixed results exactly.

intracranial EEG · hippocampus · channel loss · signal recovery · ridge regression · brain-computer interface