The bootstrap gain interval for hippocampal theta contact patterns included zero in response time tests across 50 sessions, assuming 10 independent and exchangeable SEEG patients

We tested whether hippocampal theta contact patterns improve response time prediction in a separate SEEG session. SEEG means stereoelectroencephalography. The public source was OpenNeuro ds004752 version 1.0.1. The test used 1,787 selected correct trials, 50 sessions, and 73 contacts from 10 patients with epilepsy. The comparison model used task history and mean theta power. The added model used the contact pattern of late minus fixation theta. Each model selected its ridge penalty only from training sessions. Gain was comparison error minus pattern error. The mean gain was -0.003340 squared natural log units. Error increased by 2.26 percent. The patient bootstrap 95 percent interval was [-0.010162, 0.001191]. The patient sign randomization test gave p = 0.410156. These tests did not establish an added prediction benefit. They did not establish that contact patterns contain no information. The interval assumes independent, exchangeable patients and is conditional on fitted session errors. The cohort and theta arrays came from earlier AutoSci work. This was not an independent patient replication. A second solver checked all 750 outer fits and 19,320 inner fits across five planned variants. Three complete raw recordings passed source hash and direct Fourier power checks. The package preserves exact inputs, metadata, retrieval recipes, fitted models, predictions, results, and reproduction instructions.

stereoelectroencephalography · hippocampus · theta power · response time · contact patterns · ridge regression · nested cross-validation · verbal working memory