Adding a theta score to power reduced mean comparison loss in run 02 of 15 SEEG patients under fixed models and pair rules

We tested whether a theta phase to high frequency amplitude score adds information beyond high frequency power. Models were fitted in run 01 from 15 SEEG patients. They were fixed before testing run 02 from 15 other patients. Public data came from OpenNeuro ds004100, version 1.1.3. Clinical labels defined eligible bipolar pairs and comparisons in the same contact name group. Mean comparison loss fell from 0.49271 to 0.47316 nats. The mean gain was 0.01955 nats. Its 95 percent patient bootstrap interval was 0.00432 to 0.03729. This interval is conditional on the fitted models and assumes independent, exchangeable test patients. Eleven of 15 patients had positive gains. The AUC gain interval included zero. The other recording gave a mean loss gain of only 0.00247 nats, with an interval that included zero. A contact number adjustment also gave weaker evidence. This exploratory result applies to fixed models, brief intervals, and known label pair rules in one reused cohort. The coordinate cutoff uses unconverted source numbers with unverified physical units. The package contains code, features, all test comparisons, source identities, exact retrieval recipes, results, and repeat instructions. All 60 analyzed public signal extracts reproduced the stored features exactly. Separate model and statistical checks passed.

SEEG · theta phase · high frequency power · phase-amplitude coupling · seizure onset zone · comparison loss · epilepsy