A fit ratio gave 0.68 degrees higher mean 50 ms offline phase error than amplitude in seven SEEG-labelled rest recordings

We tested a rule that accepts phase forecasts from the amplitude of a fitted sinusoid divided by its residual error. We used 28 selected bipolar signals from seven public SEEG-labelled rest recordings in OpenNeuro ds006065 version 1.0.0. Each rule accepted 25 percent of early cutoffs. Its threshold then stayed fixed. At 50 ms, the fit ratio had a mean offline phase error of 56.26 degrees. Amplitude alone had 55.58 degrees. The paired error reduction was -0.68 degrees, with a descriptive 95 percent interval of -2.34 to 1.05 degrees. The sign-flip test gave p=0.484. This test did not show better phase accuracy for the fit ratio. It did not establish equal performance. The fit ratio accepted 23.66 percent of later cutoffs; amplitude accepted 26.09 percent. We proved that the ratio and forecast phase stay fixed under constant positive voltage gain within each fit window. Direct tests with gains of 0.5 and 2 confirmed this property. A centered filter defined the scoring phase. That phase is not known biological phase. The package includes exact inputs, source hashes, code, figures, and a full numerical replay. The data reuse a prior sci study. They do not form an independent replication or a stimulation test.

phase forecasting · signal processing · SEEG · brain-computer interfaces · fit ratio · intracranial EEG · voltage gain