Starting the timer at accepted crossings increased correct and incorrect output rates in seven SEEG rest replays with a 1 s holdoff

We tested two timer rules for a phase reader on public SEEG rest data. The test used four contact pairs in each of seven people from OpenNeuro ds006065 version 1.0.0. We kept the phase readers and confidence slopes fixed. We used earlier rest to update one confidence intercept per pair. The confidence gate was 0.7, and the holdoff was 1 s. Starting the timer only at accepted crossings increased correctly timed outputs from 4.52 to 13.91 per stored minute. Incorrectly timed outputs also rose, from 1.20 to 3.64 per minute. Both rates rose in all seven people. The gain in correct outputs was 9.39 per minute, with a descriptive 95 percent person interval of 4.91 to 13.65. Mean phase accuracy changed from 75.58 to 78.97 percent. Its change interval included a decrease. Thus, a fixed confidence gate did not fix the output dose. The test compared separate later rest records with an offline phase reference. No pulse changed the voltage. The package includes voltage, fixed models, code, source records, event indices, four figures, and exact reproduction checks.

SEEG · phase forecasting · brain-computer interface · timer policy · confidence gate · output timing · offline replay