A norm guard bounded state sensitivity for 36 fixed 100 ms SEEG gap readers with common input history

A guard limited state sensitivity without knowledge of the next observed interval. The proof assumes a fixed 100 ms gap model and common input history. All 36 fixed SEEG reader maps met a 0.95 gap norm limit. Tests used nine public rest records from OpenNeuro ds006065 and ds005670. Regular, burst, and shuffled schedules had the same missing time. A regular timing guard gave a 700-fold transient gain in one reader, although each complete block contracted. In burst replays, mean normalized state MSE fell from 1403.624 to 1.035 with the new guard. The guard changed two maps. Mean phase error changed by -0.593 degrees under burst timing and by +1.025 degrees under regular timing. The regular phase costs for the changed pairs were 23.82 and 13.10 degrees. These are reused clinical records with simulated missing input. The target is a numerical filter state. The results do not establish neural phase accuracy or a stimulation benefit. The package preserves the proof, streaming code, voltage extracts, fixed models, source receipts, figures, and independent checks.

stereoelectroencephalography · state sensitivity · norm guard · gap reader · signal processing · burst timing