Power bounds were wide 100 ms after 10 ms gaps in 35 SEEG records under an assumed limit of twice an early voltage maximum

We tested power bounds after 10 ms simulated voltage gaps in OpenNeuro ds006519 version 1.0.0. The sample had 35 SEEG records from 21 patients and four fixed channels per record. The reader used a causal 70 to 110 Hz filter and a 100 ms mean square. The main assumed voltage limit was twice the maximum absolute voltage in a separate early 2 s segment. At 100 ms after the gap, a matrix calculation reduced interval width by 24.3% against an impulse envelope bound. The descriptive patient bootstrap interval was 23.5% to 25.0%. But none of 820 cases had width below 10% of power before the gap. Median times to this width rule were 250 ms for the matrix method and 285 ms for the envelope method. The main voltage limit held in 817 cases. All 82,517 tested power points in those cases were covered. The smaller one-times limit failed in 62 cases; 12 had a missed power point. All 35 saved voltage arrays matched fresh public source files exactly. The package keeps exact selected voltage, a proof, code, source hashes, results, three figures, and reproduction steps. The bound requires the stated voltage limit and known input outside the gap. These tests do not establish a neural decoder, an actual stimulation effect, or an online control rule.

power bounds · missing samples · stereoelectroencephalography · SEEG · causal filtering · signal processing · brain-computer interface