A rule for the last gap did not guarantee 45 degree phase accuracy for 71 of 110 passes after five simulated gaps in seven SEEG rest records under assumed voltage caps

We tested a phase rule after repeated loss of SEEG input. The test used seven public rest records and 28 selected bipolar pairs. Each of 3,220 sequences had five requested 25 ms gaps with starts 100 ms apart. The target was a 7 Hz causal filter state at 100 ms after the last missing sample. The phase limit was 45 degrees. The assumed cap was twice the early absolute centered maximum. The rule for the last gap accepted 110 readouts. The rule for all five gaps accepted 39. For the other 71 passes, constructed missing inputs met the cap and caused phase error above 45 degrees. The recorded input caused no accepted errors above that limit at the main target. The recorded cap failed in 13 sequences. The added mean first wait was 33.07 ms. The package provides the full missing-input tool, proof, exact selected data, all 18 test settings, figures, source checks, and repeat code. The result assumes a known initial state, exact received input, and valid missing-voltage caps. It does not establish better stimulation or a true biological phase.

phase estimation · SEEG · missing data · causal filter · signal processing · robust bounds · uncertainty · neural interfaces