Twenty reconstructed pulse blocks give higher late bipolar map agreement than five in seven people from ds006065

We tested a pulse budget question for brain-computer interfaces. The test used nine public SEEG files from seven people in OpenNeuro ds006065 version 1.0.0. Pulse times were reconstructed from the saved epoch layout. Late bipolar voltage maps used the 50–250 ms window. Mean rank agreement with the final 20 blocks rose from 0.835 with five early blocks to 0.878 with twenty. All seven people showed a gain. The mean gain was 0.043, with a descriptive 95% person-resampling interval of 0.024–0.067. Pre-event maps also had high agreement. Common mean contact maps showed a smaller gain. An exploratory check removed pairs with listed stimulation contacts. Rank agreement stayed similar, but pooled waveform agreement at twenty blocks fell from 0.986 to 0.652. Typical pair waveform agreement was weak. These limits prevent a claim of stable neural connectivity or BCI performance. The package includes figures, compact averages, a local analysis plan, source code, exact source identities, and a raw retrieval recipe. A full rebuild reproduced all saved arrays and eight numeric result files.

intracranial EEG · SEEG · brain-computer interfaces · evoked potentials · bipolar voltage maps · rank agreement · pulse calibration · reproducibility