Stimulation-pair SEEG channels had constant codes for at least 10 ms in 24 of 211 selected annotated word windows in ds005557 version 1.0.0

We tested stored SEEG values after electrical stimulation in OpenNeuro ds005557 version 1.0.0. We used 27 bipolar recordings from 15 people. We selected 211 stimulation windows and 216 controls. Each reading window used 1366 samples after the next WORD event. Timing used the event sample column. The planned 10 ms constant-value test flagged no window on at least 90 percent of channels. All 12 planned gate settings gave zero. An exploratory contact test found flags in 25 stimulation windows and two controls. Twenty-four stimulation windows flagged the labeled stimulation pair. They came from three recordings in two people. Only one contact was flagged at each positive sample. The three persistent contacts had mean flagged window fractions of 90.75, 91.37, and 58.41 percent. Their matched controls had no flags. The contact test was added after partial signal inspection. It does not establish a stimulation effect. The package provides a causal contact gate, full-channel exact-run tables for 427 excerpts, eight exact example contacts per excerpt, six complete validation excerpts, figures, source metadata, range hashes, and a version-pinned retrieval recipe. Independent checks reproduced all event tables, all 3416 example-contact gates, and six raw excerpts. A passed gate does not prove that data are suitable for decoding. Constant codes do not prove neural silence.

stereo-EEG · closed-loop stimulation · constant-code detection · signal quality · free recall