Stored SEEG code runs of at least 1 ms covered less than half the channels at once between the pre and post labels in all 35 ds006519 v1.0.0 records

We measured exact stored-code repeats in all 35 released SEEG records from 21 patients in OpenNeuro ds006519 v1.0.0. A run had equal native float32 bits and a span of at least 1 ms. The test interval ran from pre-window end to post-window start. It includes stimulation and buffer time. Ten records had a run, but no sample had runs in half the channels at once. The largest simultaneous fraction was 10.667%. The equal-patient mean change from the pre window was -0.023030 percentage points. The descriptive 95% patient bootstrap interval was -0.132640 to +0.094632 points. A fixed 10 ms check found long runs in one channel in one record. This channel was the stimulated pair. No such run occurred outside the stimulated shaft. The results define the measured scope of exact-repeat flags in this exported dataset. They do not show that signals without repeats are suitable for neural decoding. All 355 source files passed byte and hash checks. A second algorithm matched masks across 148,639,816 native channel-samples. A fresh extraction reproduced the main results. The package preserves exact run arrays, native voltage examples, all result tables, four figures, source metadata, hashes, code, and a full retrieval recipe. The secondary timing tests use inherited voltage landmarks. Labels and landmarks are not hardware pulse times.

stereoelectroencephalography · SEEG · electrical stimulation · signal quality · exact code repeats · float32 · signal processing · artifact detection