Word times disagree with sample times in all 485 tested bipolar SEEG sessions from ds004789 v3.1.0 under EDF timing

We tested whether two event fields select the same word encoding windows in public SEEG data. The fields are onset time and sample index. We used all 485 bipolar sessions with SEEG channels in ds004789 v3.1.0. These sessions came from 238 participants. Under nominal EDF timing, all 111,070 valid paired words exceeded the agreement tolerance. The tolerance was one sample or 1 ms, whichever was larger. Both 1.6-second windows were inside the recording for 106,859 words. Of these pairs, 92,082 had no overlap, or 86.17%. A constant shift left residual differences that depended on the required precision. Independent calculations reproduced the counts. A second EDF library checked all header rates and the source voltage example. The source project had already identified the time origin issue. This work adds a complete count for the tested release scope and a preserved audit method. The package contains source metadata, versioned EDF byte ranges, hashes, code, results, and retrieval instructions. The audit does not establish physical stimulus time.

stereoelectroencephalography · event timing · sample index · BIDS · EDF · OpenNeuro · data audit