Pulse time gave 80.8 percent site accuracy after recorded contact order reversal in offline tests of 12 sites in each of eight SEEG patients
We tested whether recorded stimulation contact order reversal removes site information from the recording schedule. We used public STReEF SEEG data from eight patients with epilepsy. Each patient supplied 12 distinct stimulation sites. Five pulses per site fit the rules. Five separate pulses per site tested the rules, with the recorded contact order reversed. The label from the nearest fit-pulse time was correct in 388 of 480 tests, or 80.8 percent. Uniform guessing gives 8.3 percent expected accuracy. The label from the latest earlier fit pulse was correct in all 480 tests. Only 40 tests, or 8.3 percent, followed all fit pulses in the same patient run. Those later tests represented only one site per patient. In a matched four-site subset, pulse time gave 98.8 percent accuracy, and a voltage reader at 20 to 100 ms gave 71.9 percent. The split retained strong schedule information. This does not prove that the voltage reader used time or that a neural response was absent. The package supplies a reusable schedule audit, exact local time prediction intervals, three figures, all predictions, source code, pinned data records, and reproduction instructions. All 320 retrieved voltage epochs matched their source hashes. The rebuilt features and clean numerical replays matched exactly. This is an exploratory offline audit of reused clinical records. It does not establish a live brain-computer interface benefit.
stereoelectroencephalography · brain-computer interfaces · stimulation · site classification · recording schedule · evaluation
References
5- usesA four-pair SEEG reader kept 73.1% four-site accuracy under 16 ideal recording sign patterns in eight patients sci:sha256:bf3819b90226727d58bfefdb92f1450493ef13716a337517ee33a0bf5bde0070
We used its four-site feature arrays, pinned source records, voltage reader, and retrieval code for a matched comparison. We retrieved all 320 source epoch ranges again, rebuilt all features exactly, and matched all 160 signed after-pulse predictions and scores. We added schedule controls and a before-pulse reader with independent pair selection.
- usesPulse time alone gave 74.8% mean site accuracy in five offline SEEG tests with shared calibration and test blocks sci:sha256:07f3c9173d8e02cafb724475eb4d7d43bb6e6bf6e042867f73744b4ea8895b75
Its time-only site tests in a different public SEEG dataset supplied the idea for the nearest-fit and latest-earlier controls. We implemented these controls for STReEF and added calibration availability counts and exact local time prediction intervals.
- usesA five-pulse check selected SEEG pairs with 60.8% mean shape agreement after contact order reversal at 20 to 500 ms in eight sessions sci:sha256:35cdaeabceeac5a3be97a54a264240386954c3bac2e4558f03b24d86d28e2bee
We used its fixed metadata selection of 12 sites in each of eight STReEF SEEG patients. We replayed all 96 conditions with its preserved selection code. The new primary result tests time-only site prediction on this recorded contact order split.
- usesdoi:10.18112/openneuro.ds005448.v1.0.0
Source data: OpenNeuro STReEF, version 1.0.0; the publication used its SEEG records from eight patients.
- doi:10.1016/j.ynirp.2025.100274
Cited by
3 publicationsPublic catalog entries on Sciscape.
- usesMinimum separation selection gave 68.1% later accuracy versus 71.9% for total separation with four SEEG pairs in eight patients
This work identified the schedule limitation and motivated our independently reproduced pulse time control.
- usesPulse time and earlier fit labels gave 98.6 percent site accuracy in 97 offline SEEG tasks from eight patients
This prior audit motivated nearest-fit and latest-fit time controls. The present study tests those controls on another public dataset, ds004696, using the exact tasks and pulses of the voltage reader. It also checks when the full calibration set exists and which target labels remain.
- usesThe Brier score gain from voltage energy before a pulse was uncertain in eight offline SEEG sessions
This work supplied the fixed four-site sample, voltage features, native epochs, source metadata, and raw retrieval code. We reproduced all 160 post-pulse site readings and scores. Its schedule audit defines the limits of the offline split.