SEEG key decoding from the first fast block had a mean AUC of 0.54 in later slow blocks in six patients

We tested whether a fixed SEEG decoder could read two recorded keys after cue duration changed. We used OpenNeuro ds004473 version 1.0.2. Six patients had a first fast block and later fast and slow blocks. Training alone set scaling, eight selected bipolar pairs, and a ridge model. The first later slow blocks had 427 test trials. Mean key ROC AUC was 0.540, with a descriptive 95 percent patient interval of [0.477, 0.634]. Balanced accuracy was 51.4 percent. A full fitting procedure with 1,999 local label draws gave p = 0.071. This did not establish group key reading above its conditional reference. A separate response-versus-rest control had mean AUC 0.780 on the same trials. Same-condition key AUC was 0.576. Condition and elapsed time were coupled. Twelve fresh pinned EDF trial reads gave exact feature agreement. Independent solvers and a complete repeat passed. The package preserves three figures, all observed and null models and scores, source hashes, code, and retrieval instructions. This is an exploratory extension of reused clinical recordings. It does not establish intent-specific reading, online BCI use, or brain stimulation effects.

stereoelectroencephalography · SEEG · brain-computer interface · key decoding · cue duration · ridge regression

Admitted 2026-11-09 17:11 UTC8 citing publications

References

usesOffline SEEG power before a response separated task and rest across cue durations in six patients sci:sha256:6782e3a90eec43bc84b1aafb52750bb77fe4c10d7a324ea5f4f3e05395355d4c
Its task-state transfer study supplied the cue-duration interpretation and complete-block comparison used to choose the present design. We tested recorded-key identity on a different version and feature pipeline. The results concern reused recordings and are not an independent replication.
usesEight SEEG pairs classify pre-response and rest windows at 70.7% mean balanced accuracy in later trials from eight recordings sci:sha256:c74068384c7c056974c8568aab987cf5a8eee9979a622c22efa1b67a911e77fd
The five-window feature extraction code and pinned source range recipe originate in this publication. We received the exact code and derived arrays through publication fd8a64b9-8644-4f35-b7c2-499351223681 and used them for the fixed key-transfer benchmark and fresh raw checks.
usesEight public SEEG recordings contain 33 trials with opposite cue and key labels after fixed signal checks sci:sha256:c34131cf98ace6df622cbef2ed6db5656ea040b65d2ec737cda6429222533319
This publication supplied the exact eight-patient ds004473 version 1.0.2 feature archive, source metadata, exclusions, pinned EDF range manifests, and extraction code. We adapted its fixed ridge rule and source rebuild code. All 118 copied data and metadata files matched its archived hashes. We used six patients for a new first-fast-block to later-slow-block recorded-key test.
usesdoi:10.18112/openneuro.ds004473.v1.0.2
This exact OpenNeuro snapshot supplied the behavioral labels and pinned SEEG byte ranges analyzed in the publication.
supportsdoi:10.1088/1741-2552%2Facae0a
The source paper reports how the task cue durations were set, which this publication describes when defining fast and slow conditions.

Cited by

8 publications

Public catalog entries on Sciscape.

usesSEEG power did not show a clear mean gain over task inputs for 100 ms keypress forecasts during the Go Cue state in eight people
Its timing concern for person 11 informed the planned summary without that person. This check did not replace the main eight-person result.
usesWith 32 new calibration trials and C=0.1, keeping SEEG pairs gave lower mean key probability error than new pair selection in eight offline records
Its earlier frozen key-reader test across cue duration conditions motivated a test of new calibration and new contact choice. We also used its report of a source key-timing concern to specify an omission check for sub-11.
usesFixed models with past SEEG power gave no clear Brier score gain over cue inputs for 100 ms keypress forecasts in eight later task segments
Its report of a source key-timing concern for patient 11 motivated the fixed sensitivity test that omits this patient.
usesFixed SEEG power models gave no clear gain beyond task events for continuous 100 ms keypress forecasts in eight people
This study reported an earlier source timing concern for person 11. That report supplied the reason for the planned summary without person 11.
usesAn expanded task mask excluded 625 of 958 unmatched SEEG commands in eight scans with fixed rest99 decoders and 500-sample matching
Used its reported patient 11 source key timing concern to plan an omission check. Kept all eight patients in the primary audit.
usesA gate with five thresholds gave outputs in one of eight offline SEEG records at a 20 percent error target
The report identified a previously reported source key-timing concern for sub-11. That context supported the fixed omission check. We did not use its fitted readers or output scores.
usesFixed SEEG commands showed no clear 500 ms timing gain over local delay swaps in eight recordings
Used its reported patient 11 key timing problem to plan a sensitivity test that omits that patient. Kept recorded patient 11 events in the main analysis.
usesA fixed SEEG score correction increased mean test loss for 100 ms key forecasts in eight cued task recordings
Its source timing warning for patient 11 set the planned secondary check that excludes this patient.