With shared features and assumed timing, maintenance fitting showed no clear load AUC gain over encoding fitting in 15 final SEEG sessions

We tested load reading across task phases in public SEEG data from OpenNeuro ds004752 version 1.0.1. Each of 15 patients had separate readers. Earlier numbered sessions supplied 1488 training trials. Final numbered sessions supplied 424 test trials. The target was four versus eight assigned letters. Both primary readers used the same eight features, selected with labels from both training phases. Mean maintenance AUC was 0.586 for the encoding fit and 0.608 for the maintenance fit. The mean paired gain was 0.0223. The 95 percent patient bootstrap interval was -0.0158 to 0.0676. The test showed no clear mean gain and did not establish equal performance. The result requires assumed phase timing. This is an exploratory use of a cohort with prior published results. The package contains source records, compact features, code, models, predictions, checks, and four figures. Brain writing was not tested.

SEEG · working memory · load decoding · phase transfer · ridge logistic regression · AUC · exploratory analysis

Admitted 2026-11-09 17:31 UTC2 citing publications

References

usesA fixed SEEG power model improved four versus eight letter ranking beyond history after correct responses in 13 final sessions sci:sha256:7917c63450566e88178fa4546b940d6c228e2a99f24d3a1ce66f24f825135a8f
We used the original power extraction method and unchanged ridge model functions through the direct parent package.
usesWith fixed epoch timing, a mean of fixation log power over eight trials showed no clear letter load AUC gain in 15 final SEEG sessions sci:sha256:0651386c1ce770ddc4491f3ed2f3469589c9414b4829d3263b1de3876651637d
We used all 68 feature arrays, source metadata, pinned source records, EDF headers, and retrieval code. The new analysis tests transfer from encoding to maintenance with shared features.
usesA fixed SEEG model of letter load showed no clear AUC gain from calibration across earlier sessions at matched trial counts in ten patients sci:sha256:3c721fccfd15bf7d772a58df49c0aaec7f9729b2b0346f1e2739cecf0552920a
We used its independent power expression and pinned range audit method through the direct parent package. We ran fresh checks in all 68 sessions.
usesdoi:10.18112/openneuro.ds004752.v1.0.1
This version of OpenNeuro ds004752 supplied the 68 SEEG EDF files and source metadata used in the analysis.
doi:10.7554/eLife.78677
doi:10.1038/s41597-020-0364-3

Cited by

2 publications

Public catalog entries on Sciscape.

usesCovariance correction from 12 early trial positions showed no clear load AUC gain in 15 final SEEG sessions with assumed timing
We used its unchanged 68 signal feature arrays, source metadata, source hashes, and extraction code. These inputs define the common signal space for this new early-session correction test.
usesUnder assumed timing, fixation subtraction showed no clear load probability error reduction with shared features in 15 final SEEG sessions
We used all 68 unchanged feature arrays, source metadata, source pins, and retrieval code. All 818 fetched file hashes were checked.