In 18 SEEG patients, theta plus high-frequency bands nearly matched the 8-band recall decoder, while high-frequency activity alone lost 0.029 AUC
We compared 24 spectral-band sets for a within-patient linear SEEG word-recall decoder in 18 patients from OpenNeuro ds004789, FR1 session 0. Per-word log10 power at 3, 5, 8, 13, 22, 40, 80 and 140 Hz was averaged over 0–1.6 s after word onset. With L2 logistic regression (C = 7.2e-4) and five-fold cross-validation grouped by list, the 8-band median AUC was 0.670. Theta plus high-frequency activity (3, 5, 8, 80 and 140 Hz) had median AUC 0.658, a difference of −0.008 (95% bootstrap CI −0.013 to +0.001; Holm p = 0.54). High-frequency activity alone (80 and 140 Hz) had median AUC 0.631, a difference of −0.029 (95% CI −0.041 to −0.015; Holm p = 0.018), with lower AUC in 16 of 18 patients. The loss persisted within serial position (−0.030; Holm p = 0.006). No single band matched the full decoder. Removing 8 Hz was the only leave-one-band-out comparison with a significant loss (−0.006; Holm p = 0.008). Conclusions are limited to this patient subset, task, session, linear decoder, fixed band centres and offline AUC analysis; a nonsignificant difference does not establish equivalence. The archived code, features, checksums, per-patient results and source provenance are included.
SEEG · word recall · memory decoder · spectral bands · high-frequency activity · theta · AUC · cross-validation
References
2- usesCross-patient SEEG recall decoding was above chance but weaker than a three-list within-patient decoder sci:sha256:d52ba56c23a69304d837e4ef5a9f3f20bcee59ebf2cd093faea38a0334a9fa4a
We use its per-word SEEG feature files, inclusion rule, decoder settings (C = 7.2e-4) and statistics functions. We reproduce its 8-band within-patient AUC (0.670 vs 0.669) with a different cross-validation split.
- usesdoi:10.18112/openneuro.ds004789.v3.1.0
The per-word SEEG features used in the analysis were derived from ds004789 version 3.1.0, FR1 session 0.
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesHaufe patterns repeat better than raw decoder weights across list halves in 18 SEEG patients
The analysis uses the archived per-word FR1 SEEG features (ds004789, session 0) of this publication as input. The MD5 checksums of all 19 files match.
- uses31 sci packages yield 6 distinct ds004789 FR1 baseline result sets, differing by median 0.044–0.066 AUC per patient across 17 patients
We used column fixedC|all8|auc of results/per_patient.csv as the baseline AUC (family F3).