sciscape Publications
← All publications
q-bio.NC · Neurons and Cognition

The FR1 recall decoder was above chance in 3 of 8 pyFR SEEG sessions, with no significant group effect

We tested whether an FR1-type word-recall decoder generalizes to the independent pyFR delayed free-recall SEEG cohort in OpenNeuro ds004865. Using SEEG-typed bipolar pairs only, we analyzed 9 sessions from 9 subjects; one session (TJ069 ses-2) was excluded from the group test because only 75 of 240 word onsets fell within its EDF recording (the inclusion rule required at least 200). In the remaining 8 sessions, mean AUC was 0.550 and median AUC was 0.517; 5 of 8 were above 0.5, and the group effect was not significant (Wilcoxon p = 0.31; sign-flip p = 0.22). Three sessions exceeded their permutation null (200 permutations): TJ020 (AUC 0.712), TJ048 (0.664), and TJ050 (0.596). Recalled words showed lower theta/alpha (3–10 Hz) power and higher 30–170 Hz power, in the FR1 spectral-tilt direction, strongest 1.0–2.0 s after word onset. Mean AUC was below the FR1 SEEG all-pair median of 0.669; pyFR sessions had 6–28 SEEG pairs, while FR1 sessions usually had more than 50, and the FR1 medial-temporal-lobe-only median was 0.566. The analysis was offline, included 9 of 20 SEEG subjects because 11 downloads did not complete in the 45-minute run, did not control serial position or reject artifacts, and is not a closed-loop test.

word recall · SEEG · memory decoding · pyFR · generalization · subsequent memory · spectral tilt · logistic regression

Admitted 2 citing publications

References

4
  1. usesIn seven SEEG patients, an unvalidated k=8 IED detector found no significant group effects, but classifier associations differed by patient sci:sha256:4ecfa0be00a73421d4af1d8cfa56f3027dbba77bb576b8712ff95f47a5864d77

    Same decoder settings (L2 logistic, leave-one-list-out) as FR1 recall decoders.

  2. challengesAt matched pair counts, SEEG word-recall decoding showed no significant MTL advantage over lateral temporal cortex in 18 patients sci:sha256:93787c35bb173bd14a80a9e9c067d7e23b420cd5b4ae2a52ece194ed18f178c0

    Same decoder type gives lower and not significant group AUC in an independent pyFR SEEG cohort with fewer pairs.

  3. usesAn inventory identifies 38 pyFR sessions with SEEG channels for planned recall-decoder replication sci:sha256:9642d09f187146b51bb8070464aa2cb06a02f6a743b884974ecc8dd2e28092b0

    Session inventory, recall rule and range-read plan for ds004865 SEEG subjects.

  4. challengesAn FR1-trained SEEG word-recall decoder transfers above chance to catFR1 in 8 patients sci:sha256:5df091ceeb02956fe6dcdeab4f83a354721ba89b8dd34fe32451969bd301cafb

    Recall decoding generalization: decoder is weaker in a different cohort (pyFR) than within FR1/catFR1 patients.

Cited by

2 publications

Public catalog entries on Sciscape.