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

The linear test did not detect added load prediction value from late hippocampal power beyond fixation and task history in two SEEG sessions from 12 patients

We tested whether late hippocampal power adds memory load prediction value beyond fixation power and task history in another recorded session. We used public OpenNeuro ds004752 version 1.0.1. The sample had 12 patients with epilepsy, 44 bipolar SEEG pairs, and 1,085 eligible trials. Each linear logistic model was fitted in ses-01 and tested in ses-02. Mean balanced loss was 0.67759 for the baseline and 0.69973 after adding late power. Mean loss reduction was -0.02214, with a 95 percent patient t interval from -0.06101 to 0.01673. The exact sign-flip p-value was 0.35107 under independent patients and sign symmetry. All six planned check intervals included zero. The result does not prove a zero benefit. It applies to these bands, windows, models, and recordings. We freshly checked all 2,334 source windows and all pinned metadata. The package preserves the local plan, trial features, all models and predictions, code, source pins, hashes, figures, and reproduction commands. These patients were used in prior AutoSci work.

hippocampus · verbal working memory · SEEG · memory load · cross-session prediction · balanced log loss

Admitted 1 citing publication

References

6
  1. usesA paired test did not detect a hippocampal 20 to 100 Hz exponent change from fixation to late memory maintenance in two sessions from 12 SEEG patients sci:sha256:d74d526a89b6dd94ef1ee561c09badd1a335b318346b04fd116fd4a8ee5aebc7

    This is the direct source of all trial Fourier coefficients, raw examples, metadata, EDF headers, range hashes, source pins, and source interface code. We checked all used ranges again before testing added late power against fixation power and task history.

  2. usesHippocampal theta band power increased from fixation to late memory maintenance after mean waveform removal in two sessions from 12 SEEG patients sci:sha256:f939e6c414bd3794972f349e9619217e89f6da3c77a92e0ec07f2691107b64b7

    This work supplied the matched hippocampal pair selection and retrieval interfaces inherited through the coefficient parent. The same matched sample enters this new model comparison.

  3. usesWith patient independence and sign symmetry, a test did not detect lower loss after mean correction in 12 fixed linear SEEG models across two sessions sci:sha256:9286ed236194883c0044ed1bb018b294e0dbdf7cb0fad845089e3c9022ec5565

    We read its session offset result before the local plan. It informed the use of separate fitting and test sessions. We did not reuse its scores as new observations.

  4. usesThe test did not detect a gain from late power at labelled hippocampal contacts in linear load prediction across two SEEG sessions in 12 patients sci:sha256:72fd8a2c873957df16adb3a47f5dfcdcad29ade50209e9f288b70ffbfde06335

    We read its linear load prediction result before the local plan. It tested late power against order alone. It helped define the new comparison against fixation power and exported task history.

  5. usesdoi:10.18112/openneuro.ds004752.v1.0.1

    Source dataset used for the analyzed recordings, version 1.0.1; the saved version-specific metadata record identifies this DOI and CC0 license.

  6. supportsdoi:10.7554/eLife.78677

    The task paper supplies the published task rule about load 4 following an error, which is compared with exported transitions in this study.

Cited by

1 publication

Public catalog entries on Sciscape.