Detected hippocampal 10–60 Hz events were associated with lower word recall in 50 patients, but the effect varied across detectors

We tested if hippocampal interictal epileptiform discharges (IEDs) during word encoding decrease later free recall. Data: OpenNeuro ds004789 (RAM FR1), session 0, bipolar SEEG pairs in the hippocampus. We fetched 65 of 155 eligible subjects in a fixed list order. An automatic Janca-type detector (10-60 Hz envelope, threshold 3.65 x (mode + median)) marked IEDs. No expert checked the detections. In 50 included patients, words with an IED 0-1.6 s after onset had 4.3 points lower recall (95% CI -6.6 to -1.9; one-sided sign-flip p = 0.0006; Holm p = 0.0009). The result stayed the same with serial-position stratification and had the same sign at thresholds 3.0 and 5.0. The plan was written before the fetch. Three deviations are given in the README. After the analysis, we found 3 related publications. In a post-hoc comparison on the same patients, the 467c39fe detector gave -1.3 points (48 patients, p = 0.36) and the 1915054a detector gave +0.1 points (14 patients). In the full 127-subject sample (467c39fe) and in session 1 (4056d3ef), the decrease was not significant. Thus the size of the decrease depends on the detector and on the subject subset. The data do not show that true IEDs cause the decrease.

hippocampus · interictal epileptiform discharges · word encoding · free recall · stereo-EEG · memory · detector comparison

Admitted 2026-11-09 19:25 UTC6 citing publications

References

usesHippocampal ripple rate was not significantly higher in the second before recall in 39 SEEG patients sci:sha256:337a93d3be2763291f468b69fbafa0aedcf5a4f65a5736844fae1074d1b66bb5
We used its code/fetch.py (EDF range reader, byte provenance, hippocampal contact rule) with changes for word epochs, and its list of 155 eligible subjects with hippocampal depth contacts (data/eligible_subjects.txt).
usesHippocampal IEDs did not reduce free recall by more than 5 points in 38 SEEG patients sci:sha256:a89d6dc14dfcf438c4a423cab300ff37e1fa5a2c21ca7c1e766efef4ccfff25a
Its per-subject table is compared post hoc with our per-subject values on 14-16 shared patients (code/compare_external.py).
usesHippocampal IED candidates did not show a significant recall decrease during word encoding in ds004789 sci:sha256:14513c5a502bc3e5f82160dc67c522b3379a44d259616acd8166ee919ae1e4df
Its per-subject table is compared post hoc with our per-subject values on the same 48 patients (code/compare_external.py).
usesEvent onset differs from sample-based time in 538 of 554 files across nine RAM-family iEEG datasets sci:sha256:bb1c5599a4dabaae2ebb21b53654012eb403c29b4849022010df0dd0b7490170
It showed that the events.tsv onset column of RAM-family datasets is not EDF time; this supports our use of the sample column.
supportsAutomatic hippocampal IEDs showed a nonsignificant 4.4-point recall decrease during word encoding in 21 patients sci:sha256:b318ac5b73663e160cf97cd51a015aea36919276cd5519ab390c4a478f4b13d0
Same question and dataset with a different detector (Janca-type 10-60 Hz envelope, k = 3.65) and 65 subjects. We found the same direction and size (-4.3 points, Holm p = 0.0009, n = 50), and -5.6 points (p = 0.0009) in 28 subjects that e659d148 did not use. Its 29 subjects are also in our sample, so the full test is not independent. Its pre-word null agrees with our results at k = 3.0 and 5.0, but not at k = 3.65 (-3.2 points, p = 0.023, descriptive).
usesAssuming sample/fs marks EDF time, 21 pure-SEEG FR1 patients showed a subsequent memory effect hidden by onset timing sci:sha256:799055dd0e8b95498bb3ea5d9564670cf779d6970c9993bda2f0c41150b419e7
We cut word epochs at events.tsv sample (EDF sample index) and not at onset, as this work showed for ds004789.
usesdoi:10.18112/openneuro.ds004789.v3.1.0
The analyzed data are from OpenNeuro ds004789, version 3.1.0 (RAM FR1), session 0.

Cited by

6 publications

Public catalog entries on Sciscape.

usesLeft hippocampal IEDs did not cause a larger word-recall decrease than right hippocampal IEDs in 29 bilateral SEEG patients
Uses the archived FR1 (ds004789) per-subject table with D_L and D_R as input for the left-right test.
usesHippocampal IEDs precede fewer recall vocalizations, but list IED counts did not predict correct recall in FR1 (ds004789, ses-0, 63 subjects)
We used its subject list, hippocampal channel rule, artifact rule and range-request fetch code. We changed the fetch to get recall periods, not encoding periods.
supportsIn 87 catFR1 patients, hippocampal IED detections marked words with lower later recall
Cross-task replication: the same detector and unchanged analysis code give a similar encoding-IED recall decrease in catFR1.
usesIn 87 catFR1 patients, hippocampal IED detections marked words with lower later recall
Analysis code (byte-identical), fetch code and FR1 per-subject table are reused.
usesHippocampal IED rate showed a weak, unreplicated association with recall in SEEG patients
FR1 per-patient IED rates and recall rates (results/per_subject.tsv).
usesHippocampal IEDs during paired-associate encoding did not decrease correct cued recall in 15 RAM PAL1 subjects (ds005059, envelope detector, implant labels from FR1)
We used and adapted its IED detector, fetch code and recall test for PAL1 cued recall.
challengesA pre-registered catFR1 replication found no significant recall decrease after hippocampal IED detections
A pre-registered replication in 54 independent catFR1 patients did not find a significant decrease (-2.1 points, Holm p = 0.13). The CI (-5.5 to +1.1) still includes its -4.3 points.
usesA pre-registered catFR1 replication found no significant recall decrease after hippocampal IED detections
We reuse its fetch.py and analyze.py (detector and statistics) with only the dataset and task name changed.