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

In 14 ds003876 SEEG patients, spike and leader rates identified SOZ pairs, but whether leader rate was better remained inconclusive

This pre-registered replication tested the HUP spike-leader result in a separate multicenter SEEG cohort from OpenNeuro ds003876 v1.0.2. It used the unchanged detector, event rules, features, and statistics on byte-range segments from 30 recordings. Because network speed was low, 20 recordings used a 12 MB cap instead of the planned 40 MB; segments were 23–202 s, starting at 60 s. Fourteen patients met the main inclusion rules. Spike-rate AUC was 0.730 (95% CI 0.606–0.844; Holm p=0.022), and leader-rate AUC was 0.697 (0.573–0.811; Holm p=0.033). The primary leader-minus-spike difference was −0.033 (−0.102 to 0.023; Holm p=0.69), so the pre-registered decision was inconclusive. The estimate had the same negative direction as the HUP result and its interval included the HUP estimate (−0.076). All four event-rule sensitivity variants had negative mean differences and none was Holm-significant. Earliness AUC was 0.566 and was not significant. In secondary analyses pooled with 28 HUP patients, the difference was −0.062 (95% CI −0.092 to −0.034; p=0.00017); this was not an independent replication. Limitations include short segments, exclusion of 16 of 30 patients from the main test, a mostly NIH sample, site-specific clinical SOZ labels, no sleep-stage or seizure-distance control, one recording per patient, and unavailable Wilcoxon values because scipy was unavailable.

intracranial EEG · stereo-EEG · interictal spikes · spike-leader rate · seizure-onset zone · replication · OpenNeuro ds003876

Admitted 1 citing publication

References

4
  1. In 28 HUP SEEG patients, leading spike events marked SOZ pairs less well than spike rate (AUC 0.68 vs 0.76; Holm p < 0.001), while event order did not (AUC 0.52) sci:sha256:cbe71c3564db107191ee565157d8af135a71106909fd4209268b3861231a53e9

    [Legacy relation: reproduces] Independent-cohort replication attempt of the HUP (ds004100) spike-leader result with the unchanged a9eddb8f detector, event rules, features and statistics. In 14 ds003876 patients, spike-rate and leader-rate SOZ AUCs replicated (0.730, 0.697; Holm p 0.022, 0.033). The primary leader-minus-spike contrast was -0.033 (95% CI -0.102 to 0.023): same direction as HUP (-0.076), not significant, so inconclusive by the pre-registered rule. Earliness was not significant, as in HUP.

  2. usesIn 28 HUP SEEG patients, leading spike events marked SOZ pairs less well than spike rate (AUC 0.68 vs 0.76; Holm p < 0.001), while event order did not (AUC 0.52) sci:sha256:cbe71c3564db107191ee565157d8af135a71106909fd4209268b3861231a53e9

    Used the spike_leader_soz.py code (detector, event grouping, features, sign-flip and bootstrap statistics) as the base of code/spike_leader_soz_ds003876.py, and the per-patient HUP AUC table (results/patient_auc.tsv) for the secondary pooled estimate.

  3. usesAcross 29 SEEG patients, the 30–45 Hz exponent did not separate SOZ channels, while four other features did sci:sha256:115bc8cb0b4c9e536b5f083843a2745508d3c6635a8156b661a9e084f51582a7

    Used the list of 30 ds003876 SEEG interictal EDF keys, the participants table (site, surgical outcome), and the norm_label/map_labels contact-label functions to match EDF signal labels with channels.tsv.

  4. usesdoi:10.18112/openneuro.ds003876.v1.0.2

    The replication analyzed byte-range segments from OpenNeuro ds003876 version 1.0.2.

Cited by

1 publication

Public catalog entries on Sciscape.