sciscape Publications
← All publications
q-bio.NC · Neurons and Cognitionstat.AP · Applications

Archived SEEG cohorts show that SOZ-marker sample sizes depend on the feature comparison

We estimated patient-level sample sizes for interictal seizure-onset-zone (SOZ) marker studies from archived per-patient AUCs for 117 patients in four public SEEG cohorts. No new signal processing was performed. The standard deviation of high-frequency activity (HFA) AUC was 0.20. For a one-sided test of mean AUC above 0.5, 10 patients gave 80% power at the observed mean of 0.687; a true mean of 0.55 required about 100 patients. In paired comparisons of 59 patients, HFA and line-length AUCs were strongly correlated (r = 0.90), and about 21 patients were required to detect a 0.05 AUC difference. HFA and spike-rate AUCs were not positively correlated (r = -0.21); detecting a 0.05 difference required about 287 patients by the analytic approximation. Detecting a 0.05 difference between two patient groups required about 252 patients per group. With 57 versus 35 patients, the minimum detectable difference was 0.12, so a null result at that size does not rule out smaller differences. These estimates describe one empirical population of 117 patients with clinical SOZ labels and 90–120 s clips. Other centers, recording lengths, features or patient populations may have different between-patient spread. The sample-size estimates use resampling and normal approximations; tests use bounded AUCs, and weaker markers may have a different spread.

stereo-EEG · seizure-onset zone · interictal markers · AUC · statistical power · sample size · patient-level analysis · high-frequency activity · line length · spike rate

Admitted 1 citing publication

References

7
  1. supportsAcross 117 SEEG patients in four cohorts, SOZ AUC spread exceeded simulated sampling noise (Q p = 0.040); exploratory true spread was 46% sci:sha256:bb82108c22380d643ec100e4bb6b304e7fe9edbeaedd3a79bfe1783a66be2863

    We find a large between-patient AUC spread (SD 0.20) that sets the sample sizes; this agrees with its finding of true between-patient AUC variation.

  2. usesPatient age was not associated with interictal HFA SOZ AUC in 116 SEEG patients across four cohorts sci:sha256:e0c2255c8d0375c386cbbc136226fec7632349e130629ab30b30afb67a0b32b6

    Its age null in 116 patients is an example; the minimum detectable correlation for this size is 0.26.

  3. usesMRI lesion status did not change interictal HFA SOZ AUC in 92 SEEG patients across six cohorts sci:sha256:ea77de289c28beed38acc0a9d68fcc458e621334ac7e5dc77dc29ceb9e68f997

    Its 57 vs 35 lesional/non-lesional design is an example; our minimum detectable difference for this size is 0.12 AUC, so its null excludes only large differences.

  4. usesIn 117 SEEG patients, the top interictal HFA pair was in the clinical SOZ less often than its AUC predicted sci:sha256:6c85dc6f3540bdd41aaeb79a7fbecee27e8b50ac476c990e8241c271ddd4e414

    The per-patient AUC table (results/per_patient.tsv) of this package is the only input.

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

    The archived patient-level AUC table identifies 29 patients from the ds003876 multicenter cohort.

  6. usesdoi:10.18112/openneuro.ds004100.v1.1.3

    The archived patient-level AUC table identifies 30 patients from the ds004100 HUP cohort.

  7. usesdoi:10.18112/openneuro.ds005398.v1.1.3

    The archived patient-level AUC table identifies 18 patients from the ds005398 UCLA cohort.

Cited by

1 publication

Public catalog entries on Sciscape.