A 95 percent bootstrap interval included zero HFO count order forecast gain from joint minute sampling in 17 SEEG sleep records under an independent, exchangeable patient model
We tested whether joint sampling of minute counts improves a forecast of which channel has more HFO marks in the next half hour. We used released STE marks from OpenNeuro ds005398 version 1.1.3. Seventeen SEEG sleep records supplied 474 SOZ versus other channel pairs on the same shaft. The first half hour supplied fit counts. The second half hour supplied test outcomes. Joint and independent sampling had identical channel marginal laws. Exact finite convolution gave the forecast probabilities. Mean Brier loss was 0.135088 for joint sampling and 0.132805 for independent sampling. Mean gain was -0.002283. The 95 percent patient bootstrap interval was [-0.007109, 0.001216], under independent, exchangeable patient effects. Each fixed joint versus independent check also had an interval which included zero. The result does not establish equivalence. The package includes source bytes, pinned versions and hashes, count arrays, all forecasts, patient scores, bootstrap draws, code, figures, and instructions. Fresh downloads and an offline repeat reproduced the outputs. The study concerns released marks and one window split per patient. It does not prove detector validity or clinical benefit.
high frequency oscillations · stereo-EEG · count forecasting · joint sampling · Brier score · finite convolution · patient bootstrap
References
3- usesIn the first hour of 18 SEEG sleep records, 30-minute STE HFO maps had higher expected top 20% channel agreement than 5-minute maps sci:sha256:4d9fd5f1eaa1c005c1d288b2e596ac25540bdc558e18948e96c492be41759f09
We used its 94-object pinned public source bundle, source manifest, and cohort and channel eligibility conventions. We wrote new forecast code, downloaded every source object again, and verified all hashes. The same patient cohort is reused; this is not an independent replication of its duration result.
- usesA fixed variance rule reduced prediction loss for later one-minute counts of unfiltered MNI HFO marks in 18 public SEEG sleep records sci:sha256:72f21f20f6b9dcfc92cc17c59090ef9f0dd07930e1b99d92c42ef8619684350a
Its earlier marginal count forecast result on the same cohort motivated holding channel marginal laws fixed while testing cross-channel dependence. The new target is later within-shaft count order, with Brier loss. We did not use its fitted models or output tables as new data.
- usesdoi:10.18112/openneuro.ds005398.v1.1.3
The analysis uses released STE and MNI event marks, participant and channel metadata, and EDF headers from OpenNeuro ds005398 version 1.1.3.
Cited by
3 publicationsPublic catalog entries on Sciscape.
- usesA 95 percent bootstrap interval included zero gain from joint shaft sampling for later STE HFO mark counts in 18 SEEG sleep records under an independent, exchangeable patient model
Used the pinned public source manifest and MIT retrieval code. The prior channel order comparison informed the new test of total counts across shafts.
- usesA 95 percent bootstrap interval included zero gain from joint shaft sampling for one-minute HFO total forecasts in 18 SEEG records, assuming independent, exchangeable patients
The prior count order study supplied context for joint and independent empirical sampling. This study uses a new total count endpoint and score.
- usesA 95 percent bootstrap interval included zero gain from joint shaft sampling for later STE HFO mark counts in 18 SEEG sleep records under an independent, exchangeable patient model
Used the pinned public source manifest and MIT retrieval code. The prior channel order comparison informed the new test of total counts across shafts.