Adding an inactive component increased STE mark forecast error on silent channels between consecutive 30-minute windows in 18 SEEG sleep records
We tested released HFO marks from OpenNeuro ds005398 version 1.1.3. The sample had 18 SEEG patients and 1285 bipolar channels. Of 624 channels without STE marks in the first 30 minutes, 151 had marks in the next 30 minutes. We fitted a Gamma-Poisson model and a model with an extra inactive component to early channel counts. The inactive component increased mean patient Brier loss from 0.1934 to 0.2011 on early-silent channels. The mean gain, defined as baseline loss minus mixture loss, was -0.007690. The 95 percent patient bootstrap interval was [-0.017807, -0.000735]. Both models underpredicted later marks. All fixed checks had negative mean gains. The finding concerns detector marks in these sleep windows. Biological inactivity remains untested. The package preserves source bytes, forecasts, fit checks, and code.
high-frequency oscillations · stereoelectroencephalography · forecasting · Gamma-Poisson model · Brier loss
References
4- uses20 ms detector matches gave an uncertain SOZ rank gain over combined counts in the first hour of 18 SEEG sleep records sci:sha256:e2fb21fc2b8904e6e332f4adef0382d28035541049ea83d70d8da9f15b9adbb1
This publication supplied the pinned public event tables, metadata, EDF headers, source manifest, and retrieval code. We freshly verified the inputs and wrote new forecast and verification code.
- 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 result on minute-count variance in this cohort supplied context for the question about early silence. The current test uses a new binary forecast endpoint and new model code.
- usesdoi:10.18112/openneuro.ds005398.v1.1.3
The analysis used released event tables and metadata from OpenNeuro ds005398 version 1.1.3; the version and DOI are recorded in the source manifest.
- doi:10.1111/epi.18545
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesSTE event counts gave higher mean SOZ AUC than event presence at 30 to 60 minutes in 18 SEEG sleep records
Its early-silent channel result informed the fixed question about early activity screening. We read this result before the local plan. No forecast or fitted model from that work entered the new scores. The patients are reused.
- usesSTE event counts gave higher mean SOZ AUC than event presence at 30 to 60 minutes in 18 SEEG sleep records
Its early-silent channel result informed the fixed question about early activity screening. We read this result before the local plan. No forecast or fitted model from that work entered the new scores. The patients are reused.