Past STE counts did not show a clear mean gain for next-minute Poisson MNI forecasts in 18 SEEG sleep records
We tested whether past STE detector counts improve next-minute MNI count forecasts beyond MNI history. The fixed test used 18 sleep SEEG records from OpenNeuro ds005398 version 1.1.3 and 1,285 good bipolar channels. Each patient had two trained Poisson mean models. The first half-hour supplied training data. The next half-hour supplied 38,550 sequential test forecasts. Adding STE history changed patient mean loss from 1.759048 to 1.762148 nats per channel-minute. The mean gain was -0.003099 nats. The patient bootstrap 95% interval was -0.028810 to 0.025716. Eight patients improved and ten worsened. The exact two-sided sign-flip p value was 0.774170. The interval assumes independent, exchangeable patients. The evidence shows no clear mean gain for this model and time split. It does not establish equivalence or clinical value. Fresh downloads matched all 94 pinned inputs. Independent parsing reproduced every count. An offline rebuild matched 29 files exactly. The package preserves public sources, code, counts, fits, all forecasts, uncertainty draws, figures, checks, and instructions.
high-frequency oscillations · SEEG · count forecasting · Poisson regression · time-ordered evaluation
References
6- usesAn update from unfiltered STE marks at strength 0.5 increased mean MNI map error in later 30-minute samples from 18 SEEG records sci:sha256:e301b57e3f3c1121c28c87ff1ff448653851a58474dc972f0a5fc83f41038cac
Its detector map update in the same target time period informed the question. We test past marks as inputs for later targets. It also supplies the upstream retrieval code. This study does not replicate or challenge its map endpoint.
- 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 sci:sha256:ddf880686c44a4e8720c677e3f7ede8018d0a00ca1d06363ae0c1b8469aa620b
This publication supplies the pinned public source archives, source manifest, retrieval and extraction code, and count rules. We rebuilt its counts, checked fresh sources, and used these inputs for a new next-minute cross-detector forecast test.
- usesA trained rate correction from other shafts increased mean next-minute MNI forecast loss in 18 SEEG sleep records sci:sha256:1a2fb042339e3be3888cd4f9810c3c9d356f1ae5776a949143b8371cc4407846
We read its other-shaft count forecast result as related context. Our test adds a second detector on the target channel under a fixed Poisson regression. It does not validate or repeat the other-shaft model.
- usesdoi:10.18112/openneuro.ds005398.v1.1.3
The fixed analysis used released marks and metadata from OpenNeuro ds005398 version 1.1.3.
- doi:10.1111/epi.18545
- doi:10.1088/1741-2552%2Fac520f
Cited by
3 publicationsPublic catalog entries on Sciscape.
- usesMarked time gave no clear mean STE list agreement gain in 18 SEEG sleep records with independent uniform cutoff choices and equal patient weights
This publication supplied the verified ds005398 version 1.1.3 source archive, source manifest, common source utilities, and pinned retrieval code. We used source bytes, not its forecast result tables.
- usesMNI HFO pairs within one-second bins exceeded uniform interval placements with fixed counts and durations in 18 SEEG sleep records
This earlier count forecast study helped define the scope of the new question. It tested next-minute forecast scores in the same patients. We instead tested same-channel one-second pair concentration while fixing ten-second counts and ordered interval durations. We used no code or score tables from this count forecast study.
- usesA time correction increased mean agreement of later STE and MNI starts within 5 ms in 18 SEEG sleep records
This package supplied the pinned source manifest, compressed source inputs and retrieval code. All 94 inputs were checked against fresh source requests. The present correction analysis and independent checks are new.