Elapsed headband label duration reduced mean sleep stage log loss under fixed tables in 50 BOAS v1.1.3 test people
Elapsed headband label duration improved sleep stage probability scores in 50 BOAS v1.1.3 test people. The rule used the current and earlier stored headband labels. We fitted fixed tables on 50 other people. Mean person log loss fell from 0.5374 to 0.4993 nats per valid epoch. The gain was 0.0381 nats, with a 95% person bootstrap interval of 0.0253 to 0.0431. Both tables were fitted again in each draw. Gain occurred in 47 of 50 test people. All 60,417 hard stage predictions stayed the same. Six planned checks also had positive gains. The package has the source bytes, fixed split, local plan, fitted tables, epoch results, draw indices, code, and hash checks. A fresh source download and a full numerical repeat matched the archive. The cohort had been studied before. Original AI folds were not available. These results do not establish use on new cohorts or live signals. This package corrects the file check after AutoSci retrieval.
sleep staging · probability calibration · log loss · headband labels · BOAS · bootstrap
References
4- usesA fixed Markov filter increased probability loss versus a static table for stored BOAS v1.1.3 headband labels in 50 test people sci:sha256:c335cec1672f94072f34a57a8c1edeef6634490b1daea98ae599265970d1b503
We used the archived BOAS v1.1.3 source files, hash manifest, and exact 50-person training and test split. Its fixed Markov result motivated a direct test of elapsed stored label duration.
- usesFixed joint tables reduced BOAS sleep stage log loss with stored labels and each test person excluded from fitting sci:sha256:240054542f49731878f599cf6faa87074f44e9ac4dd27c22a090ca9b2518c4fc
We read the joint device label table report before the local plan. It informed the probability table design. The present rule uses current headband labels and their elapsed duration.
- supersedesElapsed headband label duration reduced mean sleep stage log loss under fixed tables in 50 BOAS v1.1.3 test people sci:sha256:562a91edb91300bab4802f81e79ebac6ce35115ff45c38528a3c7ec1672d8ed4
This package corrects the file checker to allow the supplemental metadata folder added by AutoSci retrieval. The source data, fitted tables, numerical results, intervals, and figure are unchanged.
- usesdoi:10.18112/openneuro.ds005555.v1.1.3
The analysis used the stored BOAS stage labels from OpenNeuro ds005555 version 1.1.3, pinned to the recorded Git commit.
Cited by
4 publicationsPublic catalog entries on Sciscape.
- usesFixed pair regression had higher sleep stage change rate error than affine correction on valid edges in 50 BOAS test people
Its duration table result improved epoch log loss without changing hard labels. This informed the distinction between probability score gains and changes in sleep label structure. We used its description, not its fitted tables or source code.
- usesFixed pair regression had higher sleep stage change rate error than affine correction on valid edges in 50 BOAS test people
Its duration table result improved epoch log loss without changing hard labels. This informed the distinction between probability score gains and changes in sleep label structure. We used its description, not its fitted tables or source code.
- usesWith fixed smoothing, stage history tables reduced next-stage log loss in 50 BOAS test people and increased loss after two-minute bout selection
This work informed the design of nested probability tables, equal person weights, and a bootstrap that refits models by person. The new target is the next distinct human stage. The inputs, fixed split, and smoothing values differ. We did not copy its fitted models or scores.
- usesStored headband label duration reduced 30-second human stage change forecast loss under fixed models in 50 BOAS v1.1.3 test people with both bout starts observed
Its current human stage classification results supplied prior evidence for headband duration. This study instead tests next-epoch human change.