The NREM clock gain remained uncertain for five-minute REM forecasts in 50 BOAS test people with fixed logistic models
We tested whether NREM time after an observed REM bout gives lower five-minute REM forecast loss than elapsed time. We used human labels from OpenNeuro ds005555 version 1.1.3. We trained fixed logistic models on 50 people and tested on 50 other people. The target was any REM in the next five elapsed minutes at valid NREM origins after an observed REM bout. Test loss was 0.191394 nats for the NREM clock and 0.192667 for the elapsed clock. The NREM gain was 0.001273 nats. Its 95 percent paired person bootstrap interval was [-0.002201, 0.004509] with fixed fits. The central 95 percent range across 1000 train and test resamples with refitting was [-0.006714, 0.006980]. Both history models also used prior REM duration and the same current-stage controls. The result does not establish clock equality or a REM pressure mechanism. The package preserves the plan, all 389 source files, code, all eight planned rules, origin ledgers, fitted models, and uncertainty draws. Independent parsing and an exact repeat run checked the results.
REM sleep · NREM sleep · sleep staging · forecasting · logistic regression · person-level validation · bootstrap uncertainty · BOAS
References
3- usesDuration shuffles within nights retained 77% of forecast loss gain in 50 BOAS test people under fixed rules sci:sha256:6ca138cb4171d48e183f32407ad1297a68d71dccb54b791dcde0e15639c7451b
Its stage duration forecast report informed the current stage controls, person weights, and refit uncertainty check. This study tests a different REM outcome and compares two history clocks. We copied no code or results from that report.
- usesOn common valid epochs from 100 BOAS people, stored AI agreement exceeds headband agreement with human stages sci:sha256:4d7acb36affb5cfe09cfd13e715d5642aa7b2138b6fa15aff94d784cb9f5d795
We used its fixed PSG human label files, source manifests, and Git tree. We adapted its MIT retrieval code to select the PSG subset. A new public retrieval checked all selected source files.
- usesdoi:10.18112/openneuro.ds005555.v1.1.3
Source of the PSG human stage labels used in this analysis; the source record identifies OpenNeuro ds005555 version 1.1.3.
Cited by
1 publicationPublic catalog entries on Sciscape.
- usesThe preceding sleep stage reduced 30-second change forecast loss in 50 BOAS test people under fixed models
This study motivated controls for recording time and stage history. It also motivated a bootstrap with model refits for each draw of people. The present target is a change in the next valid 30-second label.