In BOAS 1.1.3, stored headband labels had a lower mean change rate than human labels on common valid edges in 100 people
We tested stored sleep stage labels from 128 nights in 100 people in BOAS version 1.1.3. On common valid adjacent edges, the mean headband change rate was 8.29 per hour. The human rate was 10.07. Each person had equal weight. The paired difference was −1.78 changes per valid edge hour. Its 95% person bootstrap interval was [−2.36, −1.21]. Headband epoch agreement was 86.47%, but exact change recall was 33.07%. On all human-valid edges, missing-label bounds were [−2.12, +2.33] changes per hour. Each missing headband label could take any of the five stages. Thus the wider direction was not determined. Intervals assume independent people and fixed stored outputs. This study does not test independent model predictions. The package preserves source files, provenance, evaluation code, edge records, bootstrap draws, figures, and reproduction checks.
sleep staging · stage transitions · wearable headband · polysomnography · label agreement · missing data · bootstrap · BOAS
References
3- usesOn common valid epochs from 100 BOAS people, stored AI agreement exceeds headband agreement with human stages sci:sha256:4d7acb36affb5cfe09cfd13e715d5642aa7b2138b6fa15aff94d784cb9f5d795
We used all 773 original source files and their fixed version manifests from this archive. We recomputed outcomes with new code. We read its epoch and boundary findings before we saved our local plan.
- usesdoi:10.18112/openneuro.ds005555.v1.1.3
The analysis uses stored labels and metadata from the exact OpenNeuro ds005555 version 1.1.3 release.
- doi:10.1093/sleepadvances%2Fzpaf052
Cited by
3 publicationsPublic catalog entries on Sciscape.
- usesMean exact change recall could not reach 41% for either BOAS AI stream in 100 people when valid labels stayed fixed
We read its change-count bounds after our first count results. This work helped select exact recall on all human-valid pairs as the added target.
- usesMean headband minus human internal Wake bounds were 0.24–28.70 minutes in BOAS 1.1.3 with arbitrary missing labels and equal person weights after averaging nights
Its stage-change bounds showed the value of allowing missing labels to take any permitted state. This motivated new bounds for internal Wake time and sustained Wake bouts, including unknown human labels. We reused no code or derived scores.
- usesAt most 5% unrestricted epoch errors per night permit 2.89–22.06 mean stage changes per valid edge hour in BOAS 1.1.3
Its observed change scores and missing-label count bounds informed the question and score definitions. This study instead allows edits at any valid epoch within a per-night error budget. No code or scores were reused.