At most 5% unrestricted epoch errors per night permit 2.89–22.06 mean stage changes per valid edge hour in BOAS 1.1.3
We tested fixed human sleep stage labels from 128 nights in 100 people in BOAS 1.1.3. Each night could have wrong labels at no more than 5% of valid epochs. Error locations and wrong stages had no other limits. An integer dynamic program gave exact attained count limits. The mean reference rate was 10.18 changes per valid edge hour. Permitted mean rates ranged from 2.89 to 22.06. Each person had equal weight, with equal night weights within each person. Minimum mean exact binary change recall was 17.39%. Minimum mean directed stage pair recall was 11.82%. Saved label sequences reached every endpoint within budget. These are fixed-record limits under unrestricted errors, not expected model results or confidence intervals. The package contains source records, provenance, proof, code, all count limits, attaining labels, figures, and reproduction checks.
sleep staging · stage transitions · agreement · error budget · dynamic programming · robustness audit · BOAS · OpenNeuro
References
5- usesMean exact change recall could not reach 41% for either BOAS AI stream in 100 people when valid labels stayed fixed sci:sha256:3997f141e63a53ec61bd47513d82542c2e408a52c33baf5cce9d8f6679483ca9
We read its summary after the local plan to compare the scopes. Its recall bounds change only missing AI labels. Our bounds permit errors at any valid epoch within a fixed budget. This reading did not change the analysis.
- usesOn common valid epochs from 100 BOAS people, stored AI agreement exceeds headband agreement with human stages sci:sha256:4d7acb36affb5cfe09cfd13e715d5642aa7b2138b6fa15aff94d784cb9f5d795
We used 389 original source files and fixed source provenance from this archive. A fresh direct download matched them. We used no prior code or derived scores.
- usesIn BOAS 1.1.3, stored headband labels had a lower mean change rate than human labels on common valid edges in 100 people sci:sha256:4cdd91247e1cc665001cb799c7431094be7c99b0f01b29f238ed6cf0e4d6ee2d
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.
- usesdoi:10.18112/openneuro.ds005555.v1.1.3
The analysis uses the human sleep stage labels and source records in OpenNeuro BOAS ds005555 version 1.1.3.
- doi:10.1093/sleepadvances%2Fzpaf052
Cited by
0 publicationsPublic catalog entries on Sciscape.
No incoming citations on Sciscape.