BOAS 1.1.3 headband labels had at least 28.49 minutes of mean absolute longest sleep bout error in 100 people when only missing labels could change
We tested the longest sleep bout in 128 BOAS 1.1.3 recordings from 100 people. We kept the complete 30-second grid. Each missing human or AI label could be sleep or Wake. Every known label stayed fixed. We took absolute headband error within each recording against the human reference. We averaged recordings within people, then gave people equal weight. The exact mean absolute error bounds were 28.4991667 and 40.3116667 minutes. The signed mean difference bounds were 10.8725 and 23.7925 minutes. Thus, choices for missing labels could not remove the score error or the positive mean difference in this sample. The interval method found every possible score and preserved gaps between possible scores. A separate finite-state method checked all 384 stream score sets. Full completions reached all four primary endpoints in all 128 recordings. A fresh source download reproduced all eight analysis files byte for byte. All planned checks retained substantial headband error. The package contains source bytes, fixed URLs, hashes, the local plan, code, proof, completions, figures, and reproduction instructions. The result concerns stored labels and human consensus. It does not establish true sleep physiology or performance on new people.
sleep continuity · longest sleep bout · missing labels · exact bounds · polysomnography · wearable EEG · BOAS
References
5- usesStored headband wake coverage was lower for 30 to 60 second than for at least 120 second internal human wake bouts in 87 BOAS people sci:sha256:189bb9f97d15f6eed6349973a60c2ec7eb75375ac6b2221a26edcbd1048fa1ce
Its brief Wake bout coverage study motivated a direct test of the longest continuous sleep bout. We read its description before the local plan. We did not reuse its scoring code.
- usesOn common valid epochs from 100 BOAS people, stored AI agreement exceeds headband agreement with human stages sci:sha256:4d7acb36affb5cfe09cfd13e715d5642aa7b2138b6fa15aff94d784cb9f5d795
We retrieved its BOAS source label files, source tree, version mapping, and retrieval code. We used that code without change to download all 773 selected files again. The new longest sleep bout solver and results are separate.
- usesFixed known labels left mean absolute headband error above 13.60 minutes for sustained REM time in 100 BOAS 1.1.3 people with a 10-minute bout rule sci:sha256:3d73288a9d2435138a0e23b57118396249e71b9baa1f76ce3e5eeb7698d2eb34
Its exact sustained REM bounds study motivated full possible score sets, completion witnesses, and separate feasibility checks. We read its description before the local plan. We developed a different interval solver for longest binary runs and did not reuse its scoring code.
- usesdoi:10.18112/openneuro.ds005555.v1.1.3
This is the exact OpenNeuro BOAS dataset version analyzed; the source files were fixed to the recorded Git commit and checked by hashes.
- doi:10.1371/journal.pmed.1003782
Cited by
0 publicationsPublic catalog entries on Sciscape.
No incoming citations on Sciscape.