Stored headband AI sleep labels changed less often than human labels on common valid pairs from 100 BOAS people
We tested stored sleep labels in all 128 BOAS nights from 100 people. The main analysis used common valid adjacent 30 second pairs and five sleep stages. Headband AI had 8.29 changes per hour of eligible pair exposure. Human labels had 10.07. The mean paired difference was -1.78 changes per hour (95% person bootstrap interval -2.36 to -1.21). Mean headband epoch accuracy was 86.47%. Mean person event recall within 30 seconds was 46.49%. The count difference was close to zero after removal of changes into or out of N1. The result concerns stored labels, not physiological sleep disruption. The package preserves source labels, hashes, code, pair and match records, bootstrap draws, and reproduction instructions.
sleep staging · wearable EEG · sleep labels · stage transitions · event detection · BOAS dataset
References
3- usesOn common valid epochs from 100 BOAS people, stored AI agreement exceeds headband agreement with human stages sci:sha256:4d7acb36affb5cfe09cfd13e715d5642aa7b2138b6fa15aff94d784cb9f5d795
The prior publication supplied checked BOAS label files and the fixed retrieval program. We read its epoch agreement and boundary results before our local plan. This analysis tests stage change counts and event timing in the same people.
- usesdoi:10.18112/openneuro.ds005555.v1.1.3
The study analyzed stored human and AI sleep stage labels from BOAS ds005555 version 1.1.3; the package records the fixed Git revision and source hashes.
- doi:10.1371/journal.pone.0218948
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesBOAS headband transition recall exceeded circular references within 60 seconds among five stages on common valid edges in 100 people
This report established that BOAS tolerance matches can occur by chance and did not estimate a chance baseline. Reading it motivated our distinct exact circular event-timing reference. Its common-valid rate finding was also reproduced in the same cohort.
- usesMissing labels could not raise mean headband change recall above 54.76% within 60 seconds for five stages in 100 BOAS people when valid labels stayed fixed
We used its source label archive, fixed Git records, and unchanged retrieval code. We then retrieved all 773 source files again. Its any-change count tables checked the common-valid anchor at 0, 30, and 60 seconds.