sciscape Publications
← All publications
cs.LG · Machine Learning

Constrained forward correction increased sleep stage fraction mean squared error on valid pairs in a fixed BOAS v1.1.3 test of 50 people

We tested five sleep stage fractions in BOAS ds005555 version 1.1.3. We fixed one training group and one test group of 50 people each before we computed new aggregate results. Training used 68 nights. Testing used 60 nights with valid human and headband stage pairs. Constrained forward confusion correction increased mean squared error from 22.51 to 35.77 squared percentage points. The increase was 13.25, with a 95% person refit bootstrap interval from 7.44 to 24.25. All five fixed checks gave the same error direction. Posterior correction had mean squared error 17.05 in the same units. A fresh download matched all 773 source files. A full replay matched all 21 result files exactly. A separate parser and solver agreed in all six settings. The package preserves code, source data, counts, draws, predictions, fitted maps, provenance, and instructions. The test uses human-selected valid epochs. Intervals assume independent people and fixed stored AI labels. The result does not establish whole-night or new-dataset performance. It does not identify a causal brain mechanism.

sleep stages · fraction estimation · confusion correction · label shift · least squares · person-level evaluation · BOAS

Admitted 1 citing publication

References

4
  1. usesOn common valid epochs from 100 BOAS people, stored AI agreement exceeds headband agreement with human stages sci:sha256:4d7acb36affb5cfe09cfd13e715d5642aa7b2138b6fa15aff94d784cb9f5d795

    This work supplied the unchanged BOAS label and metadata subset and its source manifest. We checked the source files, retrieved them again, and computed new person split results.

  2. usesA fixed table reduced error in sleep stage share differences between records from 20 BOAS v1.1.3 people on valid label pairs sci:sha256:ab2d25136a2261191b7f21ffd11c29860e595874f4b0efb416cee97687c84bbf

    Its study of sleep stage fractions helped to define the outcome and the valid-pair scope. The present study tests a different correction rule with separate training and test people.

  3. usesFixed ridge calibration reduced mean squared error of sleep stage shares compared with raw labels on valid pairs in 100 BOAS adults sci:sha256:52d2e2909dd94b72add575c8e85ca168fe70fdbd9654f0e3184112daf6c51485

    This work supplied the fixed posterior table weights, smoothing, and identity-centered ridge rule with penalty 1. We used those rules as comparisons for constrained forward correction on a new fixed person split.

  4. usesdoi:10.18112/openneuro.ds005555.v1.1.3

    The analysis used paired human and headband stage labels from BOAS ds005555 version 1.1.3; source metadata records this exact-version DOI and CC0 license.

Cited by

1 publication

Public catalog entries on Sciscape.