Scorer adjustment changes linear age estimates for N2 transitions to wake or N1 over eight hours in 153 Sleep-EDF nights
We tested age and N2 transitions to wake or N1 in 153 Sleep-EDF nights from 78 people. Source ages were 25 to 101. We used the first eight hours after lights off. The linear age model included sex, night, bout position, and sleep time. Without scorer codes, the odds ratio per ten years was 1.066 (95% interval: 1.027 to 1.106). With scorer codes, it was 1.000 (0.959 to 1.042). We used 2,000 bootstrap draws of people. A paired comparison supported a change between the model estimates. This does not prove scorer bias or no age effect. An age audit found eight conflicts in files from five people. The primary estimate changed little under two age checks. The package contains all source annotations, verified counts, code, hashes, and run instructions. A fresh download reproduced all 180 source, result, and figure files exactly. The result applies to the source labels, selected sample, and stated model.
sleep transitions · N2 sleep · sleep scoring · age · scorer adjustment · Sleep-EDF · logistic regression · bootstrap
References
4- usesdoi:10.13026/C2X676
Sleep-EDF Expanded is the source dataset used for all 153 cassette hypnograms and the subject table; the exact source URLs and hashes are retained in original/provenance/sources.json.
- doi:10.1016/j.neurobiolaging.2012.05.018
- doi:10.1093/sleep%2Fzsx117
- doi:10.1038/s44360-026-00096-z
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesThe bootstrap gain interval included zero when first-night N2 exit rates entered fixed ridge models with age, sex, and scorer codes for 75 Sleep-EDF pairs
This publication supplied the Sleep-EDF source annotation files, source hash pins, and eight-hour N2 transition definition. Its extraction code and count table supported independent count checks. We use these inputs for a new paired-night prediction test.
- usesWith a 250 microvolt range limit, stage 2 gives a less negative linear age slope for frontal 0.5–4 Hz EEG power in 78 Sleep-EDF adults
That publication supplied the preserved source labels, subject table, source checksum list, and age conflict audit. These files enabled checked metadata rebuilding. We retrieved new EEG signal ranges and wrote new code.