Technical challenges in <scp>REM</scp> sleep microstructure classification: A study of patients with <scp>REM</scp> sleep behaviour disorder
Bibliographic record
Abstract
While commonly treated as a uniform state in practice, rapid eye movement sleep contains two distinct microstructures-phasic (presence of rapid eye movement) and tonic (no rapid eye movement). This study aims to identify technical challenges during rapid eye movement sleep microstructure visual classification in patients with rapid eye movement sleep behaviour disorder, and to propose solutions to enhance reliability between scorers. Fifty-seven sleep recordings were randomly allocated into three subsequent batches (n = 10, 13 and 34) for scoring. To reduce single-centre bias, we recruited three raters/scorers, with each trained from a different institution. Two raters independently scored each 30-s rapid eye movement sleep into 10 × fSEM3-s phasic/tonic microstructures based on the AASM guidelines. The third rater acted as an "arbitrator" to resolve opposite opinions persisting during the revision between batches. Besides interrater differences in artefact rejection rate, interrater variance frequently occurred due to transitioning between microstructures and moderate-to-severe muscular/electrode artefact interference. To enhance interrater agreement, a rapid eye movement scoring schematic graph was developed, incorporating proxy electrode use, filters and cut-offs for microstructure transitioning. To assess potential effectiveness of the schematic graph proposed, raters were instructed to systematically apply it in scoring for the third batch. Of the 34 recordings, 27 reached a Cohen's kappa score above 0.8 (i.e. almost perfect agreement between raters), significantly improved from the prior batches (p = 0.0003, Kruskal-Wallis test). Our study illustrated potential solutions and guidance for challenges that may be encountered during rapid eye movement sleep microstructure classification.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.004 | 0.006 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.003 | 0.004 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.000 | 0.004 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".