Automatic Sleep Staging from CPAP Airflow using a Dual Fusion Multi-Period Convolutional Neural Network
Bibliographic record
Abstract
Abstract Background : Continuous positive airway pressure (CPAP) therapy is the standard treatment for obstructive sleep apnea-hypopnea syndrome, yet its use as a passive sleep dynamics monitoring remains limited. CPAP devices record only airflow signals, called CPAP-flow, which constantly interact with the pressure delivered by the device. This interaction renders the signal vulnerable to device-related artifacts and inter-/intra-patient variability, posing significant challenges to its repurposing for monitoring sleep dynamics. Methods : Motivated by neural network-based studies of sleep-wake transition from CPAP-flow, we leverage existing physiological knowledge and introduce a dual fusion multi-period convolutional neural network (DFMP-CNN) model. This deep learning architecture leverages multiple period-specific convolutional kernels and a dual-fusion mechanism to jointly encode known short- and long-range temporal dependencies in CPAP-flow, overcoming limitations of traditional fixed-scale models. Results : Extensive experiments demonstrate that DFMP-CNN achieves state-of-the-art performance in CPAP-based sleep staging. On the Yale dataset, it achieves 78.5% accuracy (Cohen’s κ = 0.605), with a best case of κ = 0.886; on the Duke dataset, it reaches 73.6% accuracy ( κ = 0.524), with a best case of κ = 0.805. Cross-dataset evaluations confirm the model’s transferability across clinical centers and device types, while feature and fusion ablation studies highlight its robustness. Conclusions : DFMP-CNN provides an unobtrusive approach for sleep monitoring using CPAP devices, providing a dual-purpose platform for therapy and longitudinal assessment. Significance: the robust performance of DFMP-CNN across datasets and device types underscores its potential to improve clinical assessment and optimize therapy management.
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 machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
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 source (direct Gemma or distilled Codex), 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".