| Title |
Apnea burden-guided framework: enhancing out-of-distribution generalization in PPG-based sleep apnea characterization |
| Authors |
Rinkevičius, Mantas ; Pfeffer, Oskar ; Alissa, Amal ; Hegemann, Nando ; Marozas, Vaidotas |
| DOI |
10.1109/ACCESS.2026.3727753 |
| Full Text |
|
| Is Part of |
IEEE Access.. Piscataway, NJ : IEEE. 2026, vol. 14, p. 133172-133187.. ISSN 2169-3536 |
| Keywords [eng] |
Apnea burden ; apnea-hypopnea index ; arterial blood oxygen saturation ; convolutional-recurrent networks ; home-based preventive monitoring ; out-of-distribution generalization ; photoplethysmographic features ; sleep apnea characterization |
| Abstract [eng] |
Background: Sleep apnea is a common sleep-related breathing disorder associated with substantial cardiovascular and metabolic risk. Although overnight polysomnography remains the reference standard for diagnosis, its complexity and cost limit its suitability for long-term preventive monitoring at home. In wearable systems, arterial blood oxygen saturation is commonly used as the main predictor, whereas additional morphological features of the photoplethysmographic pulse wave are usually underexplored. Objective: This study proposes a novel apnea burden prediction-based framework for sleep apnea severity assessment and investigates the influence of photoplethysmographic features on model performance and out-of-distribution generalization. Methods: The proposed framework first predicts apnea burden as a continuous measure, which is subsequently converted into the clinically relevant apnea-hypopnea index for subject-level classification into four severity groups in both in-distribution and out-of-distribution data. Three artificial neural network architectures were evaluated, and the performance metrics were averaged over five independent runs with different fixed random seeds. Results: During out-of-distribution testing, the combination of photoplethysmographic features and arterial blood oxygen saturation led to increases of approximately 15.72% in macro-sensitivity, 9.22% in macro-accuracy, 16.01% in macro-F1-score, 11.08% in Cohen’s kappa, and 13.22% in Matthews correlation coefficient, compared with using arterial blood oxygen saturation alone. The low-complexity convolutional-recurrent models achieved the highest overall performance. Conclusion: The results indicated that the proposed apnea burden-guided framework, combined with photoplethysmographic features and arterial blood oxygen saturation, improves sleep apnea characterization while showing encouraging out-of-distribution performance on an independent external dataset. Moreover, simpler hybrid architectures demonstrated strong potential for robust home-based preventive monitoring. |
| Published |
Piscataway, NJ : IEEE |
| Type |
Journal article |
| Language |
English |
| Publication date |
2026 |
| CC license |
|
