Automatic Sleep Staging Using Cardiorespiratory Signals: A Systematic Review of Methodologies and Performance

Clinical Snapshot

Key Findings

• P Value: Not reported for primary accuracy comparisons; significance of modality and algorithm comparisons described narratively without formal statistical testing parameters

• Effect Size: Overall accuracy of approximately 70% across included studies; no statistically significant performance differences observed between signal modalities (cardiac alone, respiratory alone, combined cardiorespiratory) or between traditional machine learning and deep learning algorithms

• Primary Outcome: Overall automatic sleep staging accuracy using cardiorespiratory signals compared to PSG reference standard across 35 included studies (2010–present)

• Nnt Or Sensitivity: Per-stage sensitivity and specificity not reported in abstract; N1 sleep stage identified as consistently and significantly underperforming across all modalities and algorithms — a clinically important finding given N1's role in sleep architecture assessment. No NNT calculable from available data.

• Confidence Interval: Not reported; absence of confidence intervals is a significant methodological limitation

Clinical Application

Cardiorespiratory-based sleep staging using wearable or ambulatory devices is technically feasible and increasingly accessible. However, clinical deployment is premature given the lack of external validation, significant heterogeneity in methodologies, and consistently poor N1 classification. Implementation would require regulatory-grade validation studies, standardised signal acquisition protocols, and integration with existing clinical workflows. Cost-effectiveness data are absent from the current evidence base. In Australia, full-attended PSG remains the gold standard for sleep disorder diagnosis and is Medicare Benefits Schedule (MBS) reimbursable (MBS items 12203–12215 for attended PSG). Home sleep apnoea testing (HSAT) using cardiorespiratory monitoring is already partially reimbursed under MBS item 12250 for suspected OSA, but is limited to respiratory event detection rather than full sleep staging. The TGA would require prospective clinical validation data meeting IVD regulatory standards before cardiorespiratory sleep staging software could be approved as a Class IIb or III medical device. RACGP guidelines on sleep health management emphasise PSG as the diagnostic reference standard; cardiorespiratory staging tools could complement but not currently replace this pathway. The Australian context of geographic remoteness and limited sleep laboratory access in rural and regional areas makes scalable cardiorespiratory monitoring particularly relevant for future development, contingent on robust validation. Adults with suspected sleep disorders (obstructive sleep apnoea, insomnia, circadian rhythm disorders) who may benefit from longitudinal or ambulatory sleep monitoring where full PSG is impractical, resource-limited, or not immediately accessible. Caution is warranted for paediatric populations, patients with significant cardiorespiratory comorbidities, and those requiring precise sleep stage differentiation for clinical decision-making.

Abstract

Cardiorespiratory-based methods offer promising alternatives to traditional PSG for longitudinal sleep monitoring, holding significant systemic medical value for scalable sleep health management. This systematic review synthesizes methodological frameworks and performance outcomes of automatic sleep staging using cardiorespiratory signals. Four databases were searched and a total of 35 studies published since 2010 were identified. The analysis revealed that cardiorespiratory signal-based sleep staging achieved a practically meaningful accuracy of 70%, with no significant performance differences observed among signal modalities (cardiac signals, cardiorespiratory signals, or cardiac/cardiorespiratory signals combined with other non-EEG modalities) or between modeling algorithms (traditional machine learning vs. deep learning). However, we identified significant methodological heterogeneity and several critical model failure modes that hinder clinical translation, including the widespread lack of external validation, consistently poor classification of the N1 sleep stage, and limited generalization across diverse patient populations. To realize the technology's potential, future research must establish consensus-driven methodological guidelines and rigorously validate algorithms on large, demographically and clinically diverse datasets. These advances are essential for integrating cardiorespiratory-based sleep staging into healthcare systems as a scalable tool for population-level screening, longitudinal monitoring, and tiered clinical decision support.

References

1. 1.Chen, W., He, X., Zheng, J., Chen, S., & Tian, X. (2026). Automatic sleep staging using cardiorespiratory signals: A systematic review of methodologies and performance. Journal of Medical Systems. https://doi.org/10.1016/j.smrv.2020.101377