Performance of cardiorespiratory-based sleep staging in patients using beta blockers

Area of Science:

• Cardiorespiratory signal analysis
• Sleep medicine
• Autonomic nervous system function

Background:

• Home sleep monitoring offers potential for automatic sleep staging.
• Machine learning algorithms show promise for sleep disorder diagnosis.
• The impact of medications like beta blockers on autonomic physiology and sleep staging is not well understood.

Purpose of the Study:

• To evaluate the performance of an existing machine learning sleep staging algorithm (CReSS) in patients with sleep disorders who are using beta blockers compared to those who are not.
• To determine if beta blocker use impacts the accuracy of cardiorespiratory-based sleep staging.

Main Methods:

• Retrospective analysis of sleep recordings from 114 patients with sleep disorders (57 using beta blockers, 57 controls).
• Sleep stages were automatically scored using the CReSS algorithm based on electrocardiography and respiratory effort.
• Algorithm performance was compared to manual polysomnography using epoch-by-epoch agreement and derived sleep parameters.

Main Results:

• The CReSS algorithm achieved substantial agreement for four-class sleep staging in both groups (beta blockers: kappa = 0.635, accuracy = 78.1%; controls: kappa = 0.660, accuracy = 78.8%).
• No statistically significant difference in performance was observed between patients using beta blockers and the control group.
• Agreement for derived sleep parameters also did not differ between the two groups.

Conclusions:

• The CReSS algorithm's performance is not compromised in patients using beta blockers.
• Cardiorespiratory-based sleep staging using machine learning is a viable approach for this patient population.
• Autonomic characteristics can be reliably used for surrogate sleep measurement even with beta blocker medication.