Time-structured detection of pre-event cardiovascular instability using CPAP-derived Cheyne-Stokes breathing

Area of Science:

• Cardiology
• Respiratory Medicine
• Biomedical Engineering

Background:

• Continuous Positive Airway Pressure (CPAP) telemonitoring generates daily respiratory metrics, such as Cheyne-Stokes breathing percentage (CSB%), which indicate ventilatory-circulatory instability.
• Fixed thresholds for CSB% have limited clinical use due to significant inter-individual variability.

Purpose of the Study:

• To develop and evaluate a time-structured detection architecture for CPAP telemonitoring data.
• To identify temporal patterns preceding cardiovascular and cerebrovascular events using this architecture.

Main Methods:

• A retrospective observational study analyzed 1265 patients with obstructive sleep apnea undergoing CPAP telemonitoring.
• A detection framework combined sustained baseline deviations and abrupt pre-event surges, incorporating central apnea index for hierarchical filtering.
• Daily CSB% values were smoothed and evaluated against individualized dynamic baselines.

Main Results:

• The architecture detected 23 out of 25 cardiovascular and cerebrovascular events (92.0%) within a D-28 window, with a median lead time of 12.0 days.
• Hierarchical filtering significantly reduced alerts (94.9%) while maintaining event detection.
• Distinct temporal phenotypes were observed across different disease categories.

Conclusions:

• CPAP-derived CSB% can serve as a time-structured digital biomarker for evolving ventilatory-circulatory instability.
• A trajectory-based detection architecture shows promise for early cardiovascular instability detection in large-scale telemonitoring.