Quantifying Respiration Effects on Cardiac Vibrations using Teager Energy Operator and Gradient Boosted Trees

Area of Science:

• Biomedical Engineering
• Cardiovascular Physiology
• Signal Processing

Background:

• Seismocardiogram (SCG) signals reflect cardiac mechanical activity.
• Respiratory variations can influence SCG morphology, impacting cardiovascular parameter analysis.
• Standardization of SCG signals is crucial for reliable clinical interpretation.

Purpose of the Study:

• To develop a novel beat scoring system for quantifying respiratory phase effects (exhalation/inhalation) on SCG signals.
• To assess the system's performance under rest and physiologically modulated conditions (classical music, recovery).
• To investigate the impact of physiological modulations on SCG morphology and respiration interaction.

Main Methods:

• Segmentation of SCG and ECG signals into exhalation and inhalation phases using respiration data.
• Construction of representative SCG beats for each respiratory phase, synchronized with ECG R-peaks.
• Extraction of SCG beat differences using the Teager-Kaiser energy operator.
• Development of a gradient-based beat scoring system employing extreme gradient boosted trees and monotonic mapping.

Main Results:

• High accuracy (AUC 0.978-0.985) achieved in distinguishing respiratory phases during rest and recovery.
• Moderate overlap (14.2%-41.2%) in score distributions between inhalation and exhalation phases across conditions.
• Demonstrated that physiological modulations alter the respiration-SCG interaction, affecting signal morphology.

Conclusions:

• The developed beat scoring system effectively quantifies respiratory influences on SCG.
• Standardization of SCG beats is essential for accurate cardiovascular monitoring, especially considering respiratory variations.
• The system offers potential for improved clinical SCG analysis by accounting for intra-recording variability.