Attractor Reconstruction of Breathing Dynamics: Characterising Respiratory Dysfunction in COPD

Area of Science:

• Biomedical Engineering
• Non-linear Dynamics
• Respiratory Medicine

Background:

• Chronic obstructive pulmonary disease (COPD) presents with fluctuating symptoms often missed between clinical visits.
• Continuous, objective monitoring is needed for effective COPD management.

Purpose of the Study:

• To investigate attractor-based phase-space reconstruction for characterizing respiratory dynamics in COPD.
• To identify novel digital biomarkers for COPD detection and monitoring.

Main Methods:

• Collected respiratory signal data using RESpeck accelerometers from 50 participants (18 COPD, 32 controls) over 2-4 weeks.
• Applied non-linear methods (attractor reconstruction) to derive 27 features from respiratory data.
• Analyzed features spanning geometric, spectral, recurrence, and complexity domains.

Main Results:

• Attractor-derived features enabled COPD classification with 84.4% accuracy.
• Identified heightened diurnal respiratory variability in COPD patients, especially during night-to-morning transitions.
• Demonstrated capture of pre-exacerbation changes missed by respiratory rate alone.

Conclusions:

• Attractor-derived features show potential as objective, high-resolution digital biomarkers for respiratory dysfunction.
• These biomarkers support passive, continuous monitoring for personalized COPD management.
• Non-linear analysis of wearable sensor data offers a promising avenue for COPD care.