ECG-based estimation of respiratory modulation of AV nodal conduction during atrial fibrillation

TL;DR

This study introduces a novel ECG-based method using deep learning to assess respiratory modulation of AV nodal conduction during atrial fibrillation, potentially aiding personalized treatment.

Contribution

A new approach combining synthetic data and deep learning to estimate respiratory modulation of AV nodal properties from ECG signals during AF.

Findings

Deep learning predicts respiratory modulation with high correlation ($ ho$ = 0.855) using ECG data.

Respiratory modulation varies significantly among patients during deep breathing.

Synthetic data effectively trains models for clinical ECG analysis.

Abstract

Information about autonomic nervous system (ANS) activity may be valuable for personalized atrial fibrillation (AF) treatment but is not easily accessible from the ECG. In this study, we propose a new approach for ECG-based assessment of respiratory modulation in AV nodal refractory period and conduction delay. A 1-dimensional convolutional neural network (1D-CNN) was trained to estimate respiratory modulation of AV nodal conduction properties from 1-minute segments of RR series, respiration signals, and atrial fibrillatory rates (AFR) using synthetic data that replicates clinical ECG-derived data. The synthetic data were generated using a network model of the AV node and 4 million unique model parameter sets. The 1D-CNN was then used to analyze respiratory modulation in clinical deep breathing test data of 28 patients in AF, where a ECG-derived respiration signal was extracted using a novel approach based on periodic component analysis. We demonstrated using synthetic data that the 1D-CNN can predict the respiratory modulation from RR series alone ($\rho$ = 0.805) and that the addition of either respiration signal ($\rho$ = 0.830), AFR ($\rho$ = 0.837), or both ($\rho$ = 0.855) improves the prediction. Results from analysis of clinical ECG data of 20 patients with sufficient signal quality suggest that respiratory modulation decreased in response to deep breathing for five patients, increased for five patients, and remained similar for ten patients, indicating a large inter-patient variability.