Propagation pattern analysis during atrial fibrillation based on the adaptive group LASSO

Area of Science:

• Biomedical Engineering
• Signal Processing
• Computational Neuroscience

Background:

• Atrial fibrillation (AF) propagation patterns are complex and challenging to estimate.
• Current methods for analyzing intracardiac AF signals have limitations in accuracy and efficiency.
• Accurate estimation of propagation patterns is crucial for understanding AF mechanisms and developing treatments.

Purpose of the Study:

• To introduce a novel sparse modeling approach for estimating intracardiac AF signal propagation patterns.
• To evaluate the performance of the proposed adaptive group least absolute selection and shrinkage operator (aLASSO) method.
• To compare aLASSO with traditional least-squares (LS) estimation.

Main Methods:

• Utilized partial directed coherence (PDC) function derived from multivariate autoregressive (MVAR) models.
• Developed and applied the adaptive group least absolute selection and shrinkage operator (aLASSO) for sparse parameter estimation.
• Conducted simulations and analyzed real intracardiac AF data.

Main Results:

• aLASSO demonstrated superior estimation performance compared to LS estimation in simulations.
• Normalized error decreased from 0.20 ± 0.04 (LS) to 0.03 ± 0.01 (aLASSO) when data samples exceeded model parameters by fivefold.
• Error reduction was more significant for shorter data segments.
• Sparsity assumption simplified the identification of propagation patterns in intracardiac AF data.

Conclusions:

• Sparse modeling with aLASSO offers a robust and accurate method for estimating AF propagation patterns.
• aLASSO provides significant improvements in estimation accuracy, especially with limited data.
• The proposed method simplifies the analysis of complex AF dynamics, aiding further research and clinical applications.