On Tuesday, 12 November 2024, our colleague Mattias Karlsson presented and defended his thesis for the degree of Doctor of Philosophy ”Model-based Analysis of Individual Atrioventricular Node Conduction Dynamics During Atrial Fibrillation”
Abstract
Atrial fibrillation (AF) is the most common arrhythmia in the world, leading to a significant burden to patients and the healthcare system. It is characterised by rapid and irregular atrial contractions stemming from disorganised electrical activity in the atria. The atrioventricular (AV) node regulates heart rate during AF by filtering electrical impulses from the atria.
However, for persistent AF, the regulating capabilities of the AV node are often insufficient in regards to maintaining a healthy heart rate. Thus, rate control drugs affecting the conduction properties of the AV node are the most common treatment, chosen empirically for each patient. This takes time and may result in a sub-optimal drug choice. Quantifying individual differences in AV-nodal function is therefore interesting in order to potentially aid in personalised treatment selection.
This thesis focuses on assessing the conduction properties of the AV node during AF from electrocardiography recordings, specifically the refractory period and conduction delay. The thesis comprises an introduction to the anatomy of the heart, AF, cardiac modelling, and parameter estimation, as well as four papers. The first paper proposes a mathematical model of the AV node where the model parameters could be estimated from 15-minute ECG recordings utilising a genetic algorithm. In the second paper, we used the proposed model and introduced a computationally efficient dynamic genetic algorithm to enable estimation of 24-hour model parameter trends, with a temporal resolution of one estimate per 1000 RR intervals, to analyse individual and drug-dependent differences in the model parameters. In the third paper, the optimisation framework was further extended to combine an Approximate Bayesian computation algorithm with the previously proposed genetic algorithm in order to quantify the uncertainty of the model parameter estimates. Additionally, a model parameter reduction step was introduced to increase interpretability of the results. In iii iv Abstract the fourth paper an improved optimisation framework consisting of a particle filter and an associated smoothing algorithm enabling beat-to-beat temporal resolution was proposed. This temporal resolution allows for analysis of beat-to-beat changes in the AV node conduction properties induced by the autonomic nervous system.
All-in-all, the work presented in this thesis has made it possible for the first time to assess the conduction properties of the AV node during AF based on ECG measurements.