Wenckebach's periods or alternating Wenckebach's periods during 4:1 atrioventricular block?

Expert-enhanced machine learning for cardiac arrhythmia classification

We propose a new method for the classification task of distinguishing atrial fibrillation (AFib) from regular atrial tachycardias including atrial flutter (AFlu) based on a surface electrocardiogram (ECG). Recently, many approaches for an automatic classification of cardiac arrhythmia were proposed and to our knowledge none of them can distinguish between these two. We discuss reasons why deep learning may not yield satisfactory results for this task. We generate new and clinically interpretable features using mathematical optimization for subsequent use within a machine learning (ML) model. These features are generated from the same input data by solving an additional regression problem with complicated combinatorial substructures. The resultant can be seen as a novel machine learning model that incorporates expert knowledge on the pathophysiology of atrial flutter. Our approach achieves an unprecedented accuracy of 82.84% and an area under the receiver operating characteristic (ROC) curve of 0.9, which classifies as "excellent" according to the classification indicator of diagnostic tests. One additional advantage of our approach is the inherent interpretability of the classification results. Our features give insight into a possibly occurring multilevel atrioventricular blocking mechanism, which may improve treatment decisions beyond the classification itself. Our research ideally complements existing textbook cardiac arrhythmia classification methods, which cannot provide a classification for the important case of AFib↔AFlu. The main contribution is the successful use of a novel mathematical model for multilevel atrioventricular block and optimization-driven inverse simulation to enhance machine learning for classification of the arguably most difficult cases in cardiac arrhythmia. A tailored Branch-and-Bound algorithm was implemented for the domain knowledge part, while standard algorithms such as Adam could be used for training.

Nodal recovery, dual pathway physiology, and concealed conduction determine complex AV dynamics in human atrial tachyarrhythmias

The genesis of complex ventricular rhythms during atrial tachyarrhythmias in humans is not fully understood. To clarify the dynamics of atrioventricular (AV) conduction in response to a regular high-rate atrial activation, 29 episodes of spontaneous or pacing-induced atrial flutter (AFL), covering a wide range of atrial rates (cycle lengths from 145 to 270 ms), were analyzed in 10 patients. AV patterns were identified by applying firing sequence and surrogate data analysis to atrial and ventricular activation series, whereas modular simulation with a difference-equation AV node model was used to correlate the patterns with specific nodal properties. AV node response at high atrial rate was characterized by 1) AV patterns of decreasing conduction ratios at the shortening of atrial cycle length (from 236.3 ± 32.4 to 172.6 ± 17.8 ms) according to a Farey sequence ordering (conduction ratio from 0.34 ± 0.12 to 0.23 ± 0.06; P < 0.01); 2) the appearance of high-order alternating Wenckebach rhythms, such as 6:2, 10:2, and 12:2, associated with ventricular interval oscillations of large amplitude (407.7 ± 150.4 ms); and 3) the deterioration of pattern stability at advanced levels of block, with the percentage of stable patterns decreasing from 64.3 ± 35.2% to 28.3 ± 34.5% ( P < 0.01). Simulations suggested these patterns to originate from the combined effect of nodal recovery, dual pathway physiology, and concealed conduction. These results indicate that intrinsic nodal properties may account for the wide spectrum of AV block patterns occurring during regular atrial tachyarrhythmias. The characterization of AV nodal function during different AFL forms constitutes an intermediate step toward the understanding of complex ventricular rhythms during atrial fibrillation.

Rhythms of high-grade block in an ionic model of a strand of regionally ischemic ventricular muscle

Electrical alternans, a beat-to-beat alternation in the electrocardiogram or electrogram, is frequently seen during the first few minutes of acute myocardial ischemia, and is often immediately followed by malignant cardiac arrhythmias such as ventricular tachycardia and ventricular fibrillation. As ischemia progresses, higher-order periodic rhythms (e.g., period-4) can replace the period-2 alternans rhythm. This is also seen in modelling work on a two-dimensional (2-D) sheet of regionally ischemic ventricular muscle. In addition, in the experimental work, ventricular arrhythmias are overwhelmingly seen only after the higher-order rhythms arise. We investigate an ionic model of a strand of ischemic ventricular muscle, constructed as a 3-cm-long 1-D cable with a centrally located 1-cm-long segment exposed to an elevated extracellular potassium concentration ([K(+)](o)). As [K(+)](o) is raised in this "ischemic segment" to represent one major effect of ongoing ischemia, the sequence of rhythms {1:1-->2:2 (alternans)-->2:1} is seen. With further increase in [K(+)](o), one sees higher-order periodic 2N:M rhythms {2:1-->4:2-->4:1-->6:2-->6:1-->8:2-->8:1}. In a 2N:M cycle, only M of the 2N action potentials generated at the proximal end of the cable successfully traverse the ischemic segment, with the remaining ones being blocked within the ischemic segment. Finally, there is a transition to complete block {8:1-->2:0-->1:0} (in an n:0 rhythm, all action potentials die out within the ischemic segment). Changing the length of the ischemic segment results in different rhythms and transitions being seen: e.g., when the ischemic segment is 2 cm long, the period-6 rhythms are not seen; when it is 0.5 cm long, there is a 3:1 rhythm interposed between the 2:1 and 1:0 rhythms. We discuss the relevance of our results to the experimental observations on the higher-order rhythms that presage reentrant ischemic ventricular arrhythmias.