Feature subset selection and classification of intracardiac electrograms during atrial fibrillation

A Novel Approach to Dual Feature Selection of Atrial Fibrillation Based on HC-MFS

This investigation sought to discern the risk factors for atrial fibrillation within Shanghai's Chongming District, analyzing data from 678 patients treated at a tertiary hospital in Chongming District, Shanghai, from 2020 to 2023, collecting information on season, C-reactive protein, hypertension, platelets, and other relevant indicators. The researchers introduced a novel dual feature-selection methodology, combining hierarchical clustering with Fisher scores (HC-MFS), to benchmark against four established methods. Through the training of five classification models on a designated dataset, the most effective model was chosen for method performance evaluation, with validation confirmed by test set scores. Impressively, the HC-MFS approach achieved the highest accuracy and the lowest root mean square error in the classification model, at 0.9118 and 0.2970, respectively. This provides a higher performance compared to existing methods, thanks to the combination and interaction of the two methods, which improves the quality of the feature subset. The research identified seasonal changes that were strongly associated with atrial fibrillation (pr = 0.31, FS = 0.11, and DCFS = 0.33, ranked first in terms of correlation); LDL cholesterol, total cholesterol, C-reactive protein, and platelet count, which are associated with inflammatory response and coronary heart disease, also indirectly contribute to atrial fibrillation and are risk factors for AF. Conclusively, this study advocates that machine-learning models can significantly aid clinicians in diagnosing individuals predisposed to atrial fibrillation, which shows a strong correlation with both pathological and climatic elements, especially seasonal variations, in the Chongming District.

Entropy Mapping Approach for Functional Reentry Detection in Atrial Fibrillation: An In-Silico Study

Catheter ablation of critical electrical propagation sites is a promising tool for reducing the recurrence of atrial fibrillation (AF). The spatial identification of the arrhythmogenic mechanisms sustaining AF requires the evaluation of electrograms (EGMs) recorded over the atrial surface. This work aims to characterize functional reentries using measures of entropy to track and detect a reentry core. To this end, different AF episodes are simulated using a 2D model of atrial tissue. Modified Courtemanche human action potential and Fenton–Karma models are implemented. Action potential propagation is modeled by a fractional diffusion equation, and virtual unipolar EGM are calculated. Episodes with stable and meandering rotors, figure-of-eight reentry, and disorganized propagation with multiple reentries are generated. Shannon entropy (ShEn), approximate entropy (ApEn), and sample entropy (SampEn) are computed from the virtual EGM, and entropy maps are built. Phase singularity maps are implemented as references. The results show that ApEn and SampEn maps are able to detect and track the reentry core of rotors and figure-of-eight reentry, while the ShEn results are not satisfactory. Moreover, ApEn and SampEn consistently highlight a reentry core by high entropy values for all of the studied cases, while the ability of ShEn to characterize the reentry core depends on the propagation dynamics. Such features make the ApEn and SampEn maps attractive tools for the study of AF reentries that persist for a period of time that is similar to the length of the observation window, and reentries could be interpreted as AF-sustaining mechanisms. Further research is needed to determine and fully understand the relation of these entropy measures with fibrillation mechanisms other than reentries.

Rethinking multiscale cardiac electrophysiology with machine learning and predictive modelling

We review some of the latest approaches to analysing cardiac electrophysiology data using machine learning and predictive modelling. Cardiac arrhythmias, particularly atrial fibrillation, are a major global healthcare challenge. Treatment is often through catheter ablation, which involves the targeted localised destruction of regions of the myocardium responsible for initiating or perpetuating the arrhythmia. Ablation targets are either anatomically defined, or identified based on their functional properties as determined through the analysis of contact intracardiac electrograms acquired with increasing spatial density by modern electroanatomic mapping systems. While numerous quantitative approaches have been investigated over the past decades for identifying these critical curative sites, few have provided a reliable and reproducible advance in success rates. Machine learning techniques, including recent deep-learning approaches, offer a potential route to gaining new insight from this wealth of highly complex spatio-temporal information that existing methods struggle to analyse. Coupled with predictive modelling, these techniques offer exciting opportunities to advance the field and produce more accurate diagnoses and robust personalised treatment. We outline some of these methods and illustrate their use in making predictions from the contact electrogram and augmenting predictive modelling tools, both by more rapidly predicting future states of the system and by inferring the parameters of these models from experimental observations.