Machine learning–guided electrophysiology: real time mapping, multimodal biomarkers, and translational tools for arrhythmia management

Electrophysiological mapping and monitoring generate complex datasets that can inform arrhythmia characterization, risk assessment, and treatment. High-density electroanatomical maps, intracardiac electrograms, surface ECGs, and ambulatory recordings provide complementary information about electrical activation and arrhythmogenic substrates. Machine learning (ML) offers opportunities to extract informative features, integrate different measurements, and support clinical decisions. Translating these approaches into electrophysiology practice requires attention to physiological relevance, data quality, reproducibility, and performance across recording conditions and patient populations. For real-time and intraprocedural applications, computational efficiency, uncertainty, and integration into clinical workflows are additional considerations. Studies that link model features and outputs to underlying electrophysiological mechanisms, and that pair methodological development with appropriate experimental or clinical validation, can help establish which approaches are suitable for further translation.This Research Topic emphasizes the translation of ML into electrophysiology practice, prioritizing real-time and intraprocedural applications. The goal is to establish how ML models and associated signal-processing tools improve mechanistic insight, procedural guidance, and patient-level decision making. Contributions should address a defined translational problem in mapping, biomarker interpretation, or arrhythmia management, including treatment selection and prediction of recurrence. Prospective clinical studies are encouraged. Retrospective analyses, experimental studies, and computational investigations are also welcome when they develop or validate methods for a specified clinical application. Evaluation should be appropriate to the study’s claims and stage of development, with a clear account of remaining requirements for clinical implementation. The collection aims to connect methodological innovation with evidence supporting clinical use.We welcome Original Research, Brief Research Report, Technology and Code, Review, Systematic Review, Hypothesis and Theory, and Perspective articles addressing one or more of the following themes: • Real-time mapping and catheter guidance• Automated substrate characterization and assessment of ablation effects• Discovery and validation of ECG- and electrogram-derived biomarkers• Multimodal integration of mapping, imaging, and longitudinal monitoring• Prediction of arrhythmia risk, treatment response, or recurrence• Interpretable models and uncertainty quantification• Hybrid approaches combining biophysical modeling with ML• Clinical validation or workflow integration. Contributions should identify the electrophysiological problem, explain the intended clinical use, link model features or outputs to underlying electrophysiological mechanisms, and critically evaluate the evidence supporting translation. Purely predictive models without mechanistic interpretation, generic ECG classification or image-segmentation algorithms, and signal-processing studies without a defined electrophysiological application fall outside the scope. Topic Editor Martin Manninger received financial support for research from Biosense Webster and Abbott (ending in 2025), and speaker fees and honoraria from Abbott, Bayer Healthcare, Biosense Webster, Biotronik, Amomed, AOP Health, Boston Scientific, Daiichi Sankyo, BMS/Pfizer and Medtronic. The other Topic Editor reports no competing interests related to this Research Topic. Topic editors: Ilija Uzelac, Martin Manninger Source: https://www.frontiersin.org/research-topics/87155/machine-learning-guided-electrophysiology-real-time-mapping-multimodal-biomarkers-and-translational-tools-for-arrhythmia-management