In our "Quick Dive" series, the authors of publications from medical societies summarise the most important information and results of the respective publication. This time we dive into:
Quick Dive: Artificial Intelligence in Clinical Electrophysiology
State of the Art of Artificial Intelligence in Clinical Electrophysiology in 2025
A Scientific Statement of the European Heart Rhythm Association (EHRA) of the ESC, the Heart Rhythm Society (HRS), and the ESC Working Group on E-Cardiology
31 March 2025 | Written by: Emma Svennberg , Janet K Han , Enrico G Caiani , Sandy Engelhardt , Sabine Ernst , Paul Friedman , Rodrigue Garcia , Hamid Ghanbari , Gerhard Hindricks , Sharon H Man , José Millet , Sanjiv M Narayan , G André Ng , Peter A Noseworthy , Fleur V Y Tjong , Julia Ramírez , Jagmeet P Singh , Natalia Trayanova , David Duncker
Reviewed by: Jacob Tfelt Hansen , Joseph Barker , Ruben Casado-Arroyo , Neal A Chatterjee , Giulio Conte , Søren Zöga Diederichsen , Dominik Linz , Arun Umesh Mahtani , Alessandro Zorzi
By:
Martin Nölke
HERZMEDIZIN editorial team
04 Jul 2025
Image source (image above): vovan / Shutterstock.com
5 questions for the first and last authors
Dr. Emma Svennberg, Karolinska University Hospital, Stockholm, Sweden
Prof. David Duncker, Hannover Medical School, Germany
What is the reason for and aim of the publication?
The aim of the publication is to address the lack of standardization in the reporting and evaluation of AI-based research in clinical electrophysiology (EP). With the rapid growth of AI applications in arrhythmia detection, risk prediction, and procedural guidance, the field requires a structured framework to ensure transparency, reproducibility, and clinical relevance. To that end, the European Heart Rhythm Association (EHRA), in collaboration with HRS and the ESC Working Group on E-Cardiology, developed a 29-item EHRA AI checklist tailored specifically for AI studies in the field of clinical EP. This publication details the creation, validation, and application of that checklist across three major domains: atrial fibrillation (AF) management, sudden cardiac death (SCD), and EP lab applications.
What are the most important take-home messages?
- There is considerable variability and frequent underreporting in AI-based EP studies, especially regarding trial registration, participant demographics, and data handling. A dedicated framework and standardization is essential to enhance scientific rigor.
- Creation of the "EHRA AI checklist": A consensus-driven, Delphi-validated checklist of 29 items was created to guide authors and readers in properly reporting and assessing AI research in EP.
- Adoption of the EHRA AI checklist can improve understanding, foster better clinical translation, and ensure AI tools are effectively applied in EP.
Fig.: Overview – scientific statement. The EHRA AI checklist provides a structured framework for reporting AI research in EP. © 2025 the European Society of Cardiology. Europace (2025) 27, euaf071 https://doi.org/10.1093/europace/euaf071.
What are challenges in practical implementation – and possible solutions?
This manuscript shows inconsistent reporting standards and lack of external validation, limiting clinical trust and application of AI tools. The widespread dissemination and use of the EHRA AI checklist in EP journals and publications can overcome these inconsistencies.
What further developments on the topic are emerging?
Tools that make AI models interpretable for clinicians are gaining traction to build trust and facilitate clinical integration (explainable AI). Digital Twins and multimodal AI allow personalized simulations based on patient-specific data and are being explored to guide ablation strategies and predict outcomes. Furthermore, the integration of wearables and telemonitoring using AI algorithms can be used for arrhythmia detection and risk stratification, offering dynamic, real-time insights.
How can the EHRA AI checklist be adopted at scale to maximize its impact?
The EHRA AI checklist will now be integrated into journal submission requirements for the Europace journal. It could also be included in institutional research protocols or educational curricula for EP fellows in training and researchers.
Continue to the publication:
Reference: Svennberg E, Han JK, Caiani EG et al. State of the Art of Artificial Intelligence in Clinical Electrophysiology in 2025: A Scientific Statement of the European Heart Rhythm Association (EHRA) of the ESC, the Heart Rhythm Society (HRS), and the ESC Working Group on E-Cardiology. Europace. 2025 May 7;27(5):euaf071. doi: 10.1093/europace/euaf071. PMID: 40163651; PMCID: PMC12123071.
ESC Document types
Document types published by the ESC, Associations, Councils, Working Groups, and ESC Committees (according to the ESC Scientific Documents Policy):
ESC Clinical Practice Guidelines present the official ESC position on key topics in cardiovascular medicine. They are based on the assessment of published evidence and consensus by an independent group of experts. The documents include standardized, graded recommendations for clinical practice and indicate the level of supporting evidence.
ESC Pocket Guidelines provide a compact, practice-oriented summary of the full guideline, including all recommendation classes and levels of evidence.
Clinical Consensus Statements provide guidance for clinical management on topics not covered or not covered in sufficient detail in existing or upcoming ESC Clinical Practice Guidelines by evaluating scientific evidence or exploring expert consensus in a structured way.
Scientific Consensus Statements interpret scientific evidence and provide a summary position on the topic without specific advice for clinical practice.
Statements outline and convey the organisation’s position or policy on non-medical issues such as education, advocacy and ethical considerations.
ESC Quality Indicators enable healthcare providers to develop valid and feasible metrics to measure and improve the quality of cardiovascular care and describe, in a specific clinical situation, aspects of the process of care that are recommended (or not recommended) to be performed.