Dear members of the ESC Working Group on e-Cardiology,
We are pleased to share a new selection of timely studies showing how digital health and artificial intelligence (AI) are reshaping cardiovascular diagnostics, patient pathways, and data-driven decision-making.
Our first highlighted study by Herman et al presents a powerful AI model for detecting occlusion myocardial infarction using a single standard 12-lead electrocardiogram (ECG). Among more than 18,000 ECGs from international cohorts, the model demonstrated excellent diagnostic performance (AUC 0.94) and significantly outperformed traditional ST-elevation myocardial infarction criteria, while achieving accuracy comparable to expert ECG interpretation. With markedly higher sensitivity for identifying acute coronary occlusions requiring urgent revascularization, this AI approach holds strong potential to improve acute coronary syndrome triage, expedite life-saving treatment, and address one of the most critical diagnostic gaps in contemporary coronary care.
The second featured article by Engels and Hermans, TeleConvert-AF trial, evaluates a photoplethysmography-based rhythm monitoring strategy for patients awaiting electrical cardioversion for presumed persistent atrial fibrillation (AF). Using structured daily monitoring with a colour-alert system, spontaneous conversion to sinus rhythm or paroxysmal AF was identified in 13.7% of patients, allowing safe cancellation of cardioversion and avoiding unnecessary hospital visits. With 94% accuracy for sinus rhythm detection, excellent adherence, and positive patient experience, this approach proves feasible, accurate, and clinically meaningful, while also highlighting that AF recurrence remains common, underscoring the need for personalized follow-up strategies.
We then turn to the rapidly evolving domain of wearable technology for post-ablation rhythm monitoring. Using continuous implantable cardiac monitor data from a large clinical trial, study by Aguilar et al simulated the performance of commercial smartwatch AF detection algorithms. The findings are compelling: smartwatch algorithms demonstrated clinically strong sensitivity for AF recurrence, outperformed traditional intermittent monitoring strategies, and showed near-perfect correlation with AF burden (r > 0.97). Sensitivity increased further with longer daily wear time. These results strongly support wearables as scalable, accessible, and reliable tools for long-term heart rhythm surveillance after AF ablation, opening new possibilities for personalized follow-up strategies.
Finally, we highlight a thought-provoking study by van der Loo et al, commented in detail by Sven Knecht, assessing the ability of large language models to extract structured clinical data from echocardiography and coronary angiography reports. While commercially available models achieved promising performance, particularly when fine-tuned, important issues emerged regarding computational demands, variability across languages and institutions, and the broader question of whether structured data should be created “by AI” or “at source.” This work provides a balanced and realistic perspective: the potential is clear, but multicentre validation, interoperability, standardization, and robust governance frameworks are urgently needed before widespread deployment in routine cardiovascular care.
Together, these studies reflect the accelerating transition from proof-of-concept innovation toward real-world implementation. They emphasize that successful digital transformation is not merely about technology - but about integration into clinical workflows, patient engagement, accessibility, and ensuring quality, safety, and trust.
On behalf of the ESC Working Group on e-Cardiology,
Monika Gawałko, MD, PhD
Community Coordinator ESC WG on e-Cardiology 2024-2026
