Dear Members of the ESC Working Group on e-Cardiology,

Artificial intelligence and digital technologies are rapidly reshaping cardiovascular care. Yet, the true measure of innovation lies not merely in technological sophistication, but in the ability to translate data into clinically meaningful decisions, more efficient care pathways, and improved patient outcomes. This month, the ESC Working Group on e-Cardiology has selected three recent publications that illustrate complementary applications of this evolving paradigm: AI-supported organisation of outpatient care, advanced cardiovascular risk prediction, and accessible multimodal monitoring of heart failure.

The first contribution addresses a persistent challenge in outpatient cardiology: identifying which patients require closer follow-up and which may safely undergo de-intensification of care. In their study, Cardiology hospital admission risk prediction: training, internal validation and technical implementation in the electronic health record, the investigators present CHARP, a machine-learning model designed to predict unplanned cardiac hospitalisation or all-cause death following an outpatient cardiology visit. Developed from more than 300,000 visits involving approximately 53,000 patients at Amsterdam UMC, the model was integrated directly into the electronic health record. Beyond its predictive performance, its clinical relevance lies in its potential to support risk-based follow-up strategies, helping allocate resources more efficiently while safely reducing unnecessary care for low-risk patients. Nevertheless, external validation and careful assessment of its impact on clinical decision-making remain essential before broader implementation.

A broader perspective on AI-based risk assessment is provided by Transformer-based models for cardiovascular disease risk prediction: what is the potential?. With particular reference to China-AIHeart, a sex-specific model developed to estimate 10-year cardiovascular risk in Chinese adults, the authors examine whether transformer architectures offer meaningful advantages over established statistical risk scores. Their ability to analyse complex longitudinal electronic health record data may enable a more nuanced representation of individual risk trajectories. However, increasing model complexity does not automatically translate into greater clinical value. Robust comparisons with conventional approaches, transparency, generalisability across populations, and integration into clinical workflows remain crucial requirements.

The third selected study shifts the focus from prediction to monitoring. Preliminary results of a single, easy-to-use device for multimodal detection of intervals associated with heart failure decompensation reports the initial evaluation of a compact device that simultaneously records phonocardiography, electrocardiography, and photoplethysmography. In this prospective, single-centre pilot study involving 36 hospitalised patients, including individuals with acute decompensated heart failure, non-expert users were able to acquire clinically relevant cardiac signals. Although preliminary, these findings highlight the potential of accessible multimodal technologies to facilitate longitudinal assessment of heart failure status. Larger studies will be required to determine whether these signals can reliably anticipate decompensation and translate into timely interventions and improved outcomes.

Together, these contributions demonstrate that digital cardiology extends far beyond the development of increasingly sophisticated algorithms and devices. Its transformative potential depends on whether such innovations can be embedded into clinical workflows, validated across diverse settings, and used to support—not burden—healthcare professionals and patients.

Finally, Roderick Treskes provides an expert commentary on Deep learning-derived biological age from preoperative chest radiographs predicts mortality following surgical and transcatheter procedures for structural heart disease beyond EuroSCORE II, highlighting its principal findings, clinical relevance, and potential implications for the future of AI-supported risk stratification in structural heart disease.

We hope that this selection, together with the accompanying expert commentary, will offer valuable insights, stimulate critical reflection, and further inspire your interest in the evolving role of artificial intelligence and digital technologies in cardiovascular care.

Maria Chiara Gatto
On behalf of the ESC Working Group on e-Cardiology