Cardiovascular disease remains one of the leading causes of death worldwide. Numerous studies have shown that a clinically significant reduction in risk can be achieved if modifiable risk factors are adequately addressed. These include hypertension, lipid management, smoking, alcohol use, physical activity, and environmental factors such as air pollution. Although often debated as less modifiable, socioeconomic status is also included in this list. Effective treatments are available to address these risk factors, but several practical challenges hinder their implementation.

The authors argue that Artificial Intelligence (AI) - an umbrella term encompassing a wide range of technologies and methodologies designed to enable machines to perform tasks that typically require human intelligence, such as learning from data (machine learning), recognizing patterns and images and adapting through reasoning and problem-solving - can improve the management of these risk factors and reduce cardiovascular risk at the population level (1). To support this claim, the paper provides an overview of both the terminology associated with AI and the available literature on its application in modifying cardiovascular risk factors. It presents evidence on the use of AI in influencing physical activity levels, obesity and diabetes, hypertension, lipid management, tobacco and alcohol use, and urban health. Additionally, it summarizes findings related to healthcare delivery, cost reduction, and legal and ethical considerations.

The paper offers a robust and accessible overview of the literature. It highlights tools that are already available or currently being tested in randomized controlled trials. Evidence suggests that for all the aforementioned cardiovascular risk factors, AI can assist with, or even replace, certain steps in diagnosis, management, or prediction of cardiovascular events. Clinically effective AI tools already exist for glucose management in patients with type II diabetes, as well as for managing hyperlipidemia and obesity. The paper also highlights challenges in using AI for healthcare delivery: both at the clinical care level (e.g., Large Language Models sometimes provide overly general recommendations due to training on non-medical datasets) and at the population level, where regulations are needed to ensure the quality of datasets used to train AI algorithms for medical purposes.

In conclusion, the narrative review by Meder and colleagues provides a comprehensive overview of currently available AI solutions in preventive cardiovascular medicine. We therefore recommend this article to any healthcare professional seeking a clear and informative introduction to AI and the evidence supporting its growing role in the delivery of preventive cardiovascular care in the near future.

On behalf of the ESC WG on e-Cardiology,
Roderick W. Treskes