[Hinews] SEOUL, South Korea — Researchers at Yonsei University College of Medicine have developed an artificial intelligence (AI) model that predicts atrial fibrillation (AF) by analyzing blood protein data, offering a novel tool for early detection of this common cardiac arrhythmia.

The approach, requiring only a small blood sample, enables identification of individuals at risk for AF before symptoms manifest. The study was led by Professors Boyoung Joung, Daehoon Kim, and Hanjin Park from the Division of Cardiology, alongside Assistant Professor Pil-sung Yang from the Department of Biomedical Sciences. Their findings were published in the latest issue of Circulation, a premier cardiology journal with an impact factor of 35.5.

From left: Professors Boyoung Joung, Daehoon Kim, Hanjin Park, and Assistant Professor Pil-sung Yang. (Photo courtesy of Severance Hospital)
From left: Professors Boyoung Joung, Daehoon Kim, Hanjin Park, and Assistant Professor Pil-sung Yang. (Photo courtesy of Severance Hospital)


Atrial fibrillation, the most prevalent heart rhythm disorder, significantly increases the risk of stroke and heart failure. Its often asymptomatic early stages frequently delay diagnosis, emphasizing the need for effective predictive strategies.

The team analyzed data from approximately 63,000 individuals in the UK Biobank, identifying proteins statistically associated with AF onset. The model’s predictive accuracy was validated using the U.S. ARIC cohort, confirming its reliability across diverse racial and environmental populations.

The AI model outperformed existing clinical-information-based prediction tools in accuracy and predictive power. It also includes a feature to estimate the time until AF onset, providing insights into disease progression beyond simple risk assessment.

The study further revealed that certain proteins are linked not only to AF but also to other cardiovascular conditions, such as stroke and heart failure, suggesting their potential as shared biomarkers.

“This blood protein-based prediction technology could be a cornerstone for personalized cardiovascular disease prevention,” said Professor Joung. “Identifying asymptomatic high-risk individuals early enables proactive interventions.”

Professor Kim emphasized the model’s development using data from European and Asian populations, ensuring its applicability across diverse demographics. “We plan further studies to facilitate clinical implementation,” he added.

The research highlights the growing role of AI in cardiovascular medicine, paving the way for blood-based tests to enhance early detection of heart disease.

저작권자 © Hinews 무단전재 및 재배포 금지