Developing an AI model for predicting atrial fibrillation with a drop of blood...Based on protein information in the blood
Jun 09, 2025
Yonsei University has developed an AI model that predicts atrial fibrillation by analyzing blood.
Professor Information Young, Kim Dae-hoon, and Professor Park Han-jin (cardiology) in the internal medicine class at Yonsei University School of Medicine, and Yang Pil-sung's assistant research team at the Medical Life Science Department announced on the 9th that they have developed an AI model that can predict the risk of atrial fibrillation based on protein in the blood.
Atrial fibrillation is the most common cardiac arrhythmia and is a major cause of increased risk of stroke and heart failure. However, in the early stages, symptoms are not clear, so it is easy to be left undiagnosed. Accordingly, a precise medical strategy is needed to accurately predict the risk before the disease occurs, select high-risk groups, and implement preventive treatment.
The research team analyzed the association between protein in the blood and atrial fibrillation on data from about 63,000 UK biobank. This identified a protein candidate group that showed a significant correlation with the occurrence of atrial fibrillation. Since then, it has been confirmed that the protein candidate group identified in cooperation with ARIC cohort researchers in the United States works equally well. When using the protein information of the proteomics model developed by the research team, it showed better accuracy than the existing clinical prediction model.
In particular, the protein information predicted the time until atrial fibrillation actually occurred, which the research team evaluated as a function that could estimate the progress of the disease beyond simple risk prediction.
In addition, some proteins have been associated with atrial fibrillation as well as the occurrence of comorbidities such as stroke and heart failure, showing the potential for expansion into new biomarkers across cardiovascular diseases.
Professor Information Young explained "By predicting the risk of atrial fibrillation through blood protein analysis, it will be able to provide an important turning point in the prevention-oriented cardiovascular treatment paradigm in the future."
Professors Kim Dae-hoon and Park Han-jin emphasized that "This study is a large-scale protein analysis in blood based on a large number of European and Asian population groups, and it is significant in that it presents a predictive model that can be used in various races and environments." The findings were recently published in the international journal Circulation (IF 35.5).
Professor Information Young, Kim Dae-hoon, and Professor Park Han-jin (cardiology) in the internal medicine class at Yonsei University School of Medicine, and Yang Pil-sung's assistant research team at the Medical Life Science Department announced on the 9th that they have developed an AI model that can predict the risk of atrial fibrillation based on protein in the blood.
Atrial fibrillation is the most common cardiac arrhythmia and is a major cause of increased risk of stroke and heart failure. However, in the early stages, symptoms are not clear, so it is easy to be left undiagnosed. Accordingly, a precise medical strategy is needed to accurately predict the risk before the disease occurs, select high-risk groups, and implement preventive treatment.
The research team analyzed the association between protein in the blood and atrial fibrillation on data from about 63,000 UK biobank. This identified a protein candidate group that showed a significant correlation with the occurrence of atrial fibrillation. Since then, it has been confirmed that the protein candidate group identified in cooperation with ARIC cohort researchers in the United States works equally well. When using the protein information of the proteomics model developed by the research team, it showed better accuracy than the existing clinical prediction model.
In particular, the protein information predicted the time until atrial fibrillation actually occurred, which the research team evaluated as a function that could estimate the progress of the disease beyond simple risk prediction.
In addition, some proteins have been associated with atrial fibrillation as well as the occurrence of comorbidities such as stroke and heart failure, showing the potential for expansion into new biomarkers across cardiovascular diseases.
Professor Information Young explained "By predicting the risk of atrial fibrillation through blood protein analysis, it will be able to provide an important turning point in the prevention-oriented cardiovascular treatment paradigm in the future."
Professors Kim Dae-hoon and Park Han-jin emphasized that "This study is a large-scale protein analysis in blood based on a large number of European and Asian population groups, and it is significant in that it presents a predictive model that can be used in various races and environments." The findings were recently published in the international journal Circulation (IF 35.5).
|
This article was translated by Naver AI translator.
