Researchers Develop WARN Model for Early Detection of Atrial Fibrillation

Atrial fibrillation is a disorder that usually affects a person’s heartbeat and makes it irregular. It is one of the most common disorders, as it affected almost 59 million people in 2019. It can increase the risk of some severe complications such as dementia, heart failure, and stroke. Even though early detection and treatment are essential for this disorder, the current methods don’t work effectively. However, researchers from the Luxembourg Centre for Systems Biomedicine (LCSB) have developed an artificial intelligence model called Warning of Atrial Fibrillation (WARN) that can work accurately to predict this disorder.   

Research published in the scientific journal Patterns has revealed that the WARN model predicts the change in heartbeats (cardiac arrhythmia) from regular to irregular with an accuracy of 80%. This model gives the idea of cardiac arrhythmia before 30 minutes. Researchers involved 350 patients in this research and tested this model on them. This model successfully gave warnings 30 minutes before atrial fibrillation started.   

This model uses the heart rate data collected by devices like smartwatches. As this model works on wearable devices like smartwatches to track changes in heart rates, it can be beneficial for ordinary people to get early warnings of atrial fibrillation.  

The current methods used to predict atrial fibrillation are based on analysis of heartbeat and electrocardiogram (ECG). These current methods can only predict atrial fibrillation when it starts but fail to provide early warnings. However, this WARN model can predict and warn about the risk of atrial fibrillation 30 minutes before it starts. So, healthcare professionals can suggest early medications or other treatments to reduce this. This practice can be beneficial as it predicts the risk earlier and reduces the need for other emergency treatments.   

As atrial fibrillation is the most common heart-related disorder that affects millions of people and increases the risk of heart failure and stroke, this WARN model can be used to save many lives. This model predicts atrial fibrillation 30 minutes before it starts with an accuracy of 80%. Healthcare professionals can make effective strategies and treatments to reduce the risk of atrial fibrillation.   

Reference Link:  

Marino Gavidia et al., Early warning of atrial fibrillation using deep learning, Patterns (2024).   

DOI: 10.1016/j.patter.2024.100970