Advanced artificial intelligence (AI) algorithms have shown potential in helping cardiologists predict bleeding and stroke risks in atrial fibrillation (AFib) patients on direct oral anticoagulants (DOACs), according to new research published in The American Journal of Cardiology.
The study, led by Dr. Rahul Chaudhary from the University of Pittsburgh Medical Center, explored whether AI could outperform traditional risk scores like HAS-BLED, ORBIT, and ATRIA in predicting major bleeding events in patients with non-valvular AFib. By analyzing electronic health record (EHR) data from over 20,000 patients who were prescribed DOACs between 2010 and 2022, researchers tested various machine learning models. The AI models demonstrated modest improvements in predicting bleeding events and hemorrhagic strokes after one, two, and five years compared to conventional methods.
AI’s predictive capabilities showed significant promise, particularly for patients with higher comorbidities. However, the models’ area under the receiver operating characteristic curve (AUC-ROC) ranged from 0.69 to 0.76, suggesting the need for further refinement. The authors emphasized the importance of future work to improve these models for broader clinical application.
This study underscores the potential of AI to enhance clinical decision-making and refine patient risk assessments, paving the way for better management of AFib patients on DOACs. Follow MEDWIRE.AI for more insights on AI in healthcare and cardiovascular innovations.
