Innovative AI Model Guides Personalized Blood Thinner Treatment for Atrial Fibrillation Patients

Mount Sinai researchers have developed a groundbreaking artificial intelligence (AI) model capable of providing tailored treatment recommendations for individuals with atrial fibrillation (AF). This innovative tool assists clinicians in determining whether prescribing blood-thinning medications—anticoagulants—is appropriate for each patient, with the primary goal of preventing strokes. Unlike traditional guidelines that rely on population-based risk scores, this model analyzes a patient’s entire electronic health record (EHR) to evaluate the unique balance between the risk of stroke and the potential for major bleeding complications.

The significance of this development lies in its ability to personalize treatment plans, potentially avoiding unnecessary anticoagulant use in nearly half of the AF patients who would previously have been recommended for such therapy. This approach could substantially reduce the risk of bleeding events while effectively preventing strokes.

AF affects an estimated 59 million people worldwide and is characterized by irregular heart rhythms that can lead to blood clot formation. These clots may travel to the brain, causing strokes. Although blood thinners are the standard preventive treatment, they carry risks—especially bleeding—that must be carefully weighed for every patient.

The AI model was trained on extensive data from electronic health records, comprising over 1.8 million patient visits, 82 million notes, and 1.2 billion data points, to generate individualized treatment recommendations. It was validated using patient data from Mount Sinai and externally on datasets from Stanford, showing high alignment with clinical objectives of reducing stroke and bleeding risks.

This advancement marks a potential paradigm shift in clinical decision-making, moving towards more precise, data-driven, and patient-centered care. Dr. Joshua Lampert, the lead researcher, emphasizes that this technology not only provides initial treatment suggestions but can also update recommendations dynamically as new patient data becomes available.

Leading experts suggest that if further trials confirm the model’s effectiveness, this AI-driven approach could dramatically improve outcomes for AF patients by enabling truly personalized anticoagulation strategies. It exemplifies how integrating AI with comprehensive health data can revolutionize clinical practice, making treatment decisions safer and more effective.

Sources: https://medicalxpress.com/news/2025-09-ai-accurately-atrial-fibrillation-patients.html