Machine Learning-Based Clinical Decision Support to Predict Bleeding Risk in Patients with Atrial Fibrillation on Direct Oral Anticoagulants

Grant Application

Rahul Chaudhary, MD, MBA

Proposed Innovation

Atrial fibrillation (AF) is the most common type of arrhythmia — a fast, irregular heartbeat that affects millions of people in the United States. Because having AF increases the risk of stroke, blood thinners are commonly prescribed to reduce the formation of blood clots. While newer direct oral anticoagulants are considered safer than warfarin, they still can increase the risk of bleeding. Some high-risk patients may benefit from having a left atrial appendage (LAA) occluder implanted to permanently seal off the area of the heart where clots form.

Through this innovative project, a state-of-the-art bleeding risk assessment tool will be created to determine whether an AF patient should be treated with a direct oral anticoagulant or an LAA occluder device.

Improvements in Action

An accurate bleeding risk assessment is crucial in making informed treatment decisions. This project will combine the use of advanced machine learning techniques and electronic health records to create an accurate and personalized bleeding risk assessment tool.

As part of the project, the AI-HEART Lab (Artificial Intelligence for Holistic Evaluation and Advancement of Cardiovascular Thrombosis) will be established. The AI-HEART Lab aims to advance precision medicine by leveraging data science and research translation using artificial intelligence.

Intended Outcomes

The development of this clinical decision support tool is expected to improve bleeding risk classification by 20%, resulting in earlier referrals of patients for an LAA occluder implant. This will save lives and enhance quality of life for patients by reducing bleeding complications and anxiety in high-risk patients.