PROJECT SUMMARY Atrial fibrillation (AF) is the most common cardiac arrhythmia with a projected prevalence of 12.1 million by 2030 in the US. Given a two-to-five-fold increased risk of stroke in this population, anticoagulation is recommended for most AF patients. However, guidelines are broad and balancing the risks of stroke and bleeding require a personalized approach. The risk of stroke and bleeding from anticoagulation can be assessed with the CHA2DS2-VASc and HAS-BLED risk assessment scores, respectively. These risk assessments are valid, but do not distinguish between the clinical outcomes, which is a significant problem given that the medical complications of a stroke are far greater than the medical complications of a bleed. The ACC/AHA clinical performance measures for AF identify shared decision making for anticoagulant therapy as a quality metric. The measure does not, however, address an important gap in implementation, namely the lack of a value clarification to demonstrate the risks of stroke and bleeding. Given the widespread adoption of electronic health records, there is a critical need to develop technologies that facilitate value clarification of stroke and bleeding risk during shared decision making for anticoagulation prescribing in AF. The proposed project will work with patients and providers to create a decision support tool that clarifies the risk tradeoffs for anticoagulation in AF, develop and implement a shared decision-making support application based on the patient and provider feedback and determine if the new decision support tool with a value clarification facilitates knowledge of the risk tradeoffs compared to standard communication. In addition, this proposal will demonstrate the use of the University of Michigan Learning Health Sciences (LHS) Knowledge Grid (KGrid), which consists of a digital Library to support large-scale knowledge curation and management and an Activator to provide rapid delivery of knowledge into practice through a customized decision support tool to recipients. These data will be crucial for a planned randomized trial to test the effects of the newly created shared decision support tool on anticoagulation outcomes in AF in multicenter electronic health record implementation. In addition, these data will allow for future research of the LHS KGrid as a scalable and standardized infrastructure for shared decision making and risk communication.