Atrial fibrillation (AF) is a highly prevalent, morbid condition. Anticoagulation to prevent thromboembolic strokes is a foremost priority in AF but adherence is challenging for patients and lapses in anticoagulation are common. Chronic disease self-management (CDSM) is a recognized program to enhance self-efficacy and improve adherence, quality of life, and patient-centered health outcomes. Rural patients with AF experience increased vulnerability to adverse outcomes due to geographic and social isolation, poor health care access, and limited health literacy. This study uses an innovative, scalable CDSM intervention to improve anticoagulation adherence in rural patients with AF.

This is a randomized clinical trial to evaluate the effect of an embodied conversational agent (ECA) on health outcomes in people with atrial fibrillation. The study will enroll 264 patients who reside in rural, Western Pennsylvania who have atrial fibrillation. Participants will be randomized to the intervention or usual care. Intervention participants will receive a contemporary mobile phone, commonly known and referred to herein as a smartphone, with a relational agent, which simulates conversation and provides coaching, guidance, and assistance with chronic disease self-management. In addition participants will receive an FDA-approved instrument for heart rate and rhythm monitoring, which is named the "AliveCor KardiaMobile" device (herein referred to as the AliveCor Kardia), that pairs with the relational agent. Usual care participants will receive a smartphone as well, which will have the general health application called "WebMD." The intervention will last 4 months and participants will have visits at baseline, 4, 8 and 12 months. The study will evaluate the improvement in adherence to anticoagulation, quality of life, and health care utilization resulting from the intervention.

Inclusion Criteria:

1. Adult, age ≥21;
2. Diagnosis of AF, identified from the electronic health record (EHR) problem list and confirmed by 2 or more reports of AF from separate monitoring events at least 2 weeks apart (CG, Holter or event monitor);
3. Prescribed use of warfarin or direct-acting oral anticoagulant (DOAC) for AF stroke prevention;
4. English-speaking well enough to participate in informed consent and this study;
5. No plans to relocate from the area within 12 months of enrollment.

Exclusion Criteria:

1. Conditions other than AF that require anticoagulation, such as mechanical prosthetic valve, deep vein thrombosis, or pulmonary embolism;
2. History of pulmonary vein isolation or foreseen pulmonary vein isolation;
3. History of atrioventricular (AV) nodal ablation or foreseen AV nodal ablation;
4. Heart failure necessitating hospital admission ≤3 months prior to study inclusion;
5. Acute coronary syndrome (defined as at least 2 of the following: chest pain, ischemic electrocardiographic changes, or troponin ≥0.1 ng/mL) ≤3 months prior to study inclusion;
6. Untreated hyperthyroidism or ≤3 months euthyroidism before inclusion;
7. Foreseen pacemaker, internal cardioverter defibrillator, or cardiac resynchronization therapy;
8. Cardiac surgery ≤3 months before inclusion;
9. Planned cardiac surgery;
10. Presence of non-cardiovascular conditions likely to be fatal within 12 months (e.g., cancer);
11. Inability to comprehend the study protocol, defined as failing to answer correctly a set of questions on orientation and short-term memory during the consent process.

The study team will share individual participant data that underlie the results reported in the study's central manuscripts after deidentification (text, tables, figures, and appendices).

Such data will be available beginning 9 months and ending 36 months following publication of the main manuscripts resulting from this clinical trial.

Data will be made available to those investigators whose proposed use of the data has been approved by an independent review committee identified for this purpose. Proposals may be submitted up to 36 months following article publication. Applicants requesting access to the data will be responsible for the minimal administrative costs to provide the data set.