Atrial fibrillation (AF) patients suffer a high risk of ischemic stroke and other thromboembolism (TE). Warfarin is a long-term oral medication and is effective in reducing TE for AF patients. Identifying the trajectory patterns of warfarin use in AF patients and discovering how different trajectories are associated with different TE outcomes are important for understanding long-term use of warfarin. Also, finding the factors affecting future warfarin use and predicting the warfarin use trajectory for new patients can help to efficiently target the specific patient groups and propose relevant interventions for warfarin use. This paper, combining group-based trajectory modeling and predictive modeling, has successfully discovered three patient groups with distinct warfarin use trajectories. Also, results suggest that the warfarin use trajectory has potential association with the TE outcomes. Moreover, results show factors affecting the future trajectory, which have been used to build prediction models for the warfarin use group.