Atrial fibrillation (AF) is a typical persistent arrhythmia, and revealing its pathogenesis and pathophysiological process is the key to AF diagnosis, prevention, and treatment. It also assists AF drug development and clinical equipment design. Experimental and clinical results can only show the local characteristics of cells or sub-cells and the macroscopic results of atrial fibrillation. However, with the development of bioinformatics acquisition technology and statistical analysis technology, using multi-scale virtual heart model to achieve the unity of macro-and micro-mechanism has attracted the attention of the research community. In this paper, we systematically reviewed the advances in virtual heart modeling at various levels, such as ion channels, cardiac cells, heart tissues and organs, and discussed the mechanisms of atrial fibrillation based on virtual heart models and the treatment of atrial fibrillation. This paper also presents the challenges and future development of atrial fibrillation.