Warfarin is an oral anticoagulant that is widely prescribed worldwide, and it has a large individual variability. Many factors contribute to the variability of warfarin. Although numerous algorithms of warfarin have also been developed by pharmacokinetics and pharmacodynamics (PKPD) and multivariate linear regression (MLR) models, there also exits some unknown factors that could affect the warfarin anticoagulant. In this study, the precision medication model of warfarin will be established based bioinformatics in cardiac surgery patients. About 200 cardiac surgery patients who will be administered warfarin will be enrolled. The demographic characters, combined drug and physiological factors will be collected from patients’ medical records. Some of single nucleotide polymorphisins (SNP) linked with warfarin PK and PD for pharmacogenetics will be performed using pyrosequencing. Fecal samples will be used for analyzing gut microbiota by 16S rDNA. S-warfarin, R-warfarin and vitamin K concentrations, and metabolomics will be detected by LC-MS/MS. International normalized ratio (INR) is the effective efficacy index for warfarin, and target INR should be within 1.5-2.5. All data will be analyzed by machine learning or neural networks. Expected the polymorphisms of pharmacogenomics (CYP2C9, VKORC1, and CYP4F2 etc.), vitamin K concentrations, and other biomarkers from gut microbiota and metabolomics were confirmed as the effect factor for the individual variability of warfarin. And an artificial intelligence model of warfarin would be established.