Idiopathic inflammatory myopathy (IIM) is hard to diagnose without a muscle biopsy. We aimed to identify a metabolite panel for IIM detection by metabolomics approach in serum samples and to explore the metabolomic signature in tissue samples from a mouse model. We obtained serum samples from IIM patients, ankylosing spondylitis (AS) patients, healthy volunteers and muscle tissue samples from IIM murine model. All samples were subjected to a targeted metabolomic approach with various statistical analyses on serum and tissue samples to identify metabolic alterations. Three machine learning methods, such as logistic regression (LR), support vector machine (SVM), and random forest (RF), were applied to build prediction models. A set of 7 predictive metabolites was calculated using backward stepwise selection, and the model was evaluated within 5-fold cross-validation by using three machine algorithms. The model produced an area under the receiver operating characteristic curve values of 0.955 (LR), 0.908 (RF) and 0.918 (SVM). A total of 68 metabolites were significantly changed in mouse tissue. Notably, the most influential pathways contributing to the inflammation of muscle were the polyamine pathway and the beta-alanine pathway. Our metabolomic approach offers the potential biomarkers of IIM and reveals pathologically relevant metabolic pathways that are associated with IIM.