Kawasaki disease (KD) and systemic juvenile idiopathic arthritis (sJIA) are two inflammatory diseases in children that usually present challenges for early differential diagnosis due to overlapping initial clinical manifestations and laboratory features. This study aims to classify KD and sJIA patients using machine learning models to reduce the early differential misdiagnosis rate of both diseases. We collected clinical and laboratory data from 240 patients (120 with KD and 120 with sJIA) and performed data preprocessing, feature engineering, and feature selection. we evaluated the classification performance of various machine learning models, including XGBoost, logistic regression, K-nearest neighbors (KNN), support vector machine (SVM), and multilayer perceptron (MLP). Our results indicate that the models can identify KD and sJIA patients efficiently, assisting clinicians in improving decision quality and efficiency when faced with early differential diagnosis of KD and sJIA.