Patients' readmission can be considered as a critical factor affecting cost reduction while maintaining a high-quality treatment of patients. Therefore, predicting and controlling patients' readmission rates would signicantly improve the healthcare service. In this study, we aim at predicting the readmission of COPD (Chronic Obstructive Pulmonary Disease) patients through the deployment of machine learning algorithms. Area Under Curve (AUC) and ACCuracy (ACC) were considered as the main criteria for evaluating models' prediction power in each time frame. Then, the importance of the variables for each outcome was explicitly identied, and dened important variables have then been differentiated. Our study could achieve the highest accuracy in predicting readmission with %91 ACC.