The traditional methods of manual heart disease monitoring are highly dependent on senior doctors with prior knowledge, and their speeds and accuracies of monitoring disease need to be improved. In order to solve these problems, a ElectroCardioGraph (ECG) monitoring algorithm based on stack classifier was proposed for the determination of cardiac anomalies. Firstly, the advantages of various machine learning algorithms were combined, and these algorithms were integrated by the way of stack classifier to make up for the limitation of learning by single machine learning algorithm. Then, Synthetic Minority Over-sampling TEchnique (SMOTE) was used to perform data augmentation to the original dataset and balance the number of samples of various diseases, so as to improve the data balance. The proposed algorithm was compared with other machine learning algorithms on MIT-BIH dataset. Experimental results show that the proposed algorithm can improve the accuracy and speed of abnormal ECG monitoring.