Abstract：Electrocardiogram (ECG) is an electrical signal of the heart recorded from the body surface by ECG recorder, which is an important tool for diagnosing cardiovascular diseases. As an important branch of artificial intelligence, machine learning can obtain information features from large ECG data sets to make accurate classification, diagnosis and interpretation of ECG, and its diagnostic efficacy even reaches the level of medical experts. Machine learning has been widely used in clinical practice. In recent years, it has been found that machine learning also can effectively extract features in ECGs that cannot be recognized by human eyes to predict unidentified cardiovascular diseases such as left ventricular systolic dysfunction, asymptomatic atrial fibrillation and paroxysmal supraventricular tachycardia. As an important algorithm of machine learning, deep learning represented by convolutional neural network also can make disease prediction more accurate. This review mainly summarizes the progress of ECGbased machine learning applied in predicting cardiovascular diseases.
