DEEP LEARNING ALGORITHM-BASED ELECTROCARDIOGRAM FEATURE EXTRACTION METHOD, APPARATUS, SYSTEM, DEVICE, AND CLASSIFICATION METHOD

摘要

A deep learning algorithm-based electrocardiogram feature extraction method, an apparatus, a system, a device, and a classification method. The deep learning algorithm-based electrocardiogram feature extraction method comprises the following steps: randomly capturing a segment of continuous electrocardiogram signals in a 12-lead electrocardiogram to be processed, the electrocardiogram signals comprising at least two cardiac cycles (S1); and inputting the captured electrocardiogram signals into a feature extraction model in the form of pictures, and extracting electrocardiogram signal features, the feature extraction model being obtained by means of training on the basis of a ResNet mode, or an Inception model, or an Inception-ResNet model (S2). According to the deep learning algorithm-based electrocardiogram feature extraction method, the apparatus, the system, the device, and the classification method, incompleteness caused by artificial design of features can be reduced, thereby improving the accuracy and diversity of deep learning algorithm-based electrocardiogram feature extraction.