| Popis: |
Electrocardiogram (ECG) has been proved to be the most common and effective approach to investigate arrhythmia. In clinical, a segment of ECG signal often indicates several arrhythmia diseases. Therefore, the automatic diagnosis algorithm of arrhythmia can be seen as a multi-label classification problem. In order to improve the classification accuracy, a method based on improved classifier chains is proposed in this paper. First, a deep neural network is pre-trained to extract the features of the ECG, and then multiple extreme random forest classifiers are used to construct a classifier chain in line with the process of clinical diagnosis, thereby completing the multi-label diagnosis of arrhythmia diseases. The experiment results show that compared with the method based on neural network, the subset accuracy of proposed method is improved from 76.62% to 83.94%, while other indicators are also improved. |
