A Novel Deep-Learning-Based Model for Medical Text Classification
Zhengfei Shen, Shaohua Zhang · 2020
In recent years, with development of the Internet hospitals and natural language processing technology, intelligent medical guidance based on machine learning has been gained increasing attentions. Medical text classification is indispensable for intelligent medical guidance. In this paper, we propose a novel deep-learning-based model named CNN-MHA-BLSTM for medical text classification. The model combines the characteristics of CNN, Multi-Head Attention and Bidirectional LSTM to capture local potential features, contextual information and contribution of each feature to the classification. We conduct numerical simulations using real dataset to verify the validation of the model. The results show that the proposed model achieves a performance with accuracy of 91.99% and F1-score of 92.03%, which outperforms some typical methods for text classification.