Research on Assistant Diagnostic Method of TCM Based on BERT and BiGRU Recurrent Neural Network

This paper proposes a model based on BERT and bidirectional GRU (BiGRU) recurrent neural network is proposed to realize disease diagnosis of patients. This method can improve the accuracy of traditional Chinese medicine (TCM) auxiliary diagnosis. First of all, this paper uses the BERT model to obtain the feature representation of Chinese medicine text and generates a text vector. Secondly, the obtained text vector is input into the BiGRU network to realize the extraction of TCM text features. Finally, the Softmax function is used to discriminate patients' diseases. The experimental results show that the accuracy, precision, recall, and F1 score of the model proposed in this paper all reach more than 80%, and it has a good disease prediction accuracy, which verifies the effectiveness of the method in this paper.