Article Details

Abstract

Aiming at the problems of insufficient context semantic capture, difficulty in effectively capturing long-distance dependence information and low classification accuracy in the study of TCM medical records diagnosis and classification, a hybrid model combining Text Convolutional Neural Network (TextCNN) and Gated Recurrent Unit (GRU) is proposed. Firstly, the Word2Vec model is used to train the word vector and construct the local word vector library. Secondly, the TextCNN is used to extract the text features of TCM medical records to capture local important information. Finally, the GRU is used to model the context information of the extracted features, thereby significantly enhancing the model's ability to process long dependencies. The experimental results show that the model performs well in the text classification task of TCM medicals records diagnosis, the prediction accuracy reaches 85.01%, and the F1 value is 81.86%.

Keywords

TCM medical record; TextCNN; GRU; Word2Vec model; text classification

References

1. [1] Yi Junhui, check Qinglin A review of research on symptom information extraction in traditional Chinese medicine [J]. Computer Engineering and Applications, 2023, 59 (17): 35-47
2. [2] Liu Gang, Gao Lu, Xie Xinsheng, etc Research progress in digital traditional Chinese medicine [J]. Chinese Journal of Traditional Chinese Medicine, 2024, 42 (9): 9-12
3. [3] THOMPSON P. Looking Back: On Relevance, Probabilistic Indexing and Information Retrieval [J].Information Processing & Management，2007，44（2）：963-970.
4. [4] ZHAO W D，ZHU L，WANG M，et al. WTL-CNN: A News Text Classification Method of Convolutional Neural Network Based on Weighted Word Embedding [J].Connection Science，2022，34（1）：2291-2312.
5. [5] HE D P，HE Z L，LIU C. Recommendation Algorithm Combining Tag Data and Naive Bayes Classification [C]//2020 3rd International Conference on Electron Device and Mechanical Engineering（ICEDME）.Suzhou：IEEE，2020：662-666.
6. [6] ZHANG M Y，AI X B，HU Y Z. Chinese Text Classification System On Regulatory Information Based On SVM [J].IOP Conference Series: Earth and Environmental Science，2019，252（2）：022133.
7. [7] CHEN Z，ZHOU L J，LI X D，et al. The Lao Text Classification Method Based on KNN [J].Procedia computer science，2020，166：523-528.
8. [8] DIENG A B，WANG C，GAO J F，et al. TopicRNN: A Recurrent Neural Network with Long-Range Semantic Dependency [J/OL].arXiv:1611.01702 [cs.CL].[2024-09-18].https://arxiv.org/abs/1611.01702.
9. [9] ZHOU Y J，LI J L，CHI J H，et al. Set-CNN: A Text Convolutional Neural Network Based on Semantic Extension for Short Text Classification [J/OL].Knowledge-based systems，2022，257：109948.[2024-09-18].https://www.sciencedirect.com/science/article/abs/pii/S0950705122010413?via%3Dihub.
10. [10] ZHANG C，GAO R Z，MA X Y，et al. W-TextCNN: A TextCNN Model With Weighted Word Embeddings for Chinese Address Pattern Classification [J/OL].Computers，environment and urban systems，2022，95：101819.https://www.sciencedirect.com/science/article/abs/pii/S0198971522000631?via%3Dihub.
11. [11] GUO B，ZHANG C X，LIU J M，et al. Improving Text Classification with Weighted Word Embeddings Via A Multi-Channel TextCNN Model [J].Neurocomputing，2019，363：366-374.
12. [12] TRANG P T Q，THANG B M，HAI D T. Single Concatenated Input is Better than Indenpendent Multiple-input for CNNs to Predict Chemical-induced Disease Relation from Literature [J].VNU Journal of Science：Computer Science and Communication Engineering，2020，36（1）：11-16.
13. [13] DEVLIN J，CHANG W M，LEE K，et al. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding [J].Conference on the North American Chapter of the Association for Computational Linguistics：Human Language Technologies，2019：3498-4195.
14. [14] Jiang Huijie, Cha Qinglin A Method for Matching Chief Complaints in Traditional Chinese Medicine Cases Based on Twin Networks [J]. Modern Information Technology, 2023, 7 (23): 122-126
15. [15] Jiang Huijie, Cha Qinglin A method for matching chief complaints in traditional Chinese medicine cases based on semantic enhancement [J]. Modern Information Technology, 2023, 7 (24): 142-146
16. [16] Xi Ningli, Zhu Lijia, Wang Lutong, etc A method for constructing a word vector model using Word2vec [J]. Computer and Information Technology, 2023, 31 (1): 43-46
17. [17] Wang Ye, Zhou Aohui, Zhou Siyuan, etc Method for Obtaining Requirements for Intelligent Computing Services [J]. Computer Applications, 2022, 42 (11): 3486-3492
18. [18] Wang Kan, Wang Mengyang, Liu Xin, etc Fusion of self attention mechanism and CNN BiGRU for event detection [J]. Journal of Xi'an University of Electronic Science and Technology, 2022, 49 (5): 181-188

Funding

Key R&D Program of Jiangxi Provincial Department of Science and Technology (20171ACG70011); Key R&D Program of Jiangxi Provincial Department of Science and Technology (20203BBG72W008)

About the authors

Qiu Xuefeng (1998-), male, Han ethnicity, from Shangqiu, Henan Province, is currently pursuing a master's degree in natural language processing; Cha Qinglin (1973-), male, Han ethnicity, from Shangrao, Jiangxi, professor, master's supervisor, master's student, research direction: Traditional Chinese Medicine Informatics; Miao Zhen (1997-), male, Han ethnicity, from Liaoyuan, Jilin Province, is currently pursuing a master's degree in data mining; Liu Ming (1996-), male, Han ethnicity, from Ganzhou, Jiangxi Province, is currently pursuing a master's degree in data mining; Li Xinyi (2000-), female, Han ethnicity, from Shangqiu, Henan Province, is currently pursuing a bachelor's degree with a research focus on natural language processing.