TCM-SD: A Benchmark for Probing Syndrome Differentiation via Natural Language Processing

CCL 2022 Proceedings of the 21st China National Conference on Computational Linguistics, pages 908-920, Nanchang, China, October 14 - 16, 2022. (c) Technical Committee on Computational Linguistics, Chinese Information Processing Society of China 908 TCM-SD: A Benchmark for Probing Syndrome Differentiation via Natural Language Processing Mucheng Ren1, Heyan Huang1, Yuxiang Zhou1, Qianwen Cao1, Yuan Bu2, and Yang Gao1 1School of Computer Science and Technology, Beijing Institute of Technology, Beijing, China 2Xuzhou City Hospital of Traditional Chinese Medicine, Xuzhou, China {renm,hhy63,yxzhou,qwcao,gyang}@bit.edu.cn buyuantcm@gmail.com Abstract Traditional Chinese Medicine (TCM) is a natural, safe, and effective therapy that has spread and been applied worldwide. The unique TCM diagnosis and treatment system requires a com- prehensive analysis of a patient’s symptoms hidden in the clinical record written in free text. Prior studies have shown that this system can be informationized and intelligentized with the aid of artificial intelligence (AI) technology, such as natural language processing (NLP). However, existing datasets are not of sufficient quality nor quantity to support the further development of data-driven AI technology in TCM. Therefore, in this paper, we focus on the core task of the TCM diagnosis and treatment system—syndrome differentiation (SD)—and we introduce the first pub- lic large-scale benchmark for SD, called TCM-SD. Our benchmark contains 54,152 real-world clinical records covering 148 syndromes. Furthermore, we collect a large-scale unlabelled textual corpus in the field of TCM and propose a domain-specific pre-trained language model, called ZY- BERT. We conducted experiments using deep neural networks to establish a strong performance baseline, reveal various challenges in SD, and prove the potential of domain-specific pre-trained language model. Our study and analysis reveal opportunities for incorporating computer science and linguistics knowledge to explore the empirical validity of TCM theories. 1 Introduction As an essential application domain of natural language processing (NLP), medicine has received re- markable attention in recent years. Many studies have explored the integration of a variety of NLP tasks with medicine, including question answering (Pampari et al., 2018; Tian et al., 2019), machine reading comprehension (Li et al., 2020; Yue et al., 2020), dialogue (Zeng et al., 2020), named entity recogni- tion (Jochim and Deleris, 2017; He et al., 2020), and information retrieval (Liu et al., 2018). Meanwhile, numerous datasets in the medical domain with different task formats have also been proposed (Pampari et al., 2018; Li et al., 2020; Tian et al., 2019). These have greatly promoted the development of the field. Finally, breakthroughs in such tasks have led to advances in various medical-related applications, such as decision support (Feng et al., 2020; Panigutti et al., 2021) and International Classification of Disease (ICD) coding (Cao et al., 2020; Yuan et al., 2022). However, most existing datasets and previous studies are related to modern medicine, while tradi- tional medicine has rarely been explored. Compared to modern medicine, traditional medicine is often faced with a lack of standards and scientific explanations, making it more challenging. Therefore, it is more urgent to adopt methods of modern science, especially NLP, to explore the principles of …

Content truncated to 500 words