Discovering Syndrome Regularities in Traditional Chinese Medicine Clinical by Topic Model

Abstract

Traditional Chinese Medicine (TCM) as one of most important approach for disease treatment in China for thousands of years. Lots of experience of famous experts in TCM is recorded in medical bibliography. The first vital work for TCM doctor is to diagnose the disease by the patients’ symptoms, and then predict the syndromes which the patient has. Generally, this process reflects the medical skill of the TCM doctors. Therefore, TCM diagnose is easy to misdiagnose and difficult to master for TCM doctors. In this paper, we proposed a probabilistic model—the symptom-syndrome topic model (SSTM) to explore connected knowledge between symptoms and syndromes. In the SSTM, symptom-syndrome are modeled by generative process. Finally, we conduct the experiment on the SSTM. The results show that the SSTM is effective for mining the syndrome regularities in TCM data.