LDA-based model for traditional Chinese medicine diagnosis

Based on Lejeune Dirichlet allocation (LDA) model and multi-relationship topic model in Diagnosis and treatment of traditional Chinese medicine (TCM), the Symptom TCM Treatment Diagnosis (STTD) model is purposed to extract topic structure of symptom, TCM, treatment and diagnosis from TCM clinical datum. This paper uses Gibbs sampling to do parameters reasoning and confirms the optimal number of topics according to the average similarity among topics. From the experimental results, given the symptoms of patient, STTD model can predicts treatment, result of diagnosis, TCM and the distribution of topic. And it is available for clinical doctors and researchers.