An Analysis and Research of Type-2 Diabetes TCM Records Based On Text Mining

This paper analyzes 152 type-2 diabetes Traditional Chinese Medicine (TCM) records via text mining methods with the aim of identifying the key medicines, prescriptions and formulae when taking patients' TCM syndromes into consideration. After structuring the TCM syndrome variables according to the diagnostic scale of TCM syndrome elements, a Chinese segmentation method was adopted at the initial stage during text mining. K-Medoids method was selected to cluster the TCM records. Eventually, a FP-Growth algorithm was applied in this paper for the purpose of discovering hidden relationships between syndromes and prescriptions whose confidence values are relatively higher. In terms of the results, this research has shown 71% accuracy in test sets. Additionally, all three senior TCM doctors deem the result feasible and acceptable.