Application of Multi-label Learning Model for Chronic Kidney Disease Syndrome Classification

Traditional Chinese Medicine (TCM) has been widely applied in Chronic Kidney Disease (CKD), syndrome classification is a very important step in the diagnosis of TCM. Based on clinical diagnosis experiences and the analogy theory of TCM, we propose a method that is grounded on analogy theory of TCM for syndrome diagnosis (ATTSCM). The ATTSCM method utilizes the feature selection algorithm to select the significant symptoms, and the clustering algorithm combine multi-label learning algorithm to distinguish the syndromes. The proposed method is evaluated by the chronic kidney disease dataset which is collated by University of Chinese Medicine. Three evaluation measures of multi-label classification are taken to evaluate the performance of syndrome classification, including classification accuracy, micro F1 measure and hamming loss. Experimental results show that the proposed method performs well in the application for TCM syndrome differentiation of CKD.