Feature selection and syndrome prediction for liver cirrhosis in traditional Chinese medicine

Traditional Chinese medicine (TCM) treatment is one of the safe and effective methods for liver cirrhosis. In the process of its treatment, a very important step, syndrome prediction is generally performed by physicians at present, which actually hinders the application prospects of TCM. Based on the data mining algorithm, a novel method called TCMSP (traditional Chinese medicine syndrome prediction) is proposed, which consists of two phases. In the first phase, based on an improved information gain method in multi-view, the critical features are filtered from the original features. In the second phase, the class label of a new case is predicted automatically based on accuracy-weighted majority voting. The proposed method is evaluated by the liver cirrhosis dataset, 20 critical features are selected from original 105 features and the corresponding syndromes of 138 new cases are identified respectively. The critical features are in sound agreement with those used by the physicians in making their clinical decisions. Finally, this new method is also demonstrated on three standard datasets (SPECT Heart, Lung Cancer and Iris) and the results are compared with some other methods. The experimental results show that TCMSP method performs well in the field of TCM diagnosis.