Ensemble learning for synthesis of the four diagnostics of TCM

This paper outlines the procedure of synthesis of the four diagnostics of traditional Chinese medicine (TCM). It is an important part of the modernization of TCM diagnosis. We apply the principle of ensemble learning, and present a systematic framework for synthesis of four diagnostic. Especially the logistic regression and LogitBoost methods are introduced. Experiment results on chronic hepatitis B dataset demonstrate that the proposed framework is suitable to the application of TCM diagnosis in clinical, and able to obtain a smaller and satisfactory critical feature subset, 15 critical features of TCM are selected from original 123 features. The critical features are in sound agreement with those used by the physicians in making their clinical decisions. At the same time, we obtain better performance in discriminating the syndromes of CHB. The classification accuracy is 95.3153%.