Classifying syndromes in traditional Chinese medicine based on ISOMAP-SVM

Syndrome, an abstract set of signs of human organism, is a unique concept in Traditional Chinese Medicine (TCM) field. The modern TCM objective process of syndromes classification, according to pattern recognition theories, is often affected by factors, such as data incompleteness and diagnostic complexity. Thus the results are not always satisfactory. In this paper, five common syndromes of Coronary Heart Disease (CHD), i.e. deficiency of heart qi, deficiency of heart yang, deficiency of heart yin, phlegm, and blood stasis, are properly classified via the analysis of quantized TCM diagnostic data from 832 CHD patients with common syndromes. Isometric Mapping (ISOMAP) was used for dimension reduction, and Support Vector Machine (SVM) was for classification. Higher syndrome classification rates were obtained via the ISOMAP-SVM method compared to normal SVM and PCA-SVM (SVM classification after PCA) method, and an accuracy of 89.69% was achieved, which indicates the feasibility of the proposed method.