A complex system entropy-based classification algorithm to extract syndrome element in Coronary Heart Disease

With the aging of population and prolong of human life, morbidity and mortality caused by Coronary Heart Disease (CHD) is growing year by year, turning CHD into a big disease all over the world. Treating CHD by Traditional Chinese Medicine (TCM) has a long history and significant curative eftect. However, the science of treatment CHD by TCM lacks interpretation and validation. Therefore, there are more and more researchers who are devoting to interpret and validate TCM by applying various data mining methods In this paper, we aim to apply a new data mining method-complex system entropy partition algorithm to analyze symptoms information of CHD patients. We carry out a clinical epidemiology survey by choosing 403 CHD patients, collecting corresponding symptoms information. Then we proposed complex system entropy partition theory as an unsupervised method to mining the symptoms combination in the data, each combination is diagnosed as a syndrome as per TCM theory. We get 8 combinations in the CHD. Finally, we validate the algorithm and use 10-fold cross validation to test three performance of the algorithm: Sensitivity, Specificity and Accuracy. The results indicate that the complex system algorithm can standardize the syndrome in CHD, more important, TCM is a science in the language of complex system.