Predicting syndrome by NEI specifications: a comparison of five data mining algorithms in coronary heart disease

  • Authors:

  • Jianxin Chen;Guangcheng Xi;Yanwei Xing;Jing Chen;Jie Wang

  • Affiliations:

  • Key Laboratory of Complex Systems and Intelligence Science, Institute of Automation, Chinese Academy of Sciences, Beijing, China;Key Laboratory of Complex Systems and Intelligence Science, Institute of Automation, Chinese Academy of Sciences, Beijing, China;GuangAnMen Hospital, Chinese Academy of Chinese Medical Science, Beijing, China;Key Laboratory of Complex Systems and Intelligence Science, Institute of Automation, Chinese Academy of Sciences, Beijing, China;GuangAnMen Hospital, Chinese Academy of Chinese Medical Science, Beijing, China

  • Venue:
  • LSMS'07 Proceedings of the 2007 international conference on Life System Modeling and Simulation
  • Year:
  • 2007

Quantified Score

Hi-index 0.00

Visualization

Abstract

Nowadays, most Chinese take a way of integration of TCM and western medicine to heal CHD. However, the relation between them is rarely studied. In this paper, we carry out a clinical epidemiology to collect 102 cases, each of which is a CHD instance confirmed by Coronary Artery Angiography. Moreover, each case is diagnosed by TCM experts as what syndrome and the corresponding nine NEI specifications are measured. We want to explore whether there exist relation between syndrome and NEI specifications. Therefore, we employ five distinct kinds of data mining algorithms: Bayesian model; Neural Network; Support vector machine, Decision trees and logistic regression to perform prediction task and compare their performances. The results indicated that SVM is the best identifier with 90.5% accuracy on the holdout samples. The next is neural network with 88.9% accuracy, higher than Bayesian model with 82.2% counterpart. The decision tree is less worst, 77.9%, logistic regression models performs the worst, only 73.9%. We concluded that there do exist relation between syndrome and western medicine and SVM is the best model for predicting syndrome by NEI specifications.