Prediction of Survival in Patients with Breast Cancer Using Three Artificial Intelligence Techniques

As medical technology advances, has accumulated a large number of health-related data. Faced with increasingly complex analytical requirements, predictive data mining has become an essential instrument for hospital management and medical research. In this study, the breast cancer dataset is collected from a regional teaching hospital in central Taiwan between 2002 and 2009. The prognostic factors composed of 8 attributes including 967 subjects, of which 861 are survival after treatment. The three techniques, artificial neural networks (ANNs), support vector machine (SVM) and Bayesian classifier, have been discussed which is used to investigated and evaluated for predicting breast cancer survival. As can be seen from the results, the prediction accuracy of a 10-fold cross validation is 90.31%, 89.79% and 88.64%, respectively. Classification results of SVM are slightly better as compared to ANN and Bayesian classifier, however, from a relatively low variance, the results show that the SVM will be the best prognosis in clinical practice.