Skip to main content
Fig. 3 | BMC Cancer

Fig. 3

From: Computational prediction of multidisciplinary team decision-making for adjuvant breast cancer drug therapies: a machine learning approach

Fig. 3

Performance of machine learning models for predicting the MDT decisions. Each point indicates the mean AUC (from ten cross-validation runs) of a classifier for correctly predicting the outcome of MDT recommendation. The error bars indicate the 95% confidence intervals estimated by the Hanley-McNeil method. The open square indicate the classifiers without bootstrap-aggregation, whereas the solid squares indicate the corresponding classifiers with bootstrap-aggregation. Legend: R: ripple down rule, J J48 classifier. A: multiclass alternating decision tree, Sp support vector machine (SVM) with polynomial kernel, Sr SVM with radial basis function kernel, D decision Table 1: OneR classifier, B naive Bayesian classifier, N nearest neighbour classifier, L Multivariate logistic regression

Back to article page