Fig. 7.

Cross-validation of the applied machine learning models for survival time prediction. Performance of three models (SVR-Linear, SVR-RBF and DTR) is compared. The fourth model, SVR-Poly has shown biased results (Fig. 6). Cross-validation is based on 5 subgroups defined by the TICF feature, obtained after using k-fold algorithm. On the vertical axis the mean values with standard deviations of success rates and error rates of the matchings of TICF feature are given