Fig. 5

Weighted accuracy of the Naïve Bayesian classification models in different comparisons. Six comparisons were investigated, whether the 6 lncRNAs, alone or in combination were suitable for biomarkers. Naïve Bayesian classifiers based on the normalized expression levels of different lncRNA combinations were created, namely (1) using each lncRNA alone, (2) all six lncRNAs (i.e. the full model), and (3) only those that showed statistically significant expression differences in case of the given comparison (i.e. the reduced model). The performances of the classification models were assessed by computing their weighted accuracy utilizing a leave-one-out cross-validation scheme. Color codes: blue: lncRNAs showing no expression differences between the two groups; red: lncRNAs showing significant differences between the two groups; green: all lncRNAs