Fig 4From: Predicting lung cancer survival prognosis based on the conditional survival bayesian networkBoxplots of the AUCs for models with different imputation methods on the following four simulated datasets. 1. Proportion of missing values = 10%, 2. Proportion of missing values = 20%, 3. Proportion of missing values = 30%, 4. Proportion of missing values = 40%. Note that the CSBN model attained the best performance. As the missing rate increased, the average AUCs of the Cox model decreased significantly, but the average AUCs of the CSBN were both good and stableBack to article page