Reviewer’s report Title: Feature Selection through Validation and Un-censoring of Endovascular repair Survival Data for Predicting the Risk of Re-intervention

The authors describe a feature selection technique suitable for right censored survival data. They transform the original regression problem (prediction of the time of an event) into a binary classification problem by stratifying patients into a low-risk group (time < 5 years) and a highrisk group (time >= 5 years). First, they reduce the dimensionality of the data by employing factor analysis. Next, they perform stagewise selection of features based on the p-value of the log-rank test statistic, which has been computed from the assignments to low/high-risk groups as predicted by an artificial neural network. To address censoring, they employ a Bayesian network that is used to impute the survival time of right censored samples. They validated their approach based on dataset concerning endovascular aortic repair.