Random forest of dipolar trees for survival prediction

In the paper the method of using the ensemble of dipolar trees for survival prediction is presented. In the approach the random forest is applied to calculate the aggregated Kaplan-Meier survival function for a new patient. The induction of individual dipolar regression tree is based on minimization of a piece-wise linear criterion function. The algorithm allows using the information from censored observations for which the exact survival time is unknown. The Brier score is used to evaluate the prediction ability of the received model.