Tree Structured Survival Analysis

A method in which a patient population is divided into 2 groups based on the feature that maximises differences in prognosis according to a log-rank statistic; this process is repeated for each subgroup, creating a tree structure, until further subdivision is not—statistically—significant or the terminal node—subgroup—has 10 cases; missing values—of which there are very few—are handled by an interpolation algorithm that uses the case values for another parameter, which most similarly ranks the cases