Dynamic estimation with random forests for discrete-time survival data

TL;DR

This paper explores various simple methods using random forests for dynamic estimation of survival and hazard functions in discrete-time survival data, considering time-varying covariates and their impact on estimation accuracy.

Contribution

It introduces and compares multiple easy-to-implement random forest approaches for dynamic survival analysis with time-varying covariates.

Findings

All methods perform well with no clear winner.

Global fits are better in difficult estimation scenarios.

Local fits excel when data and signals are strong.

Abstract

Time-varying covariates are often available in survival studies and estimation of the hazard function needs to be updated as new information becomes available. In this paper, we investigate several different easy-to-implement ways that random forests can be used for dynamic estimation of the survival or hazard function from discrete-time survival data. The results from a simulation study indicate that all methods can perform well, and that none dominates the others. In general, situations that are more difficult from an estimation point of view (such as weaker signals and less data) favour a global fit, pooling over all time points, while situations that are easier from an estimation point of view (such as stronger signals and more data) favor local fits.