Estimating heterogeneous treatment effects with right-censored data via causal survival forests

Yifan Cui, Michael R. Kosorok, Erik Sverdrup, Stefan Wager, Ruoqing Zhu
1/27/2020
stat.MEcs.LGstat.ML

Official code: grf-labs/grf (1,117 stars, GPL-3.0, last commit 2026-04-30).

Abstract

Forest-based methods have recently gained in popularity for non-parametric treatment effect estimation. Building on this line of work, we introduce causal survival forests, which can be used to estimate heterogeneous treatment effects in a survival and observational setting where outcomes may be right-censored. Our approach relies on orthogonal estimating equations to robustly adjust for both censoring and selection effects under unconfoundedness. In our experiments, we find our approach to perform well relative to a number of baselines.

View on arXivView PDF

Code Implementations(1)

grf-labs/grfOfficial100%
GPL-3.0

Generalized Random Forests

1,117282R, C++Aug 12, 20165 months agoGPL-3.0
causal-forestcausal-inferenceeconometricsmachine-learningrandom-forest+1 more

Similar Papers

Heterogeneous Treatment Effect in Time-to-Event Outcomes: Harnessing Censored Data with Recursively Imputed Trees

86% match2025

SurvHTE-Bench: A Benchmark for Heterogeneous Treatment Effect Estimation in Survival Analysis

83% match1 repo122026

Doubly Robust Learning of Heterogeneous Survival Treatment Effects under Time-Varying Treatments

82% match2026

Distilling interpretable causal trees from causal forests

81% match2024

Doubly robust identification of treatment effects from multiple environments

80% match1 repo22025

Quasi-Oracle Estimation of Heterogeneous Treatment Effects

79% match1 repo1272017

Cite this paper

@article{cui2020estimating,
  title  = {Estimating heterogeneous treatment effects with right-censored data via causal survival forests},
  author = {Yifan Cui and Michael R. Kosorok and Erik Sverdrup and Stefan Wager and Ruoqing Zhu},
  year   = {2020},
  eprint = {2001.09887},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/2001.09887}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2001.09887.svg)](https://codeofpaper.com/paper/2001.09887)

Discussion