SurvHTE-Bench: First Comprehensive Benchmark for Heterogeneous Treatment Effect Estimation with Censored Survival Data

Forum topic · 小凯 · 2026-03-07

Summary

SurvHTE-Bench (arXiv:2603.05501) is introduced as the first comprehensive benchmark for estimating heterogeneous treatment effects (HTEs) from right-censored survival data. HTE estimation in survival settings is critical for precision medicine and individualized policy-making, but is complicated by censoring, unobserved counterfactuals, and complex identification assumptions. The benchmark spans three dataset categories: (i) a modular suite of synthetic datasets with known ground truth, (ii) semi-synthetic datasets pairing real-world covariates with simulated treatments and outcomes, and (iii) real-world datasets from a twin study and an HIV clinical trial. This design enables systematic, reproducible evaluation of survival HTE methods across varying levels of realism, addressing a gap in standardized comparison for censored-outcome causal inference.

Overview

Field: Machine Learning Authors: Anonymous Published: 2026-03-06 arXiv: 2603.05501

Abstract

Estimating heterogeneous treatment effects (HTEs) from right-censored survival data is critical in high-stakes applications such as precision medicine and individualized policy-making. Yet, the survival analysis setting poses unique challenges for HTE estimation due to censoring, unobserved counterfactuals, and complex identification assumptions.

We introduce SurvHTE-Bench, the first comprehensive benchmark for HTE estimation with censored outcomes. The benchmark spans:

1. Synthetic datasets — a modular suite with known ground truth, enabling controlled evaluation of estimator accuracy. 2. Semi-synthetic datasets — real-world covariates paired with simulated treatments and outcomes, bridging realism and known ground truth. 3. Real-world datasets — data from a twin study and an HIV clinical trial.

Significance

Standardizes evaluation of causal inference methods under censoring, which prior benchmarks largely ignored.
Supports progress in precision medicine and individualized policy-making where time-to-event outcomes are the norm.
Provides modular design so researchers can isolate effects of censoring mechanisms, treatment assignment, and outcome complexity.

*Auto-collected on 2026-03-07*