Skip to content
This repository was archived by the owner on Jun 25, 2025. It is now read-only.

PhilBoileau/2025_WNAR2025_contributed-talk_TEM-VIPs

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 

Repository files navigation

A nonparametric framework for treatment effect modifier discovery in high dimensions

Authors

Philippe Boileau, Ning Leng, Nima Hejazi, Mark van der Laan, and Sandrine Dudoit

Abstract

English:

Heterogeneous treatment effects are driven by treatment effect modifiers (TEMs), pre-treatment covariates that modify the effect of a treatment on an outcome. Current approaches for uncovering TEMs are limited to low-dimensional data, data with weakly correlated covariates, or parametric data-generating processes. We resolve these issues by developing a framework for defining model-agnostic TEM variable importance parameters appropriate for high-dimensional data with arbitrary correlation structure, deriving causal machine learning estimators of these parameters, and establishing these estimators' asymptotic properties. Simulation experiments demonstrate that these estimators' asymptotic guarantees are approximately achieved in realistic sample sizes for observational and randomized studies alike. This framework is applied to gene expression data collected during a clinical trial investigating the effect of a novel treatment on disease-free survival in breast cancer.

Related Papers

Related R Packages

About

Material for WNAR 2025 Contributed Talk

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

No releases published

Packages

 
 
 

Contributors