Measuring Variable Importance in Heterogeneous Treatment Effects with Confidence
This work addresses the need for robust variable importance measures in causal machine learning, particularly for biomedical applications with limited data, though it is incremental as it builds on existing methods like Conditional Permutation Importance.
The paper tackles the problem of reliably identifying which variables drive heterogeneity in treatment effects, proposing PermuCATE, an algorithm that provides statistically rigorous variable importance assessment with lower variance than existing methods, as shown in theoretical and empirical studies.
Causal machine learning holds promise for estimating individual treatment effects from complex data. For successful real-world applications of machine learning methods, it is of paramount importance to obtain reliable insights into which variables drive heterogeneity in the response to treatment. We propose PermuCATE, an algorithm based on the Conditional Permutation Importance (CPI) method, for statistically rigorous global variable importance assessment in the estimation of the Conditional Average Treatment Effect (CATE). Theoretical analysis of the finite sample regime and empirical studies show that PermuCATE has lower variance than the Leave-One-Covariate-Out (LOCO) reference method and provides a reliable measure of variable importance. This property increases statistical power, which is crucial for causal inference in the limited-data regime common to biomedical applications. We empirically demonstrate the benefits of PermuCATE in simulated and real-world health datasets, including settings with up to hundreds of correlated variables.