Research Article

Beyond Propensity Score: The SAHR, Subgroup-Aware Holistic Ridge for Causal Inference in Observational Studies

Authors

  • Mahdi Hamisheh Bahar * Department of Industrial Engineering, Sharif University of Technology, Tehran, Iran.

https://doi.org/10.22105/raise.v3i2.97

Abstract

Causal inference is critical for decision-making in healthcare, economics, and policy, but it faces challenges such as confounding, heterogeneous treatment effects, and scalability in high-dimensional settings. Traditional methods like propensity scores often struggle with these complexities, particularly when treatment effects vary across subpopulations or datasets are large. To address these limitations, we propose Subgroup-Aware Holistic Ridge (SAHR), a novel framework that utilize subgroup analysis, kernel approximation, and adaptive covariate balancing based on BNN or regression covariates for robust causal inference. SAHR identifies subgroups with heterogeneous treatment effects using Gaussian Mixture Models (GMM), ensures scalability via the Nyström method, and balances covariates adaptively through a Holistic ridge regression framework. Experiments on benchmark datasets demonstrate that SAHR outperforms state-of-the-art methods in predictive accuracy, subgroup identification, and covariate balancing. By addressing key challenges in observational studies, SAHR advances causal inference methodology, offering an interpretable tool for applications in personalized medicine, and policy evaluation.

Keywords

Causal inference Subgroup analysis Covariate balancing Gaussian mixture models BNN Propensity scores Regression
References (28)
  1. [1] Rosenbaum, P. R., & Rubin, D. B. (1983). The central role of the propensity score in observational studies for causal effects. Biometrika, 70(1), 41–55. https://doi.org/10.1093/biomet/70.1.41

  2. [2] Austin, P. C. (2011). An Introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate behavioral research, 46(3), 399–424. https://doi.org/10.1080/00273171.2011.568786

  3. [3] Chipman, H. A., George, E. I., & McCulloch, R. E. (2010). Bart: Bayesian additive regression trees. Annals of applied statistics, 4(1), 266–298. https://doi.org/10.1214/09-AOAS285

  4. [4] Bang, H., & Robins, J. M. (2005). Doubly robust estimation in missing data and causal inference models. Biometrics, 61(4), 962–973. https://doi.org/10.1111/j.1541-0420.2005.00377.x

  5. [5] Gretton, A., Borgwardt, K. M., Rasch, M. J., Schölkopf, B., & Smola, A. (2012). A kernel two-sample test. The journal of machine learning research, 13, 723–773. https://dl.acm.org/doi/10.5555/2188385.2188410

Article Metrics

How to Cite

Hamisheh Bahar, M. . (2026). Beyond Propensity Score: The SAHR, Subgroup-Aware Holistic Ridge for Causal Inference in Observational Studies. Research Annals of Industrial and Systems Engineering, 3(2), 134-145. https://doi.org/10.22105/raise.v3i2.97

Similar Articles

1-10 of 64

You may also start an advanced similarity search for this article.