Learning Interactions Between Continuous Treatments and Covariates with a Semiparametric Model

2025-05-06

AI Summary

This paper addresses the challenge of modeling causal effects of continuous treatment variables—such as drug dosage and timing—on binary outcomes, where conventional methods relying on strong linear assumptions often fail. We propose an interpretable semiparametric regression framework that disentangles baseline risk prediction from treatment–covariate interaction effects. Our approach introduces, for the first time, a dual-score structure comprising a prognostic score and a treatment-interaction score, coupled with a nonparametric link function to flexibly characterize optimal treatment levels. The method integrates heterogeneous features—including clinical, genetic, and sociodemographic data—and enjoys theoretical consistency guarantees for estimation. Empirical evaluation on the International Warfarin Pharmacogenetics Consortium (IWPC) dataset demonstrates that our framework significantly improves the accuracy of personalized warfarin dosing recommendations, thereby enhancing both the safety and efficacy of anticoagulation therapy.

Technology Category

• Reasoning under Uncertainty: Causality
• Machine Learning: Causal Learning
• Search and Optimization: Mixed Discrete/Continuous Search

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Estimating the impact of continuous treatment variables (e.g., dosage amount) on binary outcomes presents significant challenges in modeling and estimation because many existing approaches make strong assumptions that do not hold for certain continuous treatment variables. For instance, traditional logistic regression makes strong linearity assumptions that do not hold for continuous treatment variables like time of initiation. In this work, we propose a semiparametric regression framework that decomposes effects into two interpretable components: a prognostic score that captures baseline outcome risk based on a combination of clinical, genetic, and sociodemographic features, and a treatment-interaction score that flexibly models the optimal treatment level via a nonparametric link function. By connecting these two parametric scores with Nadaraya-Watson regression, our approach is both interpretable and flexible. The potential of our approach is demonstrated through numerical simulations that show empirical estimation convergence. We conclude by applying our approach to a real-world case study using the International Warfarin Pharmacogenomics Consortium (IWPC) dataset to show our approach's clinical utility by deriving personalized warfarin dosing recommendations that integrate both genetic and clinical data, providing insights towards enhancing patient safety and therapeutic efficacy in anticoagulation therapy.

Problem

• Estimating impact of continuous treatments on binary outcomes
• Overcoming strong assumptions in traditional regression models
• Developing interpretable semiparametric framework for treatment interactions

Innovation

• Semiparametric regression framework for continuous treatments
• Nadaraya-Watson regression links interpretable parametric scores
• Personalized dosing via genetic and clinical data integration