Incorporating Partial Adherence for Estimation of Dynamic Treatment Regimes

2025-12-10

AI Summary

Standard inverse probability weighting (IPW) in observational studies of dynamic treatment regimes (DTRs) suffers from inefficient estimation and suboptimal data utilization due to its restrictive binary compliance assumption. Method: We propose two novel partial-compliance-compatible IPW estimators—the first to model compliance on a continuous or graded scale—by constructing identifiable compliance-weighting functions and establishing theoretical consistency. Simulation studies demonstrate their superior finite-sample performance over conventional IPW. Contribution/Results: Applied to the ACTG175 HIV clinical trial data, our estimators significantly improve the stability and accuracy of DTR value estimation. This work breaks the paradigmatic limitation of binary compliance modeling in DTR causal inference, offering a more efficient and robust statistical framework for optimizing dynamic interventions in chronic disease management.

Technology Category

• Intelligent Robots: State Estimation
• Reasoning under Uncertainty: Stochastic Optimization
• Humans and AI: Planning and Decision Support for Human-Machine Teams

Application Category

• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Responsible Web: Data and user privacy-enhancing technologies for the Web
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Dynamic Treatment Regimes (DTRs) provide a systematic framework for optimizing sequential decision-making in chronic disease management, where therapies must adapt to patients' evolving clinical profiles. Inverse probability weighting (IPW) is a cornerstone methodology for estimating regime values from observational data due to its intuitive formulation and established theoretical properties, yet standard IPW estimators face significant limitations, including variance instability and data inefficiency. A fundamental but underexplored source of inefficiency lies in the strict binary adherence criterion that fails to account for partial adherence, thereby discarding substantial data from individuals with even minimal deviations from the target regime. We propose two novel methodologies that relax the strict inclusion rule through flexible compatibility mechanisms. Both methods provide computationally tractable alternatives that can be easily integrated into existing IPW workflows, offering more efficient approaches to DTR estimation. Theoretical analysis demonstrates that both estimators preserve consistency while achieving superior finite-sample efficiency compared to standard IPW, and comprehensive simulation studies confirm improved stability. We illustrate the practical utility of our methods through an application to HIV treatment data from the AIDS Clinical Trials Group Study 175 (ACTG175).

Problem

• Estimating dynamic treatment regimes with partial adherence
• Improving efficiency of inverse probability weighting estimators
• Addressing variance instability in observational data analysis

Innovation

• Flexible compatibility mechanisms relax strict adherence criteria
• Novel methods integrate into existing IPW workflows efficiently
• Theoretical consistency with improved finite-sample efficiency demonstrated