Risk-Aware Quantile Learning for Personalized Dynamic Treatment Regimes

2026-08-05

AI Summary

This work addresses the challenge of simultaneously optimizing tail outcomes, controlling treatment risk, and handling multiple treatment options in dynamic treatment regimes. The authors propose Risk-aware Quantile Dynamic Treatment Regimes (RQDTR), which uniquely integrates quantile optimization with explicit risk constraints within a multi-category decision framework. The method combines quantile regression, risk-constrained optimization, and angle-based multicategory decision rules to construct three interpretable submodels, and it is supported by theoretical guarantees including Fisher consistency and finite-sample excess risk bounds. Experiments on both synthetic data and real-world datasets—All of Us for depression and MIMIC-III for sepsis—demonstrate that RQDTR significantly improves tail efficacy while achieving a superior benefit–risk trade-off.

Abstract

Sequential clinical decision-making often involves more than maximizing average efficacy. Clinicians may need to simultaneously optimize clinically relevant tails of the outcome distribution, control treatment-related risk, and choose among multiple treatment options. Existing quantile dynamic treatment regime (DTR) methods capture distributional features of treatment outcomes but remain largely restricted to efficacy-only objectives and binary treatments. To address these limitations, we propose Risk-Aware Quantile Dynamic Treatment Regimes (RQDTR), a unified framework that optimizes a prespecified quantile of the cumulative potential outcome while explicitly incorporating treatment-related risk. We also develop an angle-based formulation for jointly learning decision rules across multiple treatment categories. Our framework includes three interpretable subclasses: efficacy-only quantile learning, constraint-based learning with population-level risk control, and utility-based learning through a composite benefit-risk utility. Theoretically, we establish identification and oracle equivalence, Fisher consistency of the smoothed surrogate, consistency of the estimated regime, and finite-sample performance error rates. Extensive simulation studies and applications to All of Us major depressive disorder and MIMIC-III sepsis data demonstrate that RQDTR improves tail-oriented efficacy and achieves more favorable benefit-risk trade-offs than existing quantile DTR methods.

Problem

• dynamic treatment regimes
• quantile learning
• risk-aware decision making
• benefit-risk trade-off
• multi-category treatment

Innovation

• Risk-Aware Quantile Learning
• Dynamic Treatment Regimes
• Angle-Based Classification
• Benefit-Risk Trade-off
• Multi-Treatment Decision Making