Individualized Prediction Bands in Causal Inference with Continuous Treatments

2025-11-19

AI Summary

Existing methods primarily focus on estimating conditional means or medians, failing to fully characterize the uncertainty of individual dose–response curves under continuous treatments. Method: We propose a personalized causal inference framework for continuous exposures, framing causal effect estimation as a covariate shift problem. Our approach innovatively integrates weighted conformal prediction with quantile regression to construct statistically valid, individualized prediction bands—without requiring strong modeling assumptions. Contribution/Results: The method robustly handles covariate shift and provides rigorous uncertainty quantification for individual-level dose–response curves. Through simulations and real-world analysis of smoking behavior and healthcare expenditures, we successfully estimate the additional medical costs attributable to sustained smoking per individual, along with valid confidence intervals. This work bridges a critical theoretical and methodological gap in uncertainty quantification for personalized causal inference under continuous treatments.

Technology Category

Reasoning under Uncertainty: CausalityMachine Learning: Calibration & Uncertainty QuantificationCognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning

Application Category

User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systemsResponsible Web: Human-perceived consequences of algorithmic deployment on the webSearch and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Individualized treatments are crucial for optimal decision making and treatment allocation, specifically in personalized medicine based on the estimation of an individual's dose-response curve across a continuum of treatment levels, e.g., drug dosage. Current works focus on conditional mean and median estimates, which are useful but do not provide the full picture. We propose viewing causal inference with a continuous treatment as a covariate shift. This allows us to leverage existing weighted conformal prediction methods with both quantile and point estimates to compute individualized uncertainty quantification for dose-response curves. Our method, individualized prediction bands (IPB), is demonstrated via simulations and a real data analysis, which demonstrates the additional medical expenditure caused by continued smoking for selected individuals. The results demonstrate that IPB provides an effective solution to a gap in individual dose-response uncertainty quantification literature.

Problem

• Individualized dose-response uncertainty quantification for continuous treatments
• Leveraging covariate shift with weighted conformal prediction methods
• Providing prediction bands beyond conditional mean and median estimates

Innovation

• Leverages weighted conformal prediction for uncertainty quantification
• Treats continuous treatment as covariate shift problem
• Computes individualized prediction bands for dose-response curves