Dynamic Causal Mediation Analysis for Intensive Longitudinal Data

2025-06-24

AI Summary

Causal mediation analysis in dense longitudinal data—characterized by time-varying exposures, time-varying mediators, and distal outcomes—is challenged by high-dimensional exposures, complex dynamic pathways, and intermediate confounding, rendering conventional approaches (e.g., path-specific effects) invalid. Method: We propose natural direct and indirect incentive effects (NDIEs), which decompose the total incentive effect and precisely quantify contributions from the most proximal mediator at each stage. Leveraging efficient influence functions, we construct multiply robust estimators integrating cross-fitting and machine learning, tailored for settings with known intervention mechanisms (e.g., micro-randomized trials). Contribution/Results: The estimators are theoretically guaranteed to be consistent and asymptotically normal. Empirical validation demonstrates their effectiveness in the HeartSteps mobile health trial and the SleepHealth observational study, enabling rigorous, interpretable causal mediation inference under realistic longitudinal complexity.

Technology Category

• Machine Learning: Causal Learning
• Multiagent Systems: Mechanism Design
• Intelligent Robots: State Estimation

Application Category

• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• User Modeling, Personalization and Recommendation: Studies of user behavior, including longitudinal effects of personalized systems
• Security and Privacy: Large-scale security measurements

Abstract

Intensive longitudinal data, characterized by frequent measurements across numerous time points, are increasingly common due to advances in wearable devices and mobile health technologies. We consider evaluating causal mediation pathways between time-varying exposures, time-varying mediators, and a final, distal outcome using such data. Addressing mediation questions in these settings is challenging due to numerous potential exposures, complex mediation pathways, and intermediate confounding. Existing methods, such as interventional and path-specific effects, become impractical in intensive longitudinal data. We propose novel mediation effects termed natural direct and indirect excursion effects, which quantify mediation through the most immediate mediator following each treatment time. These effects are identifiable under plausible assumptions and decompose the total excursion effect. We derive efficient influence functions and propose multiply-robust estimators for these mediation effects. The estimators are multiply-robust and accommodate flexible machine learning algorithms and optional cross-fitting. In settings where the treatment assignment mechanism is known, such as the micro-randomized trial, the estimators are doubly-robust. We establish the consistency and asymptotic normality of the proposed estimators. Our methodology is illustrated using real-world data from the HeartSteps micro-randomized trial and the SleepHealth observational study.

Problem

• Analyzing causal mediation in intensive longitudinal data
• Addressing complex mediation pathways with time-varying factors
• Proposing robust estimators for natural direct and indirect effects

Innovation

• Novel natural direct and indirect excursion effects
• Multiply-robust estimators with machine learning
• Doubly-robust for known treatment mechanisms