Double Machine Learning for Estimating Time-Varying Delayed and Instantaneous Effects Using Digital Phenotypes

2025-09-08

AI Summary

This study addresses the challenge of estimating dynamic heterogeneous treatment effects from digital phenotyping data in mobile health (mHealth), where treatments exhibit time-varying properties and delayed responses, and observational data suffer from confounding bias. We propose a novel time-series causal inference method based on double machine learning, the first to extend the double ML framework to dynamic treatment effect decomposition. Our approach separately identifies immediate and delayed treatment effects while satisfying Neyman orthogonality and asymptotic normality—ensuring robustness to model misspecification and enabling consistent, efficient estimation under confounding. It thus mitigates confounding bias and precisely captures individual-level heterogeneity. Empirical validation on real-world mHealth data from Parkinson’s disease patients reveals that treatment effects significantly vary with age and motor fluctuation status. Simulation studies further demonstrate high accuracy and robustness even in small-sample settings.

Technology Category

• Machine Learning: Causal Learning
• Intelligent Robots: State Estimation
• Humans and AI: Human-in-the-loop Machine Learning

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Mobile health (mHealth) leverages digital technologies, such as mobile phones, to capture objective, frequent, and real-world digital phenotypes from individuals, enabling the delivery of tailored interventions to accommodate substantial between-subject and temporal heterogeneity. However, evaluating heterogeneous treatment effects from digital phenotype data is challenging due to the dynamic nature of treatments and the presence of delayed effects that extend beyond immediate responses. Additionally, modeling observational data is complicated by confounding factors. To address these challenges, we propose a double machine learning (DML) method designed to estimate both time-varying instantaneous and delayed treatment effects using digital phenotypes. Our approach uses a sequential procedure to estimate the treatment effects based on a DML estimator to ensure Neyman orthogonality. We establish the asymptotic normality of the proposed estimator. Extensive simulation studies validate the finite-sample performance of our approach, demonstrating the advantages of DML and the decomposition of treatment effects. We apply our method to an mHealth study on Parkinson's disease (PD), where we find that the treatment is significantly more effective for younger PD patients and maintains greater stability over time for individuals with low motor fluctuations. These findings demonstrate the utility of our proposed method in advancing precision medicine in mHealth studies.

Problem

• Estimating time-varying delayed and instantaneous treatment effects
• Addressing confounding factors in observational digital phenotype data
• Evaluating heterogeneous treatment effects accommodating temporal and between-subject heterogeneity

Innovation

• Double machine learning for time-varying effects
• Sequential procedure ensuring Neyman orthogonality
• Estimating instantaneous and delayed treatment effects