Incorporating Auxiliary Variables to Improve the Efficiency of Time-Varying Treatment Effect Estimation

Area of Science:

• Digital Health
• Biostatistics
• Mobile Health (mHealth)

Background:

• Smart devices and wearables are increasingly used for digital health interventions.
• Micro-randomized trials (MRTs) are crucial for evaluating mobile health (mHealth) intervention effectiveness.
• Traditional methods for covariate adjustment in randomized controlled trials (RCTs) are not fully explored for longitudinal MRT data.

Purpose of the Study:

• To propose an efficient method for assessing time-varying causal effect moderation in MRTs.
• To leverage auxiliary variables to improve the efficiency of moderated causal effect estimation.
• To address the gap in applying covariate adjustment techniques to longitudinal MRT data.

Main Methods:

• Utilized Neyman Orthogonality principles for method development.
• Proposed a straightforward method incorporating auxiliary variables to enhance efficiency.
• Compared robust standard errors of the proposed method against a benchmark.

Main Results:

• The proposed method demonstrated efficiency gains in simulation studies.
• Efficiency improvements were also observed in the analysis of the Intern Health Study data.
• The method effectively incorporates auxiliary variables for improved causal effect estimation.

Conclusions:

• The proposed method offers a more efficient way to analyze moderated causal effects in MRTs.
• Incorporating auxiliary variables is a valuable strategy for improving mHealth intervention effect estimation.
• This research contributes to more robust evidence generation for digital health interventions.