Incorporating nonparametric methods for estimating causal excursion effects in mobile health with zero-inflated count...

Area of Science:

• Mobile Health
• Biostatistics
• Causal Inference

Background:

• Micro-randomized trials are crucial for developing adaptive mobile health interventions.
• Analyzing intervention effects and covariate moderation is key.
• Existing causal inference methods often overlook count data, particularly zero-inflated outcomes.

Purpose of the Study:

• To extend the concept of causal excursion effect to zero-inflated count outcomes.
• To develop novel nonparametric estimation approaches for analyzing such data.
• To address the gap in analyzing count data within micro-randomized trials.

Main Methods:

• Revisiting the causal excursion effect for zero-inflated count data.
• Introducing novel nonparametric estimation techniques.
• Establishing bidirectional asymptotics for proposed estimators.
• Conducting simulation studies for performance evaluation.

Main Results:

• The proposed methods are evaluated through simulation studies.
• The methods are demonstrated using data from the Drink Less micro-randomized trial.
• The study provides a framework for analyzing complex count outcomes in mobile health.

Conclusions:

• Novel statistical methods are presented for causal inference with zero-inflated count data in micro-randomized trials.
• The developed techniques offer valuable tools for mobile health intervention development.
• The application to the Drink Less trial highlights practical utility.