Assessing time-varying moderation effects in micro-randomized trials
Walter Dempsey

Advances in wearable technologies and health interventions delivered by smartphones have greatly increased the accessibility of mobile health (mHealth) interventions. Micro-randomized trials (MRTs) are designed to evaluate how intervention effects change over time and are influenced by individual characteristics or context. In MRTs, scientists often have many variables that they are interested in assessing as potential moderators. High-dimensional analyses often lack interpretability, while marginal analyses yield many false positives. In this talk, we will demonstrate a simple, two-step method for inference on time-varying causal effect moderation that addresses the limitations of both high-dimensional and marginal analyses. Through simulations and real data analyses, we show that our method consistently achieves valid coverage rates and is a strong data analytic framework for exploratory analysis in MRTs.

Control Systems Engineering for Optimizing "Just-in-Time" Adaptive Behavioral Interventions
Daniel Rivera

Control systems engineering using system identification (i.e., a methodology for estimating dynamic models from data) and model predictive control (MPC; a control design technique for constrained, multivariable dynamical systems) offers a powerful framework for developing “just-in-time” adaptive behavioral interventions. The control optimization trial (COT) formalism combines these methods to create data-driven, personalized interventions that dynamically adjust to an individual's changing needs and context. The ongoing YourMove (NCT05598996), Healthy Mom Zone 2 (NCT05807594), and FLASH (NCT06244888) clinical trials are evaluating this approach for physical activity, gestational weight regulation, and weight loss maintenance, respectively. Benefits, challenges, and some practical considerations involved in effectively conducting COT trials will be illustrated with examples taken from these three interventions.