Abstract:
|
We consider the problem of designing micro-randomized trials (MRTs) in the mobile health context. MRTs are cutting-edge designs in which trial participants are sequentially randomized to available behavioral interventions (e.g., motivational messages to promote healthy behaviors) on numerous occasions throughout the study course. Interventions are typically multi-factorial in this setting (e.g., different types of messages, each with different possible versions). However, the current literature only considers two-level factors, and only continuous outcomes. In this talk, we will discuss some design and sample size considerations for MRTs where the intervention factors may have more than two levels, and the outcome may not be continuous in nature (e.g., ordinal outcome). Simulation results will be shown to validate the proposed design and sample size calculation approach. This work is motivated by a study of a mobile application designed to help people take nutritious diet and increase physical activity, which will be briefly discussed in the talk.
|