|
Activity Number:
|
2
- When Causal Inference Meets Reinforcement Learning: The Story of Mobile-Delivered Interventions
|
|
Type:
|
Invited
|
|
Date/Time:
|
Sunday, August 8, 2021 : 1:30 PM to 3:20 PM
|
|
Sponsor:
|
IMS
|
|
Abstract #316923
|
|
|
Title:
|
Assessing Time-Varying Causal Effects in the Presence of Cluster-Level Treatment Effect Heterogeneity
|
|
Author(s):
|
Jieru Hera Shi and Zhenke Wu* and Walter Dempsey
|
|
Companies:
|
University of Michigan, Ann Arbor and University of Michigan, Ann Arbor and University of Michigan
|
|
Keywords:
|
micro-randomized trials;
sequential randomization;
mobile health;
causal inference;
mobile-delivered interventions
|
|
Abstract:
|
Micro-randomized trials (MRT) sequentially randomize participants to potentially receive interventions at hundreds or thousands of decision points. For example, mobile applications enabled the delivery of push notifications to support behavioral change and maintenance. Excursion causal effect estimands have been defined along with inferential procedures under key assumptions of between-subject independence and non-interference. Deviations from such assumptions often occur, which if unaccounted for may result in bias and overconfident variance estimates. This work defines and infers the excursion causal effects under such deviations and when the moderation effect of interest depends on cluster-level moderators. We illustrate the utility of methods to estimate the effect of push notifications on mood by analyzing data from multi-institution cohorts of first year medical residents in the United States. Interactions among residents in the same specialty induce outcome dependence and intervention interference. The approach paves the way for planning just-in-time adaptive interventions by factoring realistic social network structures into MRT design.
|
Authors who are presenting talks have a * after their name.
