|
Abstract:
|
Developing data-driven optimal treatment strategies may benefit individual patients, care providers, and other stakeholders by improving outcomes and lowering healthcare costs. An optimal treatment decision rule maximizes a population-level distributional summary such as the expected value of a clinical outcome. The majority of existing methods for estimating optimal treatment decision rules assume complete data are observed, which frequently does not occur in practice. The Social incentives to Encourage Physical Activity and Understand Predictors (STEP UP) trial was a randomized trial comparing multiple interventions that aimed to increase daily step counts among employees at a large professional services company. Using simulations, we compare several estimators of optimal decision rules when data are missing at random and use our findings to guide estimation of an optimal treatment decision rule based on data from the STEP UP trial.
|
