Agent-based models and the g-formula: comparable approaches for evaluating population intervention effects?

Agent-based models and the g-formula: comparable approaches for evaluating population intervention effects?

Epidemiologists are increasingly interested in evaluating the effect(s) of interventions on disease outcomes in populations. Various methods have been developed to estimate “population health intervention” effects. These methods seek to address both pragmatic and methodological challenges (e.g., time-dependent confounding, unintended consequences of interventions, interference, etc.) inherent in predicting the efficacy of real-world interventions. In this session, we will discuss two approaches to evaluate population health interventions: the parametric g-formula and agent-based modeling. Using applied examples, we will compare the key assumptions each method requires to estimate unbiased population health intervention effects. Finally, we will discuss the specific circumstances under which both methods may be likely to produce similar effects and consider areas where methodological progress for each method could most benefit population health research.

Session Chairs:
Magdalena Cerda, University of California, Davis
Brandon Marshall, Brown University

Presenters:
Brandon Marshall, Brown University
Jennifer Ahern, University of California, Berkeley
Stephen Mooney, Columbia University
Magdalena Cerda, University of California, Davis
Katherine Keyes, Columbia University