Making evidence go further: Advancing synergy between agent-based modeling and randomized control trials

Abstract

With the world facing complex health challenges and limited resources, it's critical that we maximize the impact of evidence to find effective solutions. Randomized control trials (RCTs) are widely used in health sciences to establish causal relationships between interventions (such as new medications) and outcomes (such as a reduction in a targeted disease condition like obesity). However, RCTs have limitations that constrain the impact of the evidence they generate (1, 2). Agent-based modeling (ABM) is a computational methodology for representing simulated dynamic pathways that connect exposures to outcomes (3-5) across diverse and interacting individuals and contexts [such as how retail density, price information, and individual characteristics work together to shape tobacco use patterns across very different cities (6)]. ABM can also consider potential intervention counterfactuals-that is, projected alternative states of the world representing results of untested interventions or adaptations of intervention components (3, 4). For example, ABM has been used during the COVID-19 pandemic to understand the potential impact of interventions being considered to contain spread of the virus (7, 8). Still, a lack of empiric data can limit the utility of ABM. When used together, ABM and RCTs offer powerful synergy, each addressing key limitations of the other. We argue that more research should pair these two methods to increase significant research discoveries. When used together, ABM and RCTs offer powerful synergy, each addressing key limita­ tions of the other. Image credit: Shutterstock/ Jelena Sebik.