Using agent-based models to address non-communicable diseases: a review of models and their application to policy

Abstract

BackgroundAgent based models are a computational methodology in which systems of simulated heterogeneous agents interact with one another and their environment; they are a research tool with the potential to provide greater understanding of the complex, interdependent, and systemic determinants of population health challenges, particularly when co-produced with the decision makers, practitioners, and public who understand and experience these challenges from a variety of perspectives. Although agent-based modelling is becoming more widely used in health research, this methodology is currently underutilised in non-communicable disease prevention. We aim to highlight the potential role of agent-based modelling in supporting policy and practice decision-making in non-communicable disease prevention, using an obesity example to show how an agent-based model can capture the social influences in diet. MethodsWe identified agent-based models addressing non-communicable disease prevention from PubMed, ScienceDirect, Google Scholar, and Web of Science, using the search terms “agent-based model” and “public health” between Jan 1, 2000, and Dec 31, 2022. Case studies were selected that best illustrated the use of agent-based models for UK public health challenges. Selection criteria also included open access code documentation and models that reported core methodologies. FindingsWe identified three key case studies that best exemplified the use of agent-based models for addressing non-communicable disease prevention. A further three core sociological agent-based model case studies that were seen as core to methodology were also identified. Although we did not find any agent-based models that had been used for policy implementation to date, we did identify a model of obesity spread in networked agents. This model simulated how individuals influence each other with respect to food consumption and physical activity. We deconstructed this model to find that it is a good didactic example of agent-based methodology and shows the potential for agent-based models to be used to inform policy and practice decision-making in the future. InterpretationWell-designed agent-based models that are developed in collaboration with public health stakeholders have the potential to provide new insights into non-communicable disease prevention and provide an opportunity to test policy scenarios in silico before they are applied in the real world. FundingPopulation health Agent-based Simulation nEtwork (PHASE).