PHASE funding is supporting research projects that aim to foster collaborations between those with agent-based modelling skills and those researching and tackling non-communicable diseases (NCDs) to support the development of agent‑based modelling projects that provide insights and evidence to tackle NCDs.
Leveraging local policies to improve diet: modelling the role of local interventions impacting the food environment
Professor Martin O’Flaherty, University of Liverpool
Improving dietary quality in urban areas is increasingly important in the UK not only from economic or environmental considerations, but also from health and equity perspectives. There is growing consensus among local and national policy makers that modification of neighbourhood and environmental features offers an effective and feasible opportunity for promoting healthy diets. However, there is a lack of sufficient evidence to support action to address these policy decisions, including which policies are effective or equitable, limiting our ability to support or implement healthy food environments.
This project focusses on the role of the food environment in influencing key indicators of diet quality, including intake of fruits and vegetables and consumption of take-away meals. The project aims to support local strategies to improve urban food environments by developing a spatially explicit agent-based model based on identified key local drivers of poor diet and their connections to the food environment. Agent-based models offer an effective solution to addressing the complex and reinforcing determinants of diet-related non-communicable diseases (e.g., neighbourhood-level dietary patterns and income lead to food retail location decisions). The project will build on previous agent-based modelling studies incorporating aspects of the food environment and how key actors, including retail and fast-food outlets, households and individuals, interact with it. The model developed through this project will explore key decisions made by retailers, including open/close times, and basic inventory (e.g., stocking of fruit and vegetables). Stakeholder engagement will identify potential policy windows and feasible interventions for testing, and the model will be used to test these co-produced policy scenarios for diet outcomes.
Chronic pain, mental health and employment:
the role of firms, workers and the state
Professor Matteo Richiardi, University of Essex
There are causal links between mental health, chronic pain and employment: mental health issues reduce productivity, ambitions in job choice, and increase non-employment; and chronic pain reduces productivity and increases absenteeism and unemployment. At the same time, labour market events such as job displacement are associated with higher mortality rates and deterioration in health. These complex interactions and feedback loops between health and employment are not easily captured by standard econometric methods, and the compounding of inequalities in health and socio-economic status can lead to different conclusions when they are analysed jointly. With their focus on interaction effects, agent-based models are well suited for the task.
This project brings together policy makers, public health and simulation experts to develop a proof-of-concept agent-based model linking mental health, chronic pain, and employment outcomes, focusing in particular on the role of non-monetary characteristics of jobs. The project will contribute to understanding the importance of interactions between two widespread non-communicable diseases (NCDs) and employment. It will look at which groups are disproportionately affected by these diseases, how their labour market outcomes are affected, who bears the costs, and who could invest in decreasing prevalence of these NCDs. This will allow the identification of new policies that can address inequalities and incidence of mental health issues and chronic pain, by understanding the incentives to invest in preventing and alleviating the diseases by the individuals, firms and the state.
ABM-based Land Use-Transport Interaction (LUTI) simulation: healthier urban development and healthier travel behaviour for Greater Manchester
Dr Heeseo Rain Kwon, University of Reading
This project focusses on non-communicable disease (NCD) reduction through the creation of healthier urban development, in particular land use and transport. Obesity and air pollution-related NCDs are associated with the built environment and physical activity where individual resident and urban development behaviours interact. By using agent-based modelling, this project aims to understand the complex non-linear and recursive patterns emerging at the system-level (e.g., land use, occupancy, and mobility culture changes) and feedback loops between healthier urban development and travel behaviour.
The project will work with Greater Manchester Combined Authority and Transport for Greater Manchester to model residents’ active mobility (walking, cycling and bus). The model will include evidence on resident travel demand and real estate urban land use/occupation change, including working from home, in active mobility modelling by experimenting with the feedback loop between healthier urban development and healthier travel behaviour. Different policy scenarios will be tested to support NCD and inequality prevention.
A heterogenous agents framework for tobacco availability interventions
Dr Valerio Restocchi, University of Edinburgh
Smoking is currently the leading cause of preventable death and health inequalities in most developed countries, and is responsible for almost 100,000 deaths every year in the UK alone. Smoking is more prevalent among deprived communities, and this fuels a vicious circle of poverty and unfavourable health outcomes. The UK and Scottish Governments have identified a reduction in availability of tobacco products as a key opportunity to the creating a smoke-free society by the 2030’s. However, the specific policy interventions that might reduce the availability of tobacco products and their impact on smoking initiation and cessation remain unclear.
This project aims to develop a framework that allows evidence users to test tobacco control policies while measuring the impact of such policies on health and inequality. Using this framework, evidence users will be able to test and choose those policies that minimise inequality while still being effective on the entire population. Working alongside policymakers and practitioners, an agent-based model will be developed that captures the complex interactions between tobacco availability, social contagion, and inequality. The model will then be used to develop and test tobacco availability interventions, assessing the impact of these policies on both levels and equity of smoking. Finally, the model will be packaged into an easy-to-use interactive web-based application to allow the framework to be easily used by policy makers and researchers with no specialised knowledge of agent-based modelling and programming.
Developing a proof-of-concept agent-based model of the relationship between food advertising and food choices in England
Dr Charlotte Buckley, University of Sheffield
National and local public health policymakers need to identify policies that reduce obesity at minimum cost whilst reducing health inequalities. The modern food environment contains many contextual food cues, including advertisements for highly palatable energy dense foods often containing large amounts of sugar, salt, and fat (HFSS). Recent evidence evaluating the restriction of advertising of these foods on the London transport network suggested that this could impact purchasing behaviour. As food advertisements may contribute to the embedding of social norms in environments, a ban on food advertising may be a promising policy for reducing intake of HFSS food, and thus reducing obesity, through changing social norms about consumption. This project will develop an agent-based model to explore how exposure to food advertisements and social norms impacts food choices in different population sub-groups.
The project will take an existing conceptual framework for modelling the mechanisms underlying alcohol use in the population (Mechanism-Based Social Systems Modelling [MBSSM] framework) that incorporates social norm theories, and adapt this to develop an agent-based mode of food advertising and food choice. The model parameters will be calibrated by looking at food purchasing trends before and after the London advertising ban and control group purchasing in Northern England. The model will then be used to generate insights into health and inequality outcomes.
Modelling the spread of multiple behavioural risk factors for cardiovascular disease in social networks using an agent-based model
Prof. Nathan Griffiths, University of Warwick
Prof. Oyinlola Oyebode, Queen Mary University of London
Dr James Archbold, University of Warwick
Many public health interventions aim to change behaviour now to prevent diseases later, which means proxy and process measures are used to evaluate success. An alternative is to conduct long follow-ups of participants, but this increases costs and reduces timeliness of findings. Agent-based models offer an alternative, allowing in silico prediction of disease development based on behaviours, providing evidence on which strategies might be most successful in reducing non-communicable disease prevalence.
This project aims to construct and validate an agent-based model of how the major cardiovascular disease risk factors (e.g., tobacco-use, unhealthy diet, physical inactivity, harmful use of alcohol) spread through social networks within a population, leading to the development of cardiovascular diseases.
Once completed, the agent-based model will provide insight into who the key individuals are within community networks, and their characteristics. Behaviour change by these individuals could drive wider beneficial changes. The model will also help to identify key behaviours or combinations of behaviours that have the greatest potential to impact disease burden, and crucially reduce health inequalities, allowing future public health interventions to focus on these identified targets.
Using machine learning as a surrogate model for agent-based simulations
C. Angione, E. Silverman, E. Yaneske
PLoS One, 2022
Network structure influence on simulated network interventions for behaviour change
J. Badham, F. Kee, R.F. Hunter
Social Networks, 2021
Situating agent-based modelling in population health research
E. Silverman, U. Gostoli, S. Picascia, J. Almagor, M. McCann, R. Shaw, C. Angione
Emerging Themes in Epidemiology, 2021
PHASE: Facilitating agent-based modelling in population health
E. Silverman, U. Gostoli
Proceedings of the 2020 Winter Simulation Conference
Social and Child Care Provision in Kinship Networks: an Agent-Based Model
U. Gostoli, E. Silverman
PloS One, 2020
Sound behavioural theories, not data, is what makes computational models useful
U. Gostoli, E. Silverman
Review of Artificial Societies and Social Simulation, 2020
Modelling social care provision in an agent-based framework with kinship networks
U. Gostoli, E. Silverman
Royal Society Open Science, 2019