Modelling population health and behaviour change in the transition to a low carbon society

Many policy decisions rely on the results of computer models. Computer models are needed because there are many things we cannot directly observe. Better models should support better decision making. I will develop models in two areas where behaviour change could benefit both health and the environment; physically active travel and healthier diets.

Being more physically active, even just through walking or cycling short trips, can help reduce the risk of diseases, including heart disease, stroke, diabetes, breast cancer, and dementia. I have created a computer model to estimate how much walking and cycling more and driving less could improve public health and reduce greenhouse gas emissions. This model includes not just health benefits, such as from physical activity, but also harms from air pollution and road traffic injuries. It has highlighted the potential for cycling to replace many of our everyday car trips. The model is being used in various countries and I am currently using it to estimate the health effects from the London Barclays Cycle Hire Scheme.

In this fellowship I will improve the way the model deals with changes over time, and use the model to find out where filling gaps in the evidence could bring the greatest benefit. I will also work with policy makers to make it more user friendly and I will work with colleagues to estimate the longer term health effects from building different kinds of transport infrastructure (e.g. cycle paths).

To tackle the problems of overweight and obesity we need to not just increase physical activity but also change what we eat. I will develop another computer model that brings together physical activity and diet to study obesity. This model will also look at the cost of healthier and more sustainable diets and see which interventions might improve health, sustainability, and reduce health inequalities.

Knowing what would happen if people adopted more healthy lifestyles is not enough; we also need to know how society can better support healthier lifestyles. I will use two other computer modelling methods, agent based modelling and system dynamic modelling, to help explore why and how people's behaviour might change. Creating useful computer models of society requires bringing together many different sources of information. I will use information including scientific studies and the expert knowledge of key participants.

Agent based models are used in computer games, e.g. The Sims, as well as academic work. They create virtual agents (such as people) who interact with each other. My agent based model will look at how people's travel choices are shaped by other people: for example, people may be more influenced by those they see as similar to themselves. To make the model more accurate I will draw on research that has studied how people influence and learn from each other in local contexts like workplaces.

I will use system dynamic modelling to look at the 'big picture', asking how cycling could become a mass mode of travel and how what we eat in Britain could change. The models will build on the expertise of academics, policy-makers and advocates, through workshops, where these stakeholders will help me develop the model.

When the models have been developed, we will run them many times, test them against real world data, and try out different policies. The results of the models will be used to try and understand better how policies might play out in the long-run and to test how robust policies are to outside events (e.g. oil price rises).

This fellowship will continue my training as an independent public health researcher. I will be trained in different modelling approaches and analysis of qualitative data, and learn from working with more experienced researchers. Overall this research stands to benefit transport and food policy making and potentially other policy areas where government is interested in behaviour change, such as alcohol, and tobacco.

Modelling is needed in public health because we want to estimate that which we cannot directly observe. Models should use the best evidence, represent uncertainty, and be as transparent as possible. The research will have two foci, modelling health impacts and modelling behaviour change.

1) Modelling health impacts: My macro simulation modelling has highlighted the untapped potential of cycling in the UK both to benefit public health and reduce greenhouse gas emissions. In this fellowship I will collaborate with other researchers and policy makers to use the model to answer practical questions, to improve the model, and make it more user friendly. I will also develop a new physical activity and diet model to answer questions about how interventions, including price changes, could affect population health and health inequalities.

2) Modelling behaviour change: I will use two types of computer simulation models that can aid understanding of complex systems; agent based modelling (ABM) and system dynamic (SD). The ABMs will focus on how local cycling cultures spread through social influence, and how this affects policy. Behavioural rules will be developed through re-analysis of qualitative data. SD models, created through a group model building process, will be developed to answer questions on the transition to healthier and more sustainable food and transport systems. Insights from each model will inform the other.

Across all the simulation models I will use exploratory modelling and analysis to test robustness of policy options across plausible scenarios.

This fellowship will continue my transition into a fully independent public health researcher, with core expertise in modelling, and produce innovative models with wider application to problems of behaviour change. Formal training will include courses on ABM, SD modelling, and qualitative data analysis. Informal training will occur through collaboration with researchers with a range of expertise.

This fellowship stands to primarily benefit local, national and international policy-makers and practitioners from areas including health, environmental policy, transport, and food policy. It also stands to benefit organisations working on active travel (NGOs and businesses) and organisations seeking to promote active travel in the workplace. Most significantly the research offers the potential to reduce the burden of diseases associated with physical inactivity and poor diet and to assist with the move to a more sustainable society.

Modelling and models are too often the preserve of technical experts. I would argue that models should help us understand the complexity and uncertainty of the world not make it more opaque. Part of the aim of the research is to open up modelling through participatory approaches and improving modelling literacy of stakeholders.

Policy makers will benefit directly from the model results and through using the models to consider different policy scenarios and assess proposed changes. The models will improve the ability to predict likely effects of interventions before deciding how to invest resources in real interventions; a major benefit for resource constrained public authorities. The move of public health functions into local authorities in 2013 will bring together responsibilities for transport and health. This research can provide evidence based on integrated rather than 'silo thinking'.

Specific technical benefits will improve the usefulness of the models, e.g. the move to a dynamic model will allow policy-makers to compare potential health impacts over time from different scenarios and policies. This can help with mitigating negative effects; for example with more cycling road traffic injuries might first rise then fall. Another benefit would arise from the proposed integration of my models within the transport and land use models widely used in planning.

The SD and ABM models address gaps in policy-relevant knowledge. Specifically the SD modelling will increase understanding of likely longer-term policy impacts, including unintended consequences, in the policy areas of transport and food. The ABM modelling will develop knowledge about how micro-level social relations shape responses to policies. Both have broader implications for knowledge about public policy and its impacts, in particular on the uptake of healthier lifestyles.

Policy-makers will benefit from improved understanding of modelling outputs. For example, by engaging stakeholders in the exploratory modelling I will seek to move beyond the presentation of best estimate results by addressing questions such as risk management with respect to highly negative outcomes and the sensitivity of results to structural and parametric uncertainty.

NGOs working around health, diet, and transport are potential additional non-academic beneficiaries. Many of these organisations are increasingly involved in service delivery. Modelled results could assist them in providing better services and lead to a stronger evidence base for their work.

A recent scoping report for the WHO, which I co-authored, highlighted the potential for job creation in sustainable travel, including cycling (e.g. bike retail and public bike schemes). If industry has a better understanding of cycling trends then they are better placed to invest. The modelling of cycling in the workplace, via the ABM, stands to provide benefit to employers and employees looking to make travel within their organisations healthier and more sustainable. An increase in cycling may also lead to reductions in the economic costs from congestion.

As stakeholders will be engaged throughout, these benefits will be realised throughout the project. The models will continue to produce impact after the project, as will collaborations across policy areas. Ultimately this fellowship seeks to benefit population health, while simultaneously supporting environmental sustainability.