Simulating the impact of transport interventions on physical activity and cardiovascular disease using agent-based models

Rebalancing the travel system, so that more journeys are made using active modes, has increasingly become a priority for governments in the UK and abroad. Not only are walking and cycling good for people's physical and mental health, the reduced dependency on cars will have a positive effect on noise and air pollution, reduce the risk of road traffic injury, and prevent cardiovascular disease and obesity.

A causal relationship between the built environment and travel behaviour has been established, with the environment being strongly influenced by the car. In the worst cases, car-centric design severs communities, with roads acting as physical barriers to pedestrian movement and limiting interaction. Consequently, by improving the built environment, it is possible to remove these physical barriers to active travel. However, interventions attempting to improve the built environment do not always consider the role of individual agency, and the underlying social norms.

A social norm is an informal, unwritten understanding held collectively by members of a population as to the correct behaviour in given circumstances. Norms relating to transport behaviour, may include things such as the desirability of car ownership, or the importance of the environment.

Agent-based simulation is a bottom-up modelling process in which agents are given simple behavioural rules that define how they act in an environment and interact with one another. This is an artificial intelligence approach where simple behavioural rules can cause emergent macro-level behaviour to arise. The use of agent-based simulation allows for an inexpensive method for experimenting with public health interventions, without them having to be implemented, as well as a greater insight to the underlying interactions that occur, which may have effects on the intervention.

This project explores the following research question: how do the social norms surrounding travel behaviour minimise or maximise the effectiveness of structural active-travel interventions and their effect on personal health?

To do this a sophisticated agent-based model of transport will be used to evaluate how differing social norms impact the relative effectiveness of interventions aimed at increasing active travel will be developed. This model will be parameterised through statistical analysis of large-scale secondary data.

The model will be validated by implementing real-world interventions and comparing the results. Once a properly calibrated model has been produced, the model will be used to evaluate a number of hypothetical interventions. This allows them to be investigated without needing to be implemented, allowing for greater insight into what makes interventions successful.

This project aims to target the quantitative skills from the MRC's cross-cutting themes by the analysis of large-scale secondary data and the application of cutting-edge artificial intelligence methods (agent-based modelling). This also has a strong interdisciplinary element, with the application of computer science methods such as agent-based modelling and network science methods.

The MRC studentship has so far provided specific quantitative training in Medical Statistics (MSc-LSHTM).