PHASE: The Population Health Agent based Simulation nEtwork

Lead Research Organisation: University of Glasgow
Department Name: MRC Social & Public Health Services Unit

Abstract

Ideas that seemed like science fiction only a few years ago are now reality. Modern computers are powerful enough to simulate a city with millions of citizens inside a virtual world. We can learn how to improve health in the real world by studying "simulated populations" like this, and identifying how individual citizens, or "agents", within this virtual world interact with each other and how their health behaviours change when aspects of the simulated environment change.

The Population Health Agent-based Simulation nEtwork (PHASE) will bring together scientists, computer software developers, government workers, charity sector staff, and University researchers who are interested in improving the health of the population. By supporting different groups to work together, the UK will become a world leader in simulation methods for health, and will aim to tackle key public health issues like obesity, tobacco and alcohol use, and physical activity and exercise.

The PHASE network aims to:
1. Teach people how simulating individual and environmental factors and their influence on population health can improve our understanding of how to prevent non-communicable diseases.
2. Build stronger teams: putting together people that can create population simulations with scientific experts and those on the front line working in communities to improve public health.
3. Move science fiction to science fact: To get people around the UK talking about simulation methods for health, so that research funders, Universities, health workers and the public understand more about this 21st Century approach to improving health.

To achieve these aims, the network will hold meetings to get people talking about simulation methods and how these can be applied to public health problems, make new connections and build new teams. Training in simulation methods will be delivered through the network and expert network members will write reports and guidelines so a wide range of people can learn about the best way to use the simulation methods.

The network staff will develop an interactive website and online registry where people can find out:
Who is in the network and how to join?
What are the key health improvement issues?
What simulations are already developed and how can I use them?

In the long term, the PHASE network will increase the number of people across the UK who know about and understand simulation methods for health, and build a community of experts working in health improvement using these methods. Through network events and training we hope to build expert teams from a range of different backgrounds, working together to use simulations to tackle some of the major public health challenges facing us today.

Technical Summary

Major population health challenges including non-communicable diseases are not a product of simple, linear causal relationships, but of numerous interdependent and interacting processes, operating at multiple levels. In this context, population health research is increasingly adopting the complex systems perspective - a whole-system view centered on modelling the interactions between individuals and their social and physical environments. Agent-based modelling (ABM) is a complex systems simulation methodology that models individuals embedded in a physical and social environment, with their actions determined by simple decision rules. ABMs can model both individual health behaviours and their upstream determinants, enabling us to better understand the tangled web of processes that drive health and generate inequalities at the population level.
PHASE will facilitate the application of ABMs in population health through education, training and collaboration. Network activities will address issues of critical importance to enabling population health researchers and decision makers to apply ABMs: appropriate problem domains; best practices for model development; technical training; and ABM reporting guidelines. Working groups will develop documentation protocols, best-practice guidelines, and computational training. Our initial membership includes population health scientists, ABM specialists, health policy and practitioner groups and industrial collaborators, all of whom will be invited to contribute to our events and working groups. PHASE will also create an ABM study registry that will allow stakeholders to seek help with problems that may benefit from an ABM approach.
PHASE will particularly seek to engage with evidence users and researchers from multiple disciplines to generate new transdisciplinary teams that will apply ABMs to important challenges in the prevention of non-communicable disease.

This grant is funded by the UK Prevention Research Partnership (UKPRP) which is administered by the Medical Research Council on behalf of the UKPRP's 12 funding partners: British Heart Foundation; Cancer Research UK; Chief Scientist Office of the Scottish Government Health and Social Care Directorates; Engineering and Physical Sciences Research Council; Economic and Social Research Council; Health and Social Care Research and Development Division, Welsh Government; Health and Social Care Public Health Agency, Northern Ireland; Medical Research Council; Natural Environment Research Council; National Institute for Health Research; The Health Foundation; The Wellcome Trust.

Planned Impact

Contribution of PHASE to planned UKPRP outcomes and impacts: The motivation and rationale for PHASE are underpinned by our vision to contribute to the achievement of UKPRP Impacts 21 (Evidence-based solutions) and 22 (Decreased health inequalities and redistribution of resources). We will do this by increasing the use of ABMs in co-productive transdisciplinary multi-sectoral teams that engage and influence decision makers (Outcome 17 and 18). The key impacts we expect to achieve within the award period relate to UKPRP Outcomes 8 (Sustained community), 10 (New methodology for systems approaches), 12 (New interdisciplinary research agenda) and 14 (Transdisciplinary research group capacity).

Our vision, to normalise the application and use of ABMs among researchers and decision makers in the primary prevention of NCDs, will be delivered through a Theory of Change informed by Normalisation Process Theory. This asserts that the widespread implementation in routine practice (normalisation) of new programmes/methods (eg. Complex systems aproaches and ABMs) is dependent on capacity, capability and potential.

Who will benefit and how?
Practitioners, policy-makers and multi-sectoral public and third sector organisations: these organisations will be the principal group of beneficiaries and are critical to the achievement of UKPRP and PHASE aims. We will seek to increase the capacity, capability and potential of these communities to adopt and apply ABMs by:
1. increasing their awareness and understanding of ABMs and how they can be a useful tool in conceptualising and modelling the complex 'wicked issues' they face every day;
2. increasing their knowledge of examples of the successful application of ABMs to NCD prevention;
3. providing access to resources and support;
4. facilitating accessibility to others working across multiple sectors with an interest in applying ABMs;
5. providing pump-priming funding for user-led transdisciplinary teams to develop ABM projects;
6. improving the transparency, reliability and reporting of ABMs.

Commercial sector: we have engaged with a number of industry partners who are already developing ABMs or providing software and hardware platforms for ABMs or large-scale simulations. PHASE will provide companies from the sector with the opportunity to identify academic and user partners to work with as collaborators or customers, and to inform them of product developments that may be required to meet the user requirements of ABM developers.

Public: The key long-term impact of UKPRP, to which PHASE will contribute, is the population prevention of NCDs and reduction in health inequalities. Our communications plan will aim to achieve a better public understanding of the multiple, interdependent factors that underlie the rising incidence of NCDs and the need for complex systems informed interventions at multiple levels. PPI will be required in Research Development Groups awarded PHASE pump-priming funds.

Research funders and journal editors: PHASE will seek to increase research funding for ABMs to increase capacity in the field, and to promote publications of high quality ABMs in high impact health journals. This will be achieved through increasing funders' and journals' potential and capability to recognise and evaluate the quality and importance of ABMs by, for example, using our reporting guidance and our register as markers of high standards.

Researchers from other disciplines: we will seek to engage and learn from scientists from other disciplines who have greater experience in applying complex systems in general and ABMs in particular. An impact on them will be benefit to their research methods and practice through learning from population health scientists and the central importance of health and wellbeing in addressing applied research questions in other sectors.

Publications

10 25 50
 
Description Developing collaboration with Nvidia (Eric Silverman) 
Organisation NVIDIA
Country Global 
Sector Private 
PI Contribution Multiple meetings with Chris Emerson, Craig Rhodes, Jonny Hancox and Marjut Dieringer of Nvidia. Chris, Craig and Jonny work in Nvidia's health science team, and have developed multiple large collaborations with universities in areas like medical imaging. Marjut is part of their Deep Learning Institute which provides training on deep learning/AI. In our meetings we have discussed collaboration on the UKPRP-funded PHASE project, which focuses on agent-based modelling, and on applying deep learning methods to problems in population health. I have been introducing them to the PHASE project and developing specific areas for collaborative work on simulation approaches for population health.
Collaborator Contribution Nvidia intend to take part in our PHASE project and as a confirmation of the strategic importance of our relationship they have offered to provide deep learning training to the Unit. In upcoming meetings we will discuss their participation in PHASE events and they will provide feedback about the types of events that would be most useful to them.
Impact No outputs yet, however there will be opportunities to develop proof-of-concept projects via PHASE seed funds which can produce collaborative outputs.
Start Year 2019
 
Description Bayesian Agent-Based Population Studies workshop (Eric Silverman) 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Policymakers/politicians
Results and Impact A small group of 10 researchers from the UK, Germany and Poland gathered in Southampton for this workshop organised by the Centre for Population Change and run by Professor Jakub Bijak. Professor Bijak and his colleagues shared some outcomes of their project on Bayesian agent-based modelling for demography, and the group used these presentations as jumping-off points for discussing ways to increase the uptake of agent-based modelling across various disciplines, and to increase policy-makers' confidence in the method. Professor Bijak is producing a written summary of the event and the main discussion points raised, and we have agreed to develop an online forum for ABM practitioners to share advice and code, and to investigate the possibility of a journal focussed on methodological questions in ABM research.
Year(s) Of Engagement Activity 2020
 
Description Invited Seminar (Health and Social Care Analysis/ISD, Scottish Government) (Eric Silverman) 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Policymakers/politicians
Results and Impact 12 civil servants and analysts from ISD and the Health and Social Care Analysis team in Scottish Government invited me to give a seminar on agent-based modelling for the study of social care. The seminar took place at St Andrews House in Edinburgh and lasted 90 minutes. Participants were very enthusiastic about the possibilities presented by ABM approaches, and are keen to establish collaborative links with the Unit and with our UKPRP-funded PHASE project.
Year(s) Of Engagement Activity 2019