Dynamic simulation modelling


Read a fact sheet about dynamic simulation modelling

The Prevention Centre is promoting the use of dynamic simulation modelling in health to provide policy makers with a unique ‘what if’ tool to test the likely impact of a range of possible solutions before implementing them in the real world.

Our team’s program of work is expanding a range of applications of dynamic simulation modelling methods and approaches. We are evaluating their feasibility and usefulness in supporting policy and practice for the prevention of chronic disease

The team, working in close collaboration with relevant policy partners, has completed 6 projects to date. We are currently building an integrated model of multiple risk factors and chronic harms.

What is dynamic simulation modelling?

Dynamic simulation models are virtual representations of the real world where  individuals and communities act and react, develop health conditions, and use health services in the same way as people do in real life.

Informed by a variety of evidence sources such as research, expert knowledge, practice experience and population-based behavioural and demographic datasets, the models can represent complex human behaviours. They can follow individuals throughout life, identifying how their influences and behaviours change, and how their health is impacted over time.

They can also map the aggregate effects of thousands of people operating in a complex system, following their interactions and responses to different policies and interventions.

A dynamic simulation model is a ‘what if’ tool through which we can test various policy scenarios over time to see which are likely to have the most effect, both on individuals as well as on the system as a whole.

Background

Watch this video for a layperson’s explanation of dynamic simulation modelling

The solutions to complex public health problems such as harmful alcohol consumption, overweight and obesity and mental disorders remain elusive, despite hundreds of millions of dollars and decades of research that have led to attempts to combat them through ‘evidence-informed’ policy.

This is partly because there are many ways to tackle public health problems, multiple risk factors, and different intervention points during the lifespan. Complex chronic health problems have many inter-related causes and it can be unclear how these factors interact. In addition, there is often a broad range of possible interventions to prevent chronic disease, applied individually or in combination, which must be considered in light of their comparative costs, benefits and health system implications over the short and longer term. Health decision makers often face the challenges of differing stakeholder and community views, limitations in research evidence, political considerations and industry lobbying.

Traditional analytic tools for supporting evidence-informed policy are limited in their ability to help policy makers to plan and make decisions.

Computer simulation models provide us with a framework for mapping and quantifying complex problems. They provide a relatively low cost and risk-free way of testing the likely impacts and costs of different options for intervening before implementing them in the real world.

Dynamic simulation modelling can answer important questions, such as which risk factors are the most important, when in people’s lives we should target interventions, and which combinations of interventions work best, are most equitable and most cost-effective.

Why use dynamic simulation modelling?

  • It offers a way of helping to better focus resource allocation for optimal impact
  • It helps policy makers to test big decisions in the virtual world before implementing them in real life
  • It enables us to follow the dynamics of populations, their behaviours and outcomes over time
  • It is low risk and relatively low cost: We can test what works without the expense and ramifications of trialling new or multiple interventions in the real world
  • There is potential to include many possible effective interventions and select different combinations
  • We can test the impact of alternative interventions in situations where research consensus has not been reached
  • In situations where the likely impact of different combinations of interventions is difficult to predict, the model can provide insights into potential synergistic, additive or lagged effects.

For more information about dynamic simulation modelling, please contact Dr Jo-An Atkinson Tel: (02) 9188 9537

 

Read a fact sheet about dynamic simulation modelling

The Prevention Centre is promoting the use of dynamic simulation modelling in health to provide policy makers with a unique ‘what if’ tool to test the likely impact of a range of possible solutions before implementing them in the real world.

Our team’s program of work is expanding a range of applications of dynamic simulation modelling methods and approaches. We are evaluating their feasibility and usefulness in supporting policy and practice for the prevention of chronic disease

The team, working in close collaboration with relevant policy partners, has completed 6 projects to date. We are currently building an integrated model of multiple risk factors and chronic harms.

What is dynamic simulation modelling?

Dynamic simulation models are virtual representations of the real world where  individuals and communities act and react, develop health conditions, and use health services in the same way as people do in real life.

Informed by a variety of evidence sources such as research, expert knowledge, practice experience and population-based behavioural and demographic datasets, the models can represent complex human behaviours. They can follow individuals throughout life, identifying how their influences and behaviours change, and how their health is impacted over time.

They can also map the aggregate effects of thousands of people operating in a complex system, following their interactions and responses to different policies and interventions.

A dynamic simulation model is a ‘what if’ tool through which we can test various policy scenarios over time to see which are likely to have the most effect, both on individuals as well as on the system as a whole.

Background

Watch this video for a layperson’s explanation of dynamic simulation modelling

The solutions to complex public health problems such as harmful alcohol consumption, overweight and obesity and mental disorders remain elusive, despite hundreds of millions of dollars and decades of research that have led to attempts to combat them through ‘evidence-informed’ policy.

This is partly because there are many ways to tackle public health problems, multiple risk factors, and different intervention points during the lifespan. Complex chronic health problems have many inter-related causes and it can be unclear how these factors interact. In addition, there is often a broad range of possible interventions to prevent chronic disease, applied individually or in combination, which must be considered in light of their comparative costs, benefits and health system implications over the short and longer term. Health decision makers often face the challenges of differing stakeholder and community views, limitations in research evidence, political considerations and industry lobbying.

Traditional analytic tools for supporting evidence-informed policy are limited in their ability to help policy makers to plan and make decisions.

Computer simulation models provide us with a framework for mapping and quantifying complex problems. They provide a relatively low cost and risk-free way of testing the likely impacts and costs of different options for intervening before implementing them in the real world.

Dynamic simulation modelling can answer important questions, such as which risk factors are the most important, when in people’s lives we should target interventions, and which combinations of interventions work best, are most equitable and most cost-effective.

Why use dynamic simulation modelling?

  • It offers a way of helping to better focus resource allocation for optimal impact
  • It helps policy makers to test big decisions in the virtual world before implementing them in real life
  • It enables us to follow the dynamics of populations, their behaviours and outcomes over time
  • It is low risk and relatively low cost: We can test what works without the expense and ramifications of trialling new or multiple interventions in the real world
  • There is potential to include many possible effective interventions and select different combinations
  • We can test the impact of alternative interventions in situations where research consensus has not been reached
  • In situations where the likely impact of different combinations of interventions is difficult to predict, the model can provide insights into potential synergistic, additive or lagged effects.

For more information about dynamic simulation modelling, please contact Dr Jo-An Atkinson Tel: (02) 9188 9537

 

Although dynamic simulation modelling has been used successfully in engineering, ecology, defence and business for many years, its value for health policy and planning is being increasingly recognised.

The Prevention Centre uses a unique participatory, transparent approach to dynamic simulation modelling.

We work collaboratively with our project partners, policy agencies, researchers and other stakeholders to build and obtain the data and evidence inputs that comprise our models. This gives our policy partners confidence in the models’ integrity and validity. It also provides partners with a level of ownership, ensuring they are more likely to interact with the tool and use it to build policy evidence.

Our models are:

  • Technically advanced: Due to the pioneering work of our collaborators Dr Geoff McDonnell and Professor Nate Osgood, we have the ability to integrate multiple modelling methods in many of our projects
  • Accessible: We have a user-friendly interface
  • Participatory: We embed stakeholder engagement, consultation and consensus-building processes from the very beginning of the project
  • Diverse: Our product allows us to synthesise more diverse evidence sources compared to systematic reviews or meta-analyses
  • Robust and reliable: We test our models retrospectively against reported outcome indicators

Our models

The Prevention Centre two main modelling approaches: System Dynamics and Agent-based modelling.

  • System dynamics is a system modelling engineering tool that has been in use for over 60 years. It is a quantitative tool that was initially used in industrial management, and which is now commonly used in manufacturing, process-oriented industries, and defence logistics planning and asset lifespan. Today, system dynamics is widely used in climate science as a tool that incorporates ‘what if’ assumptions to analyse the impact of changes on our climate system, from sea level rises to historic temperature change and global CO2. System dynamics takes an aggregate, population level perspective, considering flows, feedback loops and time-delays in a complex system.
  • Agent-based, often called individual-based, models simulate the actions and interactions of individuals (‘agents’) in order to generate insights into the impact of their health and social behaviours on themselves and the wider population. These models use data or evidence to tailor the demographic characteristics, exposures, and behavioural rules that drive individual behaviours and health outcomes. Agent-based models can be developed with hundreds of thousands individual agents and are valuable for exploring equity issues such as the differing effects of exposures and interventions on the health outcomes of sub-populations

What questions can our models answer?

Our dynamic simulation models can test interventions before they are implemented in the real world. They provide reliable forecasts and insights into:

  • Which risk factors are the most important in the prevention of chronic disease
  • How different risk factors interact in complex ways to cause a number of outcomes
  • What different outcomes can be achieved by intervening on (or targeting) one risk factor, or a combination of risk factors
  • Which health interventions are most likely to work
  • Which interventions are the most equitable
  • Which interventions are the most cost-effective
  • How different interventions will impact the health system as a whole
  • How and when in people’s lives different interventions should be targeted.

Our process

  1. We work collaboratively with academics, policy makers and practitioners to map the key risk factors and likely causal pathways leading to the health or economic outcomes we are interested in. This information is combined with empirical evidence and administrative, survey and burden of disease data to create a computer model that represents the real world.
  2. We test the model to ensure it accurately represents the population of interest. We can model anything from small sections of the community in different geographic locations to the population of the entire country.
  3. We produce a baseline to see what would happen with no new interventions (business as usual).
  4. Using a simple interface, we select different interventions, individually or in combination, and target them in different ways to forecast how this is likely to impact the outcomes over the short and long term.

Once our models are completed, we have the option of sharing them amongst our key policy agency partners via easy-to-use web browser interfaces, or exporting results to statistical programs for further analysis.

Software tools

We have experience in using three software tools to develop our models:

  • Insightmaker (IM) – insightmaker.com is a free-ware simulation tool that runs in most web-browsers. It is used to build conceptual representations of complex systems which can then be converted to computational models through quantifying variables and links.
  • Stella Architect – iseesystems.com is a professional system dynamics modelling software package with user-friendly interface design capability.
  • AnyLogic – anylogic.com is a professional multimethod simulation modelling platform used for our agent-based and hybrid modelling projects. It allows for large population datasets, large-scale models, valuable visualisation and analytical tools and GIS capability.

 

The Prevention Centre has completed a range of projects as both exploratory research studies and as commissioned projects for various state health agencies, including ACT Health, NSW Health, Tasmania Department of Health and Human Services and Queensland Department of Health. Health problems to which we have applied simulation modelling methods include alcohol-related harms, tobacco use and COPD, reducing childhood overweight and obesity and improving diabetes.

Further examples of applications in mental health, suicide prevention and healthcare can be found on the Decision Analytics webpage.

Simulation modelling of alcohol consumption and the effectiveness of harm-reduction policies

The Prevention Centre developed a simulation model of alcohol use in NSW that could be used to forecast the effectiveness of a variety of approaches to reducing alcohol-related harm, both individually and in combination. The model addressed both binge drinking and high average consumption that leads to chronic disease.

Following from this project, we have produced a simulation model of alcohol use in Tasmania, with input from stakeholders including alcohol and drugs services, youth services, the Liquor Commission, Treasury, Police and the Education Department.

Read more.

System dynamics modelling to inform strategic planning for achieving the NSW Premier’s target for reducing childhood overweight and obesity

A dynamic simulation model developed in partnership with NSW Health is being used to inform strategic planning for the NSW Premier’s Priority on childhood overweight and obesity. The model was built to forecast which combinations of policies and programs would be needed to achieve the Premier’s target of reducing childhood overweight and obesity by 5% within 10 years.

Read more.

Developing a compelling case for prevention

This exploratory project is drawing together evidence about the health and economic burden of chronic disease, its preventability and the economic credentials for action. It is using this information to build a national dynamic simulation model using Australian population data that will allow decision makers to explore the likely population health and cost impacts of different interventions to reduce risk factors for chronic disease. A component of this project is building three agent-based sub-models for the ACT targeting reductions in high body mass, smoking and alcohol use. It is planned at some stage these sub-models will be linked to the national system dynamics model architecture. This model is still under development.

Read more.

Simulation modelling to support decision making in gestational diabetes care

This project is applying dynamic simulation modelling to the problem of diabetes in pregnancy in the ACT as a case study. This is a world leading tripartite model that includes system dynamics, agent-based modelling and discrete event simulation components. The model considers the short-, intermediate- and intergenerational implications of the increasing prevalence of risk factors for diabetes in pregnancy, and will inform investments in healthcare and population level interventions.

Read more.

Modelling strategies to reduce smoking in Queensland

The Queensland Government commissioned a dynamic simulation model to inform its smoking reduction strategy by identifying and testing key priority interventions to see which are most likely to make an impact.

Read more.

Modelling smoking behaviour and its influence on COPD in NSW

The research simulation study utilised the data from adults in NSW and Australia including the data from the Sax Institute’s 45 and Up Study.

The model captures detailed smoking behaviours, including smoking intensity, cessation rate and length, and relapse. The outcomes elucidate influence of smoking cessation attempt on incidence and prevalence of COPD by stage of the disease.

Read more.

 

 

 

For all inquiries about dynamic simulation modelling at the Prevention Centre, please contact Dr Jo-An Atkinson Tel: (02) 9188 9537

Lead

Dr Jo-An Atkinson, lead of evidence synthesis and simulation, Prevention Centre

Team

Jacqueline Davison, Research Officer, Prevention Centre

Louise Freebairn, PhD candidate, ACT Health

Pippy Walker, Research Officer, Prevention Centre

Jaithri Ananthapavan, Senior Research Fellow, Deakin University

Paul Crosland, Senior Research Fellow, Deakin University

 

Modellers

Dr Geoff McDonnell

Dr Ante Prodan

Mark Heffernan

Professor Nate Osgood

Adam Skinner

Alex Dumais

Dylan Knowles

 

2018

2017

2016

2015

In the media

The Mandarin, 1 November 2016: Introducing NSW liquor controls state-wide could reduce acute alcohol harms by 20%

 

Prevention Centre News

2 March 2018: Modelling drives innovation to prove the case for prevention

29 January 2018: Modelling reveals how to reduce alcohol-related harm in Tasmania

13 November 2017: Dynamic simulation modelling for a smoke-free Queensland

4 July 2017: Informing Tasmania’s Alcohol Action Framework

26 October 2017: Policy partner wins award for simulation modelling project

3 October 2017: Building a local workforce of dynamic modellers

8 June 2017: Making the compelling case for prevention

20 March 2017: Informing the NSW Premier’s Priority on childhood overweight and obesity

23 June 2017: Dynamic simulation modelling informs alcohol strategy in Tasmania

15 May 2017: Project expanded to tackle all forms of diabetes in pregnancy

1 November 2016: Dynamic simulation modelling sheds light on effective alcohol harm reduction approaches

12 May 2017: PhD project builds network to tackle diabetes in pregnancy

1 June 2016: Simulation modelling helps to unpick causes of gestational diabetes

9 May 2016: Keeping it real: the potential of big data in simulation modelling

9 February 2017: Keep a ‘human in the loop’ when using big data for health policy: Nate Osgood

10 August 2016: Experts meet to map complex factors in childhood obesity

10 August 2015: Hands-on modelling workshop tackles complexity of alcohol misuse

 

Prevention Centre blog

23 March 2015: ‘What if’ tool explores complex problems

 

Videos

A what if tool to better understand complex health problems: Lay person’s video explains what dynamic simulation modelling is

Integrating big data and simulation models to support health decision making: International data and system science expert Professor Nate Osgood describes how big data and dynamic simulation modelling is providing policy makers with unprecedented opportunities to address complex problems in public health.

Five minutes with … Professor Nate Osgood: Canadian systems modeller Nate Osgood explains the role that system science modelling can play in unpicking complex problems in public health