Regional outcomes of national health policies are difficult to forecast. This is partly due to a lack of realistically complex models that can be used to appraise policy options and partly a lack of accessible and adaptable tools that can be used to simulate the consequences of policy decisions. These barriers might be overcome by exploiting the commoditization of massively parallel computing architectures, advances in machine learning, and the increased availability of large-scale linked healthcare data. This paper presents a novel modelling methodology, The Stock of Health, for harnessing emerging data and computational resources to simulate health policy, with application initially to coronary heart disease. We detail the use of multi-core graphical processing architectures to facilitate a micro-simulation approach. The simulation tools have been deployed through the IMPACT Framework. We explore how this framework can be extended to support the sharing and reuse of policy models and simulations based on the digital publishing concept of e-Lab. © 2013 IMIA and IOS Press.
- in-silico parallel simulation
- policy decision support
- Policy modelling
- public health