Splash is a research project aimed at building a framework that supports the integration of multiple existing models, simulations, and data that represent parts of the broader health ecosystem. Specifically, the goal is to create a platform that takes expert models of constituent real-world systems related to health, synthesising and integrating those models, resulting in an interoperating complex composite system model with which policy-makers can try out alternatives in a low-cost, highly responsive way. The key research question is whether such integration of independently created, deep domain models can be made feasible, practical, flexible, cost-effective, attractive, and usable

Type of content: Assets
Type of asset:
Framework
Big data potential
No
Policy domains: Health
Phase in the policy cycle:
Policy Design and Analysis
Open license availability
No
Ease of use
Low
Tags: E-Governance Strategic planning
Addresses:
SWOT Analysis for
SPLASH
Helpful Harmful
Internal
Strengths• Building a framework that supports the integration of multiple existing models, simulations, and data that represent parts of the broader health ecosystem.
• Create a platform that takes expert models of constituent real-world systems related to health, synthesising and integrating those models, resulting in an interoperating complex composite system model with which policy-makers can try out alternatives in a low-cost, highly responsive way.
• Focus primarily on integrating and composing, i.e., “mashing up,” simulation and statistical models aimed at informing health policy decision making,
Weaknesses• Policies and interventions that will have reliable and effective impact require changes to a large set of interconnected systems
• The health ecosystem comprises a large set of complex, intricately connected subsystems.
• Different categories of models are constructed, maintained, and used by different people and organisations, each using different terms, conventions, and approaches. If models are not too diverse or too complex, it might be feasible to manually and tightly integrate them into a common framework, But if models are very heterogeneous and complex, it might not be feasible to combine them at all.
• Today there are no reliable and manageable means to explore the health system consequences of many interacting subsystems. Although individual deep-domain models can be made more comprehensive, there will always be important factors that lie outside the expertise of a given group of modelers, and it can be difficult to nimbly adapt very large and complex models as new questions arise.
External
Opportunities• Requirement for combination of multiple deterministic and stochastic simulation models, as well as statistical models and data sources, to project the effects of policy or investment choices into the future
• Health-Focused Policy Decision Making: Recent years have seen great innovations in both technologies and organisational mechanisms for promoting human health, but effective policy and investment decisions are needed to reap the benefits of these advances.
• Creating a framework for integrating disparate individual models to create effective and useful composite models.
• Need to enable a community of disparate stakeholders – those with health-related data, deep domain models, and health policy issues or questions – to work together, using the platform to make progress and solve problems. The community must be set up as an effective service system, creating more value through interaction than through isolation
• There exists no standard way to describe models in sufficient depth to determine compatibility: Here, the challenge is to create mechanisms and methods for describing models so that it is easy to determine how to integrate them into larger, more complex models of larger, more complex systems. The goal is for the Splash platform to semi-automatically identify models that are potentially compatible with a specified model, perhaps after some transformation of model inputs and outputs.
• There is no targeted technology and set of practices to facilitate collaboration between the varied people and organisations that develop and use deep-domain models: We envision an active community of participants contributing models and data, combining models, discussing models, exploiting previous results, and optionally sharing their models and modelling results. The final challenge is to develop a deep understanding of what is required for such an open integrated community system to successfully enable cooperation among all
Threats• Health-related investment and policy decisions by government agencies, healthcare providers, insurers, and other stakeholders may lead to complex interactions and have far-flung consequences, many of which may be difficult to foresee.
• Not all models can be combined in a sensible way. The assumptions, time scales, capabilities, level of detail, and indeed the selection of the key aspects to represent may be quite different: What factors characterise the models that are compatible with one another?
• There exists no standard way to describe models in sufficient depth to determine compatibility: Here, the challenge is to create mechanisms and methods for describing models so that it is easy to determine how to integrate them into larger, more complex models of larger, more complex systems.
• There are no tools or platforms to support mashing up independently created models and datasets in a simple, flexible, and useful way. This adds the challenge of providing efficient mechanisms for searching and identifying applicable models, for establishing an appropriate execution environment, for semi-automatically generating connectors between models and between models and datasets, and for enabling reuse, result pruning, data transformations, flexible model transformations, experiment management, visualisation, simulation output analysis, and so on
• There is no targeted technology and set of practices to facilitate collaboration between the varied people and organisations that develop and use deep-domain models

Open data - Download the Knowledge base

You are free to download the data of this Knowledge base.

To do this you must be an authenticated user: log in or sign in now.

All the data are licensed as Creative Common CC-BY 4.0.