Keywords: Agent Based Modelling
Need: Agent based modelling is a powerful tool for simulating complex problems such as the spread of epidemics, consumer behaviour or congestion. However, there are not many successful examples of applied agent based simulation in the NHS, partly due to finding an appropriate use-case and partly due to the level of complication in the model development. HASH have created an online intuitive tool which allows for this technique to be demonstrated in a transparent manner. Using HASH or a similar tool, this project would seek to investigate the power of agent based simulations and their appropriate application for managing health services.
Current Knowledge/Examples & Possible Techniques/Approaches: Whilst other suitable tools may exist (Mesa (Python) or Repast (Java) frameworks), our discovery work has demonstrated that HASH would allow a model to be created in the timeframes for this project and that the learnings would be transferable and open. Example simulations can be found on the HASH website as well as a guide to using the HASH process modeling library and visual interface. However, if a candidate brings experience/knowledge of an alternative open framework for building these models, then the project could focus on demonstrating a use-case through their recommended tool.
Related Previous Internship Projects: SynPath
Enables Future Work: Demonstration of technique feeding into further SynPath developments as well as supporting pathway analysis and system modelling programmes.
Outcome/Learning Objectives: A&E capacity model built in HASH or alternate framework. Staff acting as agents deciding which patient to prioritise based on patient characteristics, current queue and A&E capacity (triage model). Learning would highlight when benefit can be realised from this approach and how to validate the model behaviour against reality.
Datasets: n/a for development, but SUS and ECDS data for validation.
Desired skill set: When applying please highlight any experience around agent based simulations, coding experience (including any coding in the open), and any other data science experience you feel relevant.