WMLH-1: Standards for decision analytic modelling to support healthcare decision making
Project description. (Jointly
supervised by Jim Chilcott,
The ScHARR Technology Assessment Group (ScHARR-TAG) supports the National Institute for Health and Clinical Excellence (NICE) in making its guidance to the NHS. Part of this work programme involves constructing decision analytic models of diseases and treatments to assess the clinical and economic impact of new and existing therapies. These models directly impact on treatment decision within the NHS.
There has been an increasing recognition of the potential for errors in models to impact adversely on NICE treatment decisions. In many domains such as defence standards for requirements capturing and system specification have been developed. The existing standards on undertaking economic assessments for NICE, do not address the process of model building either in problem formulation, model specification or model implementation.
The student will work alongside a ScHARR-TAG project team in undertaking an assessment of a new drug or treatment to support NICE decision making and explore the potential for defining methodological standards in decision analytic model building.
NICE is the UK National Institute for Health and Clinical Excellence, it is the body responsible for deciding what drugs and treatments will be approved for use in the National Health Service.
These decisions are based on evaluation reports produced by researchers in units such as ScHARR. These assessments are based on extensive research into the medical literature together with economic models that try to identify the costs and benefits of different medical treatments. Peoples live depend on them getting these assessments right.
There is concern, however, that some of the models may not be correct or accurate and this project will look at a current assessment project being undertaken in ScHARR in order to identify the main issues in verifying the correctness of the models being used.
Potential outcomes might be recommendations about the way the models are specified – the sort of language needed, the sort of validation and testing that is required to have confidence in the models and possible opportunities for automated analysis of the models – either dynamically using machine learning techniques (eg. DAIKON) or through static techniques such as probabilistic model checking (eg. PRISM), symbolic execution etc.
Project description. This will involve the use of the X-agent computational framework FLAME
in the modelling of a social network – a number of individuals together located in various positions with transport links between them. We will investigate how a disease, such as Avian Flu, Foot and Mouth etc. can spread through the interactions of individuals in locations where there is movement along specific routes. Currently all the modelling of the spread of disease that is done by governments etc. is based on the use of systems of differential equations that track changes at the population level. However, to catch a disease the individual will normally have to come into close contact with a carrier. We will model this process and investigate a variety of scenarios including the use of vaccination etc.
Books of epidemiology from the library.
The X-agent framework – this involves the use of the C programming language and, possibly, the use of parallel computers.
Project description. This will involve the use of the X-agent computational framework FLAME http://www.flame.ac.uk in the modelling of a social network – a number of individuals and organisations together located in various positions with transport links between them. We will use an agent-based approach to investigate how a social and economic system adapts under changing circumstances. We will use a case study which contains data about the effect of a factory closure in Leeds has on the local community – individuals and families, suppliers, local shops etc. We will design agents to represent all of these and use available data to create realistic models. These will include spending patters, receipt of benefits etc. movement away form the area etc. Such simulations will be valuable for policy makers in Governments and industrial companies.
The FLAME framework – this involves the use of the C programming language and, possibly, the use of parallel computers
WMLH-4: Analysis of swarm behaviour and tools for FLAME
Project description. This will involve the use of the X-agent computational framework FLAME http://www.flame.ac.uk in the modelling of swarm systems. We want to try to identify the emergent behaviour that is associated with different types of agents and rules. This will involve an experimental approach and the use of some tools such as DAIKON and the development of our own data mining tools to extract useful information from the data files exported from large numbers of simulations of swarms generated by FLAME.
The FLAME framework – this involves the use of the C programming language.