 |
| |
Research |
|
|
|
|
Developing Simulation-based Parametrisation Methods for Models of Sexually Transmitted Infection: Making Policy Recommendations Evidence-based |
|
Listen to what the team aims and expects of their research project
 Alex Richard Cook |
|
Abstract
Sexually-transmitted infections (STIs) are a major cause of preventable ill health in many parts of Asia. Mathematical models of infectious diseases provide a way to assess the efficiency of intervention programmes for which it would be infeasible to perform trials for cost or ethical reasons. For these studies to be relevant to the realities of public health requires they be realistic and evidence-based. However, STIs differ from many other infectious diseases, such as respiratory illnesses, by the strong heterogeneity of human sexual behaviour, making them much harder to parametrise realistically; this has led to drastically conflicting conclusions on effectiveness of interventions even on the same study population.
Recently, novel simulation based statistical methods have been developed to fit complex genetic, ecological and engineering models to highly structured data, using particle filtering or approximate Bayesian computation. Both rely on being able to simulate “data” to be compared to the actual data via a model-derived metric; both also inherently deal with the search over the high dimensional parameter space. Heretofore, their use in infectious disease epidemiology has been limited to simple, non-STI models, but they provide the clearest route to statistically rigorous estimation of STI parameters. A particular benefit is that both are forms of Bayesian estimation, which facilitates incorporation of multiple streams of data, thereby allowing parameters that cannot be well resolved from the primary data set to be determined from auxiliary data, and concomitantly permitting the information content of the primary data to be exploited in other ways.
This project will provide a rigorous methodology to use complex models---in our case of STIs, though the method will be more general---to evaluate the effectiveness of prospectively useful but costly interventions such as partner notification programmes. Alongside development of the approach we will demonstrate its utility by applying it to local and regional case studies. |
|
|
Team Members |
 |
Alex Richard Cook
Assistant Professor
Department of Statistics & Applied Probability, Faculty of Science,
National University of Singapore
Singapore |
|
 |
Adrian Röllin
Assistant Professor
Department Statistics & Applied Probability, Faculty Of Science,
National University of Singapore
Singapore |
|
 |
Mark I-Cheng Chen
Associate Consultant
Communicable Disease Centre, Tan Tock Seng Hospital
Singapore |
|
 |
David Bruce Matchar
Professor
Program in Health Services & Systems Research,
Duke-NUS Graduate Medical School
Singapore |
|
