Modelling disease emergence in heterogeneous populations

Diseases that can make the leap from animal to human are an important focus for research. Some, such as SARS, have already caused outbreaks; others, like avian influenza, have the potential to do so. However, there is still much we do not understand about these infections. In particular, can we distinguish between pathogens that transmit only from animal to human, and those than can transmit from person to person? And how can we estimate the rate at which a new infection spreads?

Part of the answer lies in Southeast Asia. The region forms an evolutionary hub for many pathogens, and events there are likely to have a major impact on global health in the future. In particular, how people interact with each other, and animals, in these countries could affect disease emergence and subsequent transmission in a population. Therefore London School of Hygiene and Tropical Medicine recently led a study that collected data on social contact networks in Vietnam, Thailand, Indonesia and Taiwan. However, to understand the dynamics of emerging infections, this data needs to be incorporated into mathematical models, which can be used to simulate outbreaks. This will form the first part of the project; specifically, I will work on methods to identify infections that have adapted to spread between humans, based on case reports stratified by age. The potential for these approaches to be applied during the early stages of an outbreak will be investigated using data on the 2009 H1N1 pandemic in the UK.

The second part of the project will develop statistical methods to better understand disease transmission following the emergence of a new pathogen. These techniques could provide essential information in the early stages of an outbreak, and will be tested against real data from Vietnam and Indonesia.

The project will use ideas and techniques from several fields to achieve these aims, with collaborators including specialists in evolutionary biology, statistics and epidemiology. The insights generated by the project will help improve our understanding of pathogen emergence, create better ways to spot human-adapted strains, and develop tools to provide crucial information about new outbreaks.

Data on social interactions will be used to approximate contact structure, including interaction with animals. Disease transmission will be modelled stochastically; in particular, I will use multi-type branching processes to develop a framework in which heterogeneity in social contact - and hence transmission - can be incorporated along with pathogen adaptation. Analysis of the age distribution of disease cases during an outbreak will then be used to generate an inference method of distinguishing between different situations - for instance, stuttering transmission, whereby the disease fails to establish in the human population, or a full-scale epidemic. This will involve using approximate Bayesian computation (ABC), which bypass exact likelihood calculations by using summary statistics (such as the distribution of cases at the end of the outbreak). The potential for these techniques to be applied during an outbreak will be explored using data on the early progression of the 2009 H1N1 pandemic in the UK.

Using simulated epidemic data generated from these models developed, I will also explore a statistical framework for inference of the effective reproductive ratio, with contact data used as a prior. Existing 'gold-standard' data augmentation approaches are computationally intensive for large populations, so I will look at alternative such as ABC and ABC based on sequential Monte Carlo (ABC-SMC), as well as branching process approximations. The performance of these frameworks will be tested against real disease case data from Southeast Asia, as well as simulated data.

Disease surveillance and control is a major concern for public health agencies. The research will provide insights into how to effectively monitor for new infections, and tools with which to assess the likely impact of a new outbreak. The work may also help inform surveillance measures, such as what data are best to collect before, and during, an outbreak. These are critical pieces of information for public health officials, both at the national and international level.

The burden disease brings - both in Asia and abroad - can have a negative impact on economies, both in terms of healthcare costs and absenteeism. Understanding the emergence of new infections is an important step to reducing this burden. By investigating the dynamics of zoonoses, the project also has the opportunity improve the effectiveness of disease control for affected countries.

As pandemic emergence is a topic of national importance, the research will also be relevant to the public. The proposed research into new outbreaks - communicated effectively to this general audience - will improve individuals' knowledge and understanding of such events.