Minimum Models for Optimal Epidemic Monitoring and Control

点击翻译按钮获取中文摘要

Mathematical models have become indispensable to public health. Infectious-disease models convert observed epidemic data into predicted outbreak responses by simulating and providing insight into disease transmission. Realistic predictions allow policymakers to efficiently and economically evaluate the risks and benefits of different interventions before trial in the field. Models were crucial to controlling the devastating 2014 African Ebola outbreak, for example, by guiding the allocation of people and resources. However, attempts to describe disease transmission more precisely have led to increasingly complex models. Complex models are difficult to interpret and validate and may conceal epidemic assumptions. Policy based on them can be unreliable and overconfident, resulting in resource mismanagement and even mortality. This can be especially costly in the UK, where modelling directly informs vaccination programmes. Event-triggered control is an engineering subfield that recommends control actions as responses to informative events, called triggers. By directly linking monitoring to control, it prioritises interpretability and exposes innate assumptions. Recent research suggests that refocusing modelling efforts on outbreak management, instead of disease transmission, could remove unnecessary complexity. Event-triggered control may be precisely the tool for this job. Ebola and influenza are two priority diseases in the UK where model complexity and uncertainty are salient concerns. By working with officials at Public Health England and the World Health Organisation to identify important Ebola and influenza triggers, I will derive models of justifiable complexity and provable reliability.