Paul Birrell – firstname.lastname@example.org
Potential PhD Project
Inference in agent-based models of infectious disease transmission
Classical transmission modelling proceeds by classifying populations into disease states (such as the susceptible-exposed-infectious-recovered class of models) with further stratification e.g. by age, location and/or risk group, used to incorporate heterogeneity. From such models it is a relatively straightforward task to carry out Bayesian estimation of model parameters and make forecasts for the future course of infection that account for all inherent sources of uncertainty (See  with examples in  and ).
Individual Based Models (IBMs) can be thought of the limit of this population stratification into compartments to the smallest possible unit of infection, the individual. This facilitates the incorporation of as much detailed information about the composition of the afflicted population as is available and may include (though is far from limited to) age, sex, ethnicity, type of work, home location, household composition. Such models are particularly useful when it comes to assessing the impact of a pandemic intervention, either prospectively or retrospectively. Such a task is, however, dependent on the model being well-calibrated against relevant data and, in comparison to compartmental models, this is not an easy task. These models are prohibitively computationally expensive in the sense that it is not possible, in a timely fashion, to run them the number of times that typical algorithms for Bayesian computation would require.
Bayesian emulation is a method for developing a computationally simple proxy statistical model that is capable of approximating certain key features of the more complex system. From a small number of training runs from the IBM, the statistical proxy model can be trained, and will then provide predictive distributions for model outputs given a set of parameters for the more complex system [e.g. 4]. This statistical model, or emulator, then replaces the complex model in the calibration step, enabling the estimation of epidemic parameters. In partnership with the UKHSA, this project will look at the appropriate calibration of models developed to provide guidance as part of the SARS-CoV-2 pandemic response, including the JUNE model  and models for tracking nosocomial infections in hospitals [e.g. 6], applying and developing the state of the art for emulation, encoding this in software and enhancing the modelling toolkit for pandemic decision support.
This work would be jointly supervised by Prof Daniela De Angelis
 Birrell PJ, De Angelis D, Presanis AM. Evidence synthesis for stochastic epidemic models. Statistical science: a review journal of the Institute of Mathematical Statistics. 2018;33(1):34.
 Birrell P, Blake J, Van Leeuwen E, Gent N, De Angelis D. Real-time nowcasting and forecasting of COVID-19 dynamics in England: the first wave. Philosophical Transactions of the Royal Society B. 2021 Jul 19;376(1829):20200279.
 Shubin M, Lebedev A, Lyytikäinen O, Auranen K. Revealing the true incidence of pandemic A(H1N1)pdm09 influenza in Finland during the first two seasons — An analysis based on a dynamic transmission model. PLoS computational biology. 2016 Mar 24;12(3):e1004803.
 Farah M, Birrell P, Conti S, Angelis DD. Bayesian emulation and calibration of a dynamic epidemic model for A/H1N1 influenza. Journal of the American Statistical Association. 2014 Oct 2;109(508):1398-411.
 Vernon I, Owen J, Aylett-Bullock J, Cuesta-Lazaro C, Frawley J, Quera-Bofarull A, Sedgewick A, Shi D, Truong H, Turner M, Walker J. Bayesian Emulation and History Matching of JUNE. medRxiv. 2022 Jan 1
 Stephanie Evans, James Stimson, Diane Pople, Alex Bhattacharya, Russell Hope, Peter J White, Julie V Robotham, Quantifying the contribution of pathways of nosocomial acquisition of COVID-19 in English hospitals, International Journal of Epidemiology, Volume 51, Issue 2, April 2022, Pages 393–403.
How to apply
For details of the MRC BSU application process please see How to apply
To be considered for funding applications need to be submitted to the University of Cambridge application system by 23:59 (GMT) on January 5th 2023