Inference for Aggregated Point Processes with Application to Disease Mapping

Point processes are a natural class of models for describing the spatial locations of a disease as they arise over time. Point processes can be used to describe how the locations of cases of a disease might evolve in the presence of environmental or socioeconomic risk factors and are applicable whether incident cases are self-limiting or have the potential to act as infective agents. They can be used to model multiple types of point simultaneously, thus allowing the modelling of both host and vector populations, for instance.

Methods for point process inference are currently well developed in the case that the true location of points is known. Unfortunately in real data applications, this is rarely the case, since collection of this information is both expensive and can also pose ethical challenges. To complicate matters, an individual’s residential location is not necessarily the optimal proxy for environmental exposure.

It is much more common in epidemiology to need to handle point data that has been aggregated to administrative, or politically defined units, such as health authorities, or hospital catchment areas. In developing country settings, data collection is also not reliable, with some health facilities reporting on a regular basis and others reporting intermittently. Furthermore, in deprived and particularly in deprived and physically isolated areas, there is likely to be substantial under-reporting of cases due to inability of individuals to report to facilities and inability of facilities to collect and transmit relevant data to central information offices. To compound matters, the risk factors we seek to understand are often measured at disparate spatial scales.

This PHD will develop statistical methodology to handle complex data of this sort.

Specifically, it will primarily be concerned with developing inferential methods (e.g. Markov chain Monte Carlo algorithms) for modelling aggregated counts of infectious disease in the presence of known environmental risk factors using the framework of point-processes.

Where does the project lie on the Translational Pathway?

T1 – Basic Research & T3 – Evidence into Practice

Expected Outputs

- new methodologies for handling aggregated point process data

- a reusable open-source software implementation of the algorithms

- an analysis of a real infectious disease dataset

- further work could include deriving optimal intervention strategies, given a set of resources and other


Training Opportunities

As well as standard Lancaster University researcher training

( the student would also be

encouraged, funding permitting, to attend other relevant courses e.g. APTS


Skills Required

- they should be highly numerate with a first degree in statistics or something with substantial mathematical and

computational content

- they should be interested in learning about and developing MCMC and other complex algorithms

- they should have strong programming skills

- they should be interested in learning more about point processes

- they should be motivated by health applications

Key Publications associated with this project

Continuous Inference for Aggregated Point Process Data. Benjamin M. Taylor, Hugh

Sturrock, Ricardo Pacheco. Journal of the Royal Statistical Society: Series A (2018). Vol. 181 Iss. 4. [ArXiV]

Bayesian Inference and Data Augmentation Schemes for Spatial, Spatiotemporal and

Multivariate Log-Gaussian Cox Processes in R. Benjamin M. Taylor, Tilman M. Davies,

Barry S. Rowlingson, Peter J Diggle. Journal of Statistical Software (2015). Vol. 63 Iss. 7.

Spatial and Spatio-Temporal Log-Gaussian Cox processes: Extending the Geostatistical

Paradigm. Peter J Diggle, Paula Moraga, Barry Rowlingson and Benjamin M. Taylor.

Statistical Science (2013). Vol. 28(4): 542-563.



LSTM Themes and Topics – Key Words

Malaria and other Vector-Borne Diseases

The call for applications for the 2020-21 round of studentships is now OPEN. Deadline for receipt of complete application 23:59 13th February 2020
Further information on the programme and application process can be found here