Real-time statistical anomaly detection for spatiotemporal streaming data

The Small Animal Veterinary Surveillance Network (SAVSNET) collects large volumes of anonymised health records from a sentinel network of UK veterinary practitioners in real-time (savsnet.co.uk). Recently, we have used these data to describe antibiotic prescription in veterinary practice, to look at the seasonality of important parasites like ticks and flystrike. We also publish surveillance reports regularly in the Veterinary record. Data is fed back to participants in the form of anonymised benchmarks.

The aim of this project is to speed up the flow of meaningful research surveillance measures so that they can be actioned in a clinically relevant time frame to improve individual and population canine health. These interlinked PhDs are i) in rapid statistical analysis of “big” streaming data for anomaly detection (University of Lancaster), ii) in providing rapid decision support to clinicians to help better manage chronic disease (Liverpool) and iii) developing a novel national framework to respond effectively to significant events like outbreaks in dogs (Bristol / Animal Health Trust). These PhDs will work closely together and with the rest of the SAVSNET team including software development and machine learning capability.

This PhD project will focus on developing statistical models and algorithms for spatiotemporal streaming data. It aims to model stochastic variation in disease cases as reported in real-time by SAVSNET on a national scale, detecting spatial “hotspots” of disease and unusual “spikes” of disease incidence to guide clinical practice in terms of diagnosis and treatment. The methodology developed in this project will be embedded into the SAVSNET system and automated to provide daily situation reports of the current small animal disease landscape in England.

We are looking for motivated graduates with either a high undergraduate or masters degree in a data science related subject. During the course of this PhD, the successful applicant will undergo training in:
• Spatiotemporal statistical methods, such as Gaussian Process and infectious disease models;
• Statistical scientific computing in R and/or Python;
• Skills in translational science, in particular those of working in a multidisciplinary team.

This exciting PhD will complement the work of the other PhDs, the wider SAVSNET team and the Dogs Trust, to transform the way big data is used in veterinary practice, surveillance and research when confronted with important canine disease.
During the PhD there will be regular meetings to link the work of the three PhDs into the wider SAVSNET team, and to the Dogs Trust,
As well as traditional research publications, the results of this degree will feed directly back to practitioners through existing benchmarking sites. This will ensure the results of this work are rapidly translated into best veterinary practice.
This project will be based at Lancaster University, working in CHICAS (Centre for Health Informatics, Computing, and Statistics) within Lancaster Medical School.

For further information about the project contact Chris Jewell [Email Address Removed]

Applications are made by completing an application for PhD Statistics and Epidemiology October 2019 through our online application system. Closing date: midnight 20th July 2019