Identifying anomalies in canine health data

Analysing UK canine health data, we're developing new real-time methods to identify clinically relevant patterns and anomalies to support the early detection of any future outbreaks of disease in the UK's dog population.

The development of spatiotemporal statistical methods for anomaly detection enables the detection of regions or seasons in which symptoms or diseases of interest are unusually prevalent.

This research uses data from veterinary practices and diagnostic laboratories participating in the Small Animal Veterinary Surveillance Network (SAVSNET). This includes both the syndromes selected by the veterinary surgeon and also text mining methods. The research is closely linked with the work being done at the University of Bristol and the Animal Health Trust which focuses on disease outbreaks and setting disease outbreak thresholds. 

Charlotte Appleton at Lancaster University has been working closely with the University of Liverpool and the University of Manchester in exploring SAVSNET data and the syndromes recorded. Linking with the University of Bristol and Animal Health Trust, discussions have begun regarding thresholds and relaying information back to stakeholders.