Defra - Department for Environment, Food and Rural Affairs.

Science Search

Science and Research Projects

Return to Science Search homepage   Return to Project List

Investigation of potential factors which improve the effectiveness of early outbreak detection methods for scanning data. - SE4301

Description
It is very important to be able to detect the emergence of any new farm animal diseases as rapidly as possible. A highly useful source of information on the occurrence of farm animal diseases is from scanning surveillance submissions at VLA, the results of which are held on the Farmfile database. In particular the data on
"diagnosis not reached" (DNR), where undiagnosed cases are reported according to their clinical syndrome, provides a set of data which could be potentially very useful at detecting any outbreaks of new or emerging diseases.

Descriptive statistics and temporal methods are routinely carried out for detecting unexpected high values of disease submissions. The use of statistical methods in surveillance is a dynamic research field that is evolving fast, and several methods have been developed which can be used for the early detection of outbreaks. However, since there have been no previous outbreaks detected in FarmFile, it is not known how effective these methods will be when applied to DNR data from Farmfile, and there are several factors which could influence the sensitivity of statistical methods when applied to FarmFile data. Firstly, there is a background occurrence of DNR’s for each syndrome. Secondly, there could be variability in how a new or emerging disease would be classified in terms of syndrome, making diseases with a variable clinical presentation harder to detect. Thirdly, there are factors that will influence whether a submission is made to VLA at all. Fourthly, there is also an issue of limited testing, where a small number of animals and/or a very limited range of diagnostic tests is carried out on farm due to resource implications. This could result in a large number of DNR’s where diagnosis could have been possible.

We will also explore the application of data mining methods to develop a clinical decision support system based on disease profiles/patterns, clinical historical data and other potential factors available in Farmfile database.

Objective
1. Production of a modified version of the Farmfile simulation model (project ED1039) considering the variability of the classification by syndrome, the rate of submission and the unit level (month 6)

2. Application of a selected pool of statistical methods for outbreaks detection with the simulated data and present results to project team (month 12).


3. Assessment of the impact of the ‘limited’ versus ‘reasonable’ testing DNRs on the sensitivity of the selected outbreak detection model as in point 1 (month 18).

4.- Development of a clinical diagnosis support system (CDSS) for cattle submissions to VLA-RL (month 24).

Project Documents
• FRP - Final Report : Final report for SE4301 May 2013   (1409k)
Time-Scale and Cost
From: 2010

To: 2012

Cost: £98,671
Contractor / Funded Organisations
Veterinary Laboratories Agency
Keywords
Animal Health              
Control              
Plants and Animals              
Statutory and Exotic Viruses              
Veterinary Surveillance