Approaches for evaluating veterinary epidemiological models: verification, validation and limitations

• Published in: Revue scientifique et technique (International Office of Epizootics) Vol. 30; no. 2; p. 499
• Main Authors: Reeves, A, Salman, M A, Hill, A E
• Format: Journal Article
• Language: English
• Published: France 01.08.2011
• Subjects: Animal Diseases - epidemiology; Animals; Computer Simulation - standards; Disease Outbreaks - statistics & numerical data; Disease Outbreaks - veterinary; Models, Biological; Reproducibility of Results
• ISSN: 0253-1933
• DOI: 10.20506/rst.30.2.2053

The evaluation of models of the spread and control of animal diseases is crucial if these models are to be used to inform decisions about the control or management of such diseases. Two key steps in the evaluation of epidemiological models are model verification and model validation. Verification is the demonstration that a computer-driven model is operating correctly, and conforms to its intended design. Validation refers to the process of determining how well a model corresponds to the system that it is intended to represent. For a veterinary epidemiological model, validation would address such issues as how well the model represents the dynamics of the disease in question in the population to which this model is applied, and how well the model represents the application of different measures for disease control. Just as the development of epidemiological models is a subjective, continuous process, subject to change and refinement, so too is the evaluation of models. The purpose of model evaluation is not to demonstrate that a model is a 'true' or accurate' representation of a system, but to subject it to sufficient scrutiny so that it may be used with an appropriate degree of confidence to aid decision-making. To facilitate model verification and validation, epidemiological modellers should clearly state the purpose, assumptions and limitations of a model; provide a detailed description of the conceptual model; document those steps already taken to test the model; and thoroughly describe the data sources and the process used to produce model input parameters from those data.