Development of new strategies to model bovine fallen stock data from large and small subpopulations for syndromic surveillance use

The continuous monitoring of fallen stock mortality in bovine farms has been demonstrated in different studies to have potential as an important component of veterinary syndromic surveillance. However, as far as we know, the usefulness of these systems to detect abnormal events in near real-time in the field has not been assessed. To implement this type of system, a number of challenges must be faced. The main difficulties are associated with the non-specific nature of fallen stock data, since multiple events may cause bovine mortality at farm level. Moreover, these data are originated from heterogeneous subpopulations that can be clustered and studied in accordance with different traits (e.g. production type, type of farm and/or individuals, husbandry and environmental conditions, or administrative level). In this study, we present the main pillars of a syndromic system to collect continuous fallen stock data from a specific region and to model time series and detect abnormal events at large and small scale.