Translating Big Data into Smart Data for Veterinary Epidemiology

VanderWaal, Kimberly; Morrison, Robert B.; Neuhauser, Claudia; Vilalta, Carles; Perez, Andres M.

The use of universally recognized data formats is important for enhancing connectivity of data between different sources (e.g., different laboratories and clinics) (47, 48) . Human diagnostic laboratories have long had standardized vocabularies for health records, such as HL7, LOINC, and SNOMED, but standardized vocabularies are typically underused in veterinary medicine (49) . To overcome this challenge, the Clinical Wildlife Health Initiative has worked to create a standardized terminology for clinical signs in admitted wildlife. This standardization enhances the ability to pool data from multiple clinics into a common dataset, thus creating a powerful network of clinics in which health trends could be tracked (50) . Similar efforts are underway for swine diagnostic data (48) . Related to the current lack of data standardization, many sources of animal health data are not readably usable in statistical models, such as "unstructured" text-heavy data (15, 49, 51) . While it is possible to use text-mining techniques to extract information from unstructured text fields in clinical or diagnostic records (52, 53) , homogenizing and naturalizing data into uniform and standardized formats is critical for maximizing accuracy and ensuring smooth automation (8) .