Accuracy of near infrared spectroscopy to predict quality of pork and pork products including samples of Krškopolje and Turopolje pigs

Il s'agit d'un article correspondant à un poster présenté en conférence Animal Science Days, Heilbruun, Austria, Sept 2017 Notice à Reprendre pas de clé UTInternational audienceStudy demonstrates the preliminary results of the evaluation of pork and pork products of local Slovenian (Krškopolje) and Croatian (Turopolje) pig breeds using near infrared spectroscopy (NIRS) conducted in the frame of European Union H2020 project TREASURE. For that purpose, samples from meat and products of two local pig breeds were collected, scanned with near infrared spectroscopy apparatus and chemically analysed (for proximate composition, fatty acids composition, proteolysis index, salt content and water activity). Data obtained were added to the database of previously collected samples and prediction models were recalibrated and reassessed. In general, NIRS calibration models are considered to be fit for purpose when the requirements (chemometric parameters) for screening purposes are met. In the present study, the quality of recalibrations using the samples from local pig breeds confirmed practical applicability for majority of studied quality traits. Further efforts are needed to enlarge the database with additional samples from local pig breeds to improve the robustness of the models and to test the calibrations on the independent sets of samples (i.e. with external validation)