The spoilage of dairy products often manifests itself via the generation of offensive off-odors and off-flavors. Typically, this is due to microbial metabolism and oxidative reactions that cause the accumulation of volatile organic compounds (VOCs) in the package over storage time. There has thus been great interest towards identifying spoilage-indicating VOCs and their critical concentration levels corresponding with different spoilage statuses. However, this is a complex task that calls for advanced data analysis. Each food product has its own characteristic volatilome that changes over storage time; furthermore, the volatilome typically consists of several compounds that may be present in different concentration magnitudes and have different olfactory impact. Analyzing and interpreting the VOC datasets thus calls for multivariate statistical analysis and modelling. In this presentation, different multivariate methods for characterizing the volatilome of dairy products packed under gaseous atmospheres are discussed. Firstly, an overview is given of the identification criteria that should be taken into account when aiming at identifying volatile spoilage indicators. A comparison of classical and novel multivariate methods is presented, using examples from different dairy products packed under gaseous atmospheres and stored at refrigeration temperatures (4 °C). Special emphasis will be given on Latent Dirichlet Allocation (LDA), a novel method with promising potential in food spoilage analysis.

Multivariate methods in dairy spoilage characterization

ACS Spring 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.