Data-Driven MOX Chemosensing for Beer Discrimination: Towards Rapid Food Quality Screening

Beer quality assessment increasingly requires rapid and scalable analytical tools for product discrimination and authenticity control. In this study, a data-driven metal oxide semiconductor (MOX) chemosensing approach was investigated for the discrimination of commercial lager beers with different alcohol contents and brands. Alcoholic and alcohol-free beer samples from four commercial brands were analyzed using a six-element SnO₂-based MOX sensor array, and the resulting response patterns were classified using supervised machine-learning algorithms. Headspace solid-phase microextraction gas chromatography-mass spectrometry (HS-SPME-GC-MS) was employed as a reference technique to characterize volatile organic compound profiles and support the interpretation of sensor-based fingerprints. GC-MS analysis highlighted a shared volatile backbone dominated by fermentation-related compounds, while also revealing brand- and category-dependent differences in VOC distribution. The MOX sensor array captured these differences as multidimensional volatile fingerprints. Machine-learning models achieved high classification performance in brand-matched alcoholic versus alcohol-free comparisons, with balanced accuracy ranging from 0.937 to 1.000, while brand discrimination within the same category reached balanced accuracy values of 0.875 (alcoholic) and 0.933 (alcohol-free). These results highlight MOX-based chemosensing combined with data-driven analysis as a rapid, portable platform for beer discrimination, with applications in food quality screening, authenticity assessment, and at-line monitoring.