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

Area of Science:

• Analytical Chemistry
• Food Science
• Sensor Technology

Background:

• Rapid and scalable analytical tools are crucial for beer quality assessment, product discrimination, and authenticity control.
• Metal oxide semiconductor (MOX) chemosensing offers a promising avenue for developing such tools.

Purpose of the Study:

• To investigate a data-driven MOX chemosensing approach for discriminating commercial lager beers based on alcohol content and brand.
• To evaluate the performance of supervised machine-learning algorithms in classifying sensor data.

Main Methods:

• Utilized a six-element SnO2-based MOX sensor array to analyze alcoholic and alcohol-free beer samples from four commercial brands.
• Employed headspace solid-phase microextraction gas chromatography-mass spectrometry (HS-SPME-GC-MS) as a reference technique for volatile organic compound (VOC) profiling.
• Classified MOX sensor response patterns using supervised machine-learning algorithms.

Main Results:

• MOX sensor array successfully captured multidimensional volatile fingerprints, reflecting brand- and category-dependent VOC differences.
• Machine-learning models achieved high classification performance: balanced accuracy of 0.937-1.000 for alcoholic vs. alcohol-free comparisons.
• Brand discrimination within the same category reached balanced accuracy of 0.875 (alcoholic) and 0.933 (alcohol-free).

Conclusions:

• MOX-based chemosensing combined with data-driven analysis provides a rapid and portable platform for beer discrimination.
• This approach has significant applications in food quality screening, authenticity assessment, and at-line monitoring.
• The study demonstrates the potential of sensor technology for objective and efficient beverage analysis.