Rapid and Accurate Differentiation of Fresh and Frozen-Thawed Pork Using Rapid Evaporative Ionization Mass Spectrometry-Based Chemometrics and Machine Learning

• REIMS method was used for identifying fresh and frozen-thawed pork • Fat-varied pork model overcomes single-cut limitations with enhanced generalizability • The validation test of the proposed method with 83.85% accuracy was performed • Key frozen-thawed biomarkers identified via chemometrics and machine learning Temperature fluctuations during cold chain transportation and storage of meat induce frozen-thawed cycles, significantly degrading meat quality and posing risks to food safety. The misrepresentation of frozen-thawed meat as fresh is a critical concern. This study addresses this challenge by integrating Rapid Evaporative Ionization Mass Spectrometry (REIMS) with chemometrics and machine learning to rapidly and accurately distinguish between fresh and frozen-thawed pork. A PCA-LDA model classified high-fat and lean meat cuts classification (foreleg, hindleg, backfat), achieving an accuracy of 99.72%. Model optimization further improved its generalizability. OPLS-DA analysis identified key metabolite markers associated with frozen-thawed cycles, including m/z 594.3049 (PS 21:1; O), m/z 595.2889 (LPI 18:2), and m/z 279.2330 (FA 18:2), reflecting metabolic changes such as fat oxidation and cell membrane disruption. This approach enables real-time meat quality monitoring, offering high accuracy for fraud detection and food safety assurance.