Real-time assessment and future forecasting of extra virgin olive oil oxidation using near-infrared spectroscopy throughout shelf life

Lipid oxidation is a key determinant of the shelf life of extra virgin olive oil (EVOO). Conventional lipid oxidation assessments using near-infrared (NIR) spectroscopy are largely constrained in “snapshot” analysis at a single time point under specific conditions. This study addresses this limitation with two novel approaches: (1) real-time assessment of ongoing lipid oxidation; (2) future forecasting of lipid oxidation progression under realistic conditions using 50 commercial EVOO products. Samples were subjected to short-term thermal oxidation (90°C, 12 hours) and long-term simulated storage (24°C, 32 months). Unified partial least squares regression (PLSR) models, incorporating three distinct oxidation states (pre-oxidation, after short-term heating, and after long-term storage), demonstrated predictive capacity suitable for rough screening (R 2 =0.930 and 0.823; RPD = 3.78 and 2.37 for calibration and validation). Furthermore, future forecasting models developed solely from pre-oxidation NIR spectra (at package opening) indicated the feasibility of early shelf-life prediction for both short-term heating (R 2 =0.818 and 0.502; RPD = 2.36 and 1.36 for calibration and validation) and long-term storage (R 2 =0.801 and 0.470; RPD = 2.24 and 1.28 for calibration and validation). Spectral analysis identified a water-related band at 5264 cm -1 as a critical physicochemical indicator, suggesting that the distribution of prooxidants and/or antioxidants, possibly reflected by water content, may play a critical role in forecasting future oxidation progression. This work introduces a framework for both continuous assessment and early forecasting of EVOO shelf life by expanding NIR spectroscopy from a descriptive technique to a powerful forecasting platform. • Real-time & pre-oxidation NIR for extra virgin olive oil (EVOO) assessment. • Unified PLSR model accurately predicts peroxide value (POV) with real-time NIR. • Moderate shelf-life forecasting was achieved using NIR spectra at package opening. • Water influence highlights the importance of pro/antioxidant localization. • Developed PLSR models enhance efficiency for EVOO shelf-life assessment.