Feasibility of a portable NIR-based PAT approach for discriminating fresh and thawed octopus along the supply chain

The distinction between fresh and thawed octopus is a critical issue in seafood quality control, due to the high commercial value of fresh products and the risk of fraudulent practices. Although other techniques, such as histological analysis, have been evaluated as a routine screening technique, their destructive nature makes them unsuitable for routine on-site inspections along the supply chain. Near-infrared (NIR) spectroscopy combined with chemometrics represents a promising non-destructive alternative; however, industrial variability and process-oriented implementation must be considered. In this study, the performance of a portable MicroNIR system was evaluated for the discrimination of fresh and thawed Octopus vulgaris under realistic supply chain conditions. The dataset was designed to reflect industrial variability, including different suppliers, fishing areas, fishing methods, and thawing procedures. The Partial Least Squares Discriminant Analysis (PLS-DA) models achieved classification rates of approximately 85% in calibration and cross-validation, while external prediction performance decreased to approximately 75%, particularly for thawed samples. To translate the model results into an operational quality control strategy, a Process Analytical Technology (PAT)-oriented decision logic was developed based on the mean and variability of replicate predictions. An acceptance zone was defined for fresh samples, ensuring that no thawed sample was incorrectly accepted as fresh in the external test set. These results demonstrate the feasibility of integrating portable NIR spectroscopy into a risk-based and online PAT framework for octopus authenticity assessment, prioritizing fraud prevention and robust decision-making.