Isotope-assisted data mining techniques for authenticating rice across Chinese markets

Area of Science:

• Food authenticity and traceability
• Agricultural science
• Analytical chemistry

Background:

• Accurate labeling of rice origin and cultivar is crucial for consumer trust and fair trade.
• Stable isotope analysis offers a powerful tool for geographical and varietal authentication.
• Current market practices may involve misrepresentation of rice products.

Purpose of the Study:

• To authenticate the geographical origin and cultivar (Japonica vs Indica) of rice sold in Chinese markets.
• To evaluate the effectiveness of stable isotope analysis combined with chemometrics and machine learning for rice verification.
• To identify potential mislabeling issues in the rice supply chain.

Main Methods:

• Stable isotope analysis of carbon (δ13C), nitrogen (δ15N), oxygen (δ18O), and hydrogen (δ2H) using isotope ratio mass spectrometry (IRMS).
• Multivariate statistical analyses including Principal Component Analysis (PCA) and t-distributed Stochastic Neighbor Embedding (t-SNE) for data visualization.
• Development and application of machine learning models (kNN, SVM, RF) for classification and origin verification.

Main Results:

• Significant isotopic differences were observed between authentic field samples and market-sourced rice, indicating potential mislabeling.
• t-SNE provided superior visualization of regional and cultivar separation compared to PCA.
• Machine learning models achieved high accuracy on authentic samples but showed reduced performance on market samples, highlighting widespread origin mislabeling.

Conclusions:

• Stable isotope analysis is a reliable method for authenticating rice origin and cultivar.
• Machine learning models, particularly RF, show promise for rice authentication, but require robust training data.
• The study strongly suggests prevalent mislabeling of rice origin and cultivar in the Chinese market, necessitating improved regulatory oversight.