Multivariate fusion for identification of geographic origins and varieties of coix seed.

To ensure the authenticity of geographic origins and varieties of coix seeds and their preliminary processed products, this study focused on their viscosity characteristics and stable isotopes before and after gelatinization. The feasibility of identifying coix seed origins from Guizhou and Liaoning, China, and distinguishing between big coix seed (BCS) and small coix seed (SCS) based on viscosity parameters, protein and lipid content, stable isotope ratios and near-infrared spectra was investigated. Significant differences were observed in peak viscosity, trough viscosity, final viscosity, δ²H, and δ¹⁸O values between Guizhou and Liaoning coix seeds, as well as between BCS and SCS. Stable isotopic changes before and after gelatinization were consistent, with higher δ¹⁸O values in gelatinized coix seeds. PLS-DA models achieved prediction accuracies of 92.00% for Guizhou samples, 88.00% for Liaoning samples, and 84.00% for BCS and SCS, providing effective methods for the large-scale traceability of coix seeds' geographic origins and varieties.