Rapid detection of yam adulteration using portable NIRS combined with chemometrics

Yam is widely utilized for both culinary and medicinal purposes in China. However, it is frequently subject to adulteration in its powdered form. This study proposed a novel hierarchical modeling strategy integrating portable NIRS (900-1700 nm) for rapid detection of yam powder adulteration. First, 38 pure yam samples, 96 corn starch-adulterated and 96 cassava starch-adulterated samples were used to develop PLS-DA model, achieving 100% classification accuracy. Subsequently, two PLSR models were developed to predict the adulteration levels of corn starch and cassava starch, respectively. Through the spectral preprocessing and variable selection, the corn starch adulteration PLSR model yielded r_val of 0.9979, RMSEP of 1.29% and RPD of 15.40; while the cassava starch adulteration PLSR model achieved r_val of 0.9934, RMSEP of 1.83% and RPD of 8.74, both indicating excellent predictive capability. The results demonstrated that the portable NIRS combined with hierarchical modeling strategy successfully achieved rapid detection of yam adulteration.