Non-invasive prediction of maca powder adulteration using a pocket-sized spectrophotometer and machine learning techniques

Abstract

AbstractDiscriminating different cultivars of maca powder (MP) and detecting their authenticity after adulteration with potent adulterants such as maize and soy flour is a challenge that has not been studied with non-invasive techniques such as near infrared spectroscopy (NIRS). This study developed models to rapidly classify and predict 0, 10, 20, 30, 40, and 50% w/w of soybean and maize flour in red, black and yellow maca cultivars using a handheld spectrophotometer and chemometrics. Soy and maize adulteration of yellow MP was classified with better accuracy than in red MP, suggesting that red MP may be a more susceptible target for adulteration. Soy flour was discovered to be a more potent adulterant compared to maize flour. Using 18 different pretreatments, MP could be authenticated with R2CV in the range 0.91–0.95, RMSECV 6.81–9.16 g/,100 g and RPD 3.45–4.60. The results show the potential of NIRS for monitoring Maca quality.