Convenient use of near-infrared spectroscopy to indirectly predict the antioxidant activitiy of edible rose (Rose chinensis Jacq "Crimsin Glory" H.T.) petals during infrared drying.

Abstract Infrared drying (IRD) was used for the dehydration process of rose petals for the purpose of improving drying efficiency as well as retaining product quality. A methodology to predict the antioxidant capacities of rose petals which include DPPH, ABTS radical scavenging capacities and ferric-ion reducing antioxidant power (FRAP) values during infrared drying (IRD) was established in this study. Partial least squares regression (PLSR) and back propagation-artificial neural network (BP-ANN) modelling were used to establish the relationships between the near infrared (NIR) spectrum and the antioxidant capacities. Results of model fitting showed that BP-ANN model displayed higher prediction accuracy than PLSR model for determining the DPPH, ABTS radical scavenging capacities and FRAP of rose petals during IRD based on NIR spectral data. The results obtained indicate that NIR spectroscopic parameters combined with multivariate calibration could be used reliably to predict the antioxidant capacities of IR-dried rose petals via appropriate mathematical models.