Establishment of a Non-Destructive Analysis Method for the Polyphenol, Flavonoid Content and Antioxidant Activity of Sweetpotato using Near-Infrared Reflectance Spectroscopy
This study developed a calibration model using near-infrared spectroscopy (NIR) for the rapid and non-destructive prediction of functional components—total polyphenols, total flavonoids, and antioxidant activities—in sweetpotato. Analysis of 473 sweetpotato genetic resources revealed significant variation and diversity in total polyphenols (4.67 to 2,419 mg/100g, CV 74.26%), total flavonoids (7.71 to 1,057.67 mg/100g, CV 122.88%), DPPH radical scavenging activity (7.83 to 532.56 mg/100g, CV 111%), and ABTS radical scavenging activity (11.84 to 615.87 mg/100g, CV 70.02%). The NIR spectra (400 to 2,500 nm), following pretreatment and modified partial least squares (MPLS) regression, were utilized to create the calibration model. Cross-validation indicated high predictive performance for total polyphenols (R2 cv 0.895), total flavonoids (0.940), DPPH (0.955), and ABTS (0.940). External validation yielded R2 ev values ranging from 0.623 to 0.726 and RPD values between 1.58 and 2.13, confirming the model's practical applicability for large-scale screening and classification. However, further enhancements are needed to improve the accuracy of quantitative analysis. Future studies should focus on refining the predictive model by incorporating samples from diverse genetic resources and cultivation environments. This research demonstrates that NIR spectroscopy can be effectively employed for the rapid, non-destructive evaluation of sweetpotato functional quality, facilitating the efficient selection of high-value cultivars.