Application of a data fusion strategy combined with multivariate statistical analysis for quantification of puerarin in Radix puerariae

Abstract Motivated by the wide use of Radix puerariae (RP) in the food and pharmaceutical industries, a reliable approach was developed for the quantitative analysis of puerarin from RP. Data fusion strategy based on near infrared (NIR) and ultraviolet (UV) spectra is proposed herein to establish a reliable partial least squares (PLS) regression model for predicting the puerarin content, with critical variables being selected by iPLS algorithm. The developed PLS model performed better than that established only using NIR or UV spectra. Compared with an independent NIR or UV spectra model, low-level data fusion (LLDF) reduced the predicted error to a lower root mean square error of prediction (RMSEP) of 0.418, and a higher Rp value of 0.974 and RPD value of 4.295, indicating that there was a synergistic effect between the NIR and UV spectra for determination of puerarin. It was shown that the data fusion strategy coupled with chemometric methods effectively enhanced the model performance, and this combination could be a promising tool for accurate determination of components that cannot easily be quantified with individual spectral data.