Development and validation of NIR model using low-concentration calibration range: rapid analysis of Lonicera japonica solution in ethanolprecipitation process

Abstract

A strategy for both low-concentration calibration range and large number of sample sets selection in the development of a PLS model is studied. A novel approach based on accuracy profile validated the accuracy and precision of the PLS model. The strategy was applied to the determination of chlorogenic acid content in Lonicera japonica using ethanol precipitation by near-infrared (NIR) transmission spectroscopy. The results found the determination coefficient (R²), standard errors of calibration and prediction (SEC and SEP) were 0.9648, 71.2 ppm and 74.9 ppm, respectively. The further study showed that PLS model could be used to determine chlorogenic acid content based on accuracy profile, which has a lower limit of quantification (LLOQ) ∼1700 ppm. Analytical properties such as accuracy, precision, range and linearity from validation criteria also demonstrated the feasibility of the strategy using a low-concentration calibration set in the PLS model, paving the way for analyses in Chinese Herbal Medicine (CHM) applications.