Moisture-adaptive corrections of NIR for the rapid simultaneous analysis of 70 chemicals in tobacco: A case study on tobacco.

[Display omitted] • A comprehensive strategy for investigating tobacco quality. • The strategy including a moisture-prediction model, a moisture-correction model and a compound-prediction model. • NIR prediction models with moisture-adaptive corrections. • Rapid and efficient quantification of 70 chemicals in tobacco. • A Mahalanobis-distance-based method for the geographical discrimination and part identification of samples. Tobacco is one of the most widely cultivated non-food cash crops worldwide, and the quality of tobacco procured from different geographical locations varies considerably. This study proposes a comprehensive strategy for investigating tobacco quality using near-infrared (NIR) prediction models with moisture-adaptive corrections. This strategy enables the rapid and efficient quantification of 70 chemicals in tobacco by reducing the effect of moisture on the NIR spectra of tobacco samples. Additionally, this strategy has been proposed for the geographical discrimination and part identification of tobacco samples, with the Mahalanobis distance analysis, the accuracy of predicted values is higher than 81.5%. This study confirms that the tobacco chemical composition from different regions in China is inconsistent. [ABSTRACT FROM AUTHOR]