NIR spectroscopy for quality assessment and shelf-life prediction of kiwifruit

In this study, a non-destructive quality testing method along with shelf-life prediction of Xu Xiang ready-to-eat kiwifruit were developed using near-infrared spectroscopy (NIR) techniques. Several traditional quality indicators (hardness, soluble solids content, and dry matter) were evaluated. Partial least squares regression (PLS) was used to predict the intrinsic quality attributes of the samples. Competitive adaptive reweighted sampling algorithm (CARS) and uninformative variable elimination (UVE) algorithm were used to select the characteristic wavelengths. Prediction models for hardness, soluble solids content and dry matter were developed. The results showed that the prediction ability of the models could be improved by screening the characteristic wavelengths of CARS and UVE. Among them, the CARS-SNV-PLS model based on soluble solids had the best prediction ability (RMSEP of 0.430 and Rp² of 0.958). Then, an NIR-based residual shelf-life prediction model was obtained by linking the measured quality indicators to the residual shelf-life, which was well validated with an RMSEP of 1.64 and an Rp² of 0.939. Therefore, this study demonstrated the potential of combining CARS, SNV, and PLS for the non-destructive testing of ready-to-eat kiwifruit to provide technical support and solution.