Spectroscopic evaluation demonstrates accurate soluble solids prediction and storage stage classification in kiwifruit, highlighting an effective non-destructive framework for automated sorting.
Kiwifruit is prone to deterioration during storage and transportation. Rapid detection of soluble solids content (SSC) is crucial for evaluating fruit maturity and quality, whereas traditional methods struggle to meet the demands of large-scale production. In this study, ‘Jinyan’ kiwifruit was selected as the research material, and a dynamic online visible/near-infrared spectroscopy (Vis/NIR spectroscopy) system was employed to monitor quality changes in the fruit during the storage phase. Generalized two-dimensional correlation spectroscopy (2D-COS) was applied to investigate the variation patterns of molecular structures corresponding to spectral characteristics under the influence of storage duration. The competitive adaptive reweighted sampling (CARS) algorithm was utilized to select effective characteristic wavelengths. The microscopic structural changes revealed by 2D-COS were compared with the macroscopic spectral features, and classification and prediction models were constructed in combination with machine learning algorithms. The results indicated that the partial least squares-support vector classification (PLS-SVC) model preprocessed by standard normal variate combined with Savitzky-Golay smoothing (SNV+SG) achieved the best performance in distinguishing fruit from different storage periods, with a maximum classification accuracy of 96.3% and an weighted F1-score of up to 0.962. Meanwhile, the partial least squares-support vector regression (PLS-SVR) model built based on the raw spectra exhibited the strongest predictive capacity for SSC, yielding the highest coefficient of determination for the prediction set (RP2) of 0.802 and residual predictive deviation (RPD) of 2.3, along with the lowest root mean square error of prediction (RMSEP) of 0.968°Brix. Consequently, the integration of online Vis/NIR spectroscopy, mechanistic analysis via 2D-COS, and machine learning can not only accurately classify the storage periods of kiwifruit and reliably predict SSC, but also elucidate the internal material transformation rules during fruit storage. This approach provides a technical basis for kiwifruit quality inspection and online sorting.
No takes yet. Share an insight, caveat, or question.
Xiao et al. (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: