Machine Learning Approach Using a Handheld Near-Infrared (NIR) Device to Predict the Effect of Storage Conditions on Tomato Biomarkers

Abstract

Minimizing food waste is critical to future global food security. This study aimed to assess the potential of near-infrared (NIR) spectroscopy combined with machine learning to monitor the stability of tomato fruit during storage. Freshly harvested U.K.-grown tomatoes (n = 135) were divided into five equally sized groups, each stored in different conditions. Absorbance spectra were obtained from both the tomato exocarp and locular gel using a portable NIR spectrometer, capable of connecting to a mobile phone, before subsequent chemometric analysis. Results show that support vector machines can predict the storage conditions and time-after-harvest of tomatoes. Molecular biomarkers highlighting key wavelength and molecular changes due to time and storage conditions were also identified. This method shows potential for the development of this approach for use in the field to help mitigate the environmental and economic impacts of food waste.