Artificial neural networks application on the prediction of vacuum cooling process in commercial scale and quality of ready-to-eat mixed fresh-cut lettuces

Abstract

Consumer demand for fresh-cut leafy vegetable tended to be increasing due to the image of nutritious and health benefit. However, the product still faced with challenge of short shelf life and fast quality reduction, especially enzymatic browning which was considered as one of the most important attributes limiting the shelf life of product. This study aims at investigating the possibility of applying non-chemical technologies, namely, vacuum precooling technology and equilibrium modified atmosphere packaging (EMAP), to prolong the shelf life of ready-to-eat mixed fresh-cut lettuces. Moreover, artificial neural networks (ANNs) were also evaluated for their performance on vacuum precooling process parameters prediction. Firstly, the optimal vacuum precooling process parameters for each fresh-cut lettuce, including frillice iceberg, romaine, and red oak leaf lettuce, were optimized using response surface methodology (RSM). The process parameters which could reduce the produce’s temperature to 4±1°C with the lowest amount in weight loss percentage, cooling time, and energy consumption, were chosen. In this study, the optimal vacuum precooling process parameters of initial produce’s temperature, target pressure, and holding time, were investigated for frillice iceberg lettuce (22.5°C, 6 mbar, and 6 min), romaine lettuce (22.0°C, 6 mbar, and 7 min), and red oak leaf lettuce (22.2°C, 6 mbar, and 5 min). Vacuum precooled fresh-cut lettuces were mixed with the ratio of 1:1:1 to obtain the net weight of 120 g, covered by polypropylene (PP) packaging, and kept in a commercial refrigerator at 4±1°C for further study on quality changes during the storage time, as well as, for indicating the shelf life using the microbiological and visual quality score as criteria. The study revealed that the shelf life of vacuum precooled ready-to-eat mixed fresh-cut lettuces (7.2±0.4 d) was longer than non-precooled one (3.8±0.9 d). On the 3rd day of storage, vacuum precooled ready-to-eat mixed fresh-cut lettuces presented significantly better quality (p≤0.05) than the non-precooled sample, as a higher content of chlorophyll, total phenolic compounds, and anthocyanins, as well as, higher visual scores for freshness, browning at the cutting edge, and overall acceptance attributes. Furthermore, a significantly lower (p≤0.05) amount of enzyme activity, total aerobic bacteria, yeast, and molds, observed in the vacuum precooled sample proved the efficiency of the vacuum precooling process on quality preservation and shelf life extension for ready-to-eat mixed fresh-cut lettuces. Consequently, vacuum precooled ready-to-eat mixed fresh-cut lettuces were packed in equilibrium modified atmosphere packaging (EMAP) of 3 different oxygen transmission rates (OTR) of 10,000-12,000, 12,000-14,000, and 14,000-16,000 cm3/m2 d atm, before studying on their quality changes during storage at 4±1°C and pointing out the shelf life. The results revealed that the efficacy of EMAP on maintaining of oxygen concentration close to the normal atmospheric condition than polypropylene packaging, example, headspace oxygen concentration on the 7th day of storage in PP and EMAP from low to high OTR were 12.2±1.2, 17.4±0.7, 17.9±0.9, and 18.2±0.9%, respectively. Besides, ready-to-eat mixed fresh-cut lettuces in EMAP remained higher chlorophyll residual, as well as, obtained higher visual evaluation scores for browning at the cutting edge, and overall acceptance attributes than the sample in PP packaging, significantly (p≤0.05). Although EMAP caused significantly higher (p≤0.05) weight loss percentage in the sample (0.39±0.06, 0.40±0.03, and 0.46±0.02%, respectively) than PP packaging (0.13±0.03%), no significant effect on the freshness attribute was observed. EMAP could extend the shelf life of ready-to-eat mixed fresh-cut lettuces to 9.0±1.0, 8.8±0.8, and 8.6±0.9 d, respectively, longer than the shelf life of PP packaged sample (7.2±0.4 d). EMAP with OTR of 10,000-12,000 cm3/m2 d atm presented as the most appropriate packaging for preserving the quality and prolonging the shelf life of vacuum precooled ready-to-eat mixed fresh-cut lettuces. In the last part of research, the data of vacuum precooling process parameter optimization were analyzed using feed-forward backpropagation for the purpose to test the possibility to use artificial neural networks (ANNs) on vacuum precooling process parameters prediction. Thirty-four data were randomly separated into 3 groups, called training (approximately 70% of all data, totally of 24 data), testing (approximately 15% of all data, totally of 5 data), and validating set (approximately 15% of all data, totally of 5 data). The optimal artificial neural network models for the final produce's temperature prediction of frillice iceberg, romaine, and red oak leaf lettuce were 3-18-1, 3-15-1, and 3-21-1, respectively, while the optimal models for weight loss percentage prediction were 3-9-1, 3-18-1, and 3-18-1, respectively. Predicted and actual values were compared, ANNs presented better prediction performance on final produce's temperature and a weight loss percentage of fresh-cut frillice iceberg, romaine, and red oak leaf lettuce (R2adj 0.894-0.952 and 0.819-0.952, respectively), over the performance of RSM (R2adj 0.657-0.959 and 0.586-0.764, respectively). Furthermore, ANNs models exhibited a higher possibility for applying with the other data (R2adj 0.761-0.925) than RSM equations (R2adj 0.330-0.725). Thus, artificial neural networks were one of the proper methods to be used for the vacuum precooling process parameters of ready-to-eat fresh-cut lettuces.