IMPLEMENTING BIG DATA ANALYTICS APPROACHES TO IMPROVE FOOD QUALITY AND MINIMIZE FOOD WASTE AND LOSS

Abstract

This study was conducted to explore the application of non-invasive, rapid advanced technological enhancements and to combine big data analytics methods to detect muscle quality issues such as woody breast (WB), white striping (WS), and spaghetti meat (SM) conditions that arise in fast-growing broilers within the poultry industry. Results obtained from the rapid identification of myopathies in chicken breast fillets experiments using BIA (Hand-held and plate BIA) collected data analyzed with supervised and unsupervised machine learning algorithms (Support Vector Machine (SVM), Back Propagation Neural Network (BPNN), K-nearest neighbor (k-NN), Fuzzy C Means (FCM) clustering and K-means clustering) have shown that model developed with SVM separated WB with a higher accuracy of 71.0% for normal , 59.9% for moderate, 81.4% for severe WB. Compared to SVM, the BPNN training model accurately (100%) separated normal WB fillets with and without SM demonstrating the ability of BIA to detect SM. While on the other hand, the modified BIA showed better detection ability for normal chicken breast fillets than the probe BIA setup. In the plate BIA setup, fillets were 80.0% for normal, 66.6% for moderate, and 85.0 % for severe WB. However, hand-held BIA showed 77.78%, 85.71%, and 88.89% for normal, moderate, and severe WB. Plate BIA setup is more effective in detecting WB myopathies and could be installed without slowing the processing line. Breast fillet detection on the processing line can be significantly improved using a modified automated plate BIA. Radio wave frequencies were also used in detection of these myopathic conditions. Results obtained from this experiment indicates that pre-processed data (False discovery rate, predictor screening, and variable clustering) with identified signature frequencies were used to develop classification-based models using Back Propagation Neural Network (BPNN) and Support vector machines (SVM). The BPNN model effectively predicts bird myopathies with varying accuracy in stages: Live Birds (83% variance, 87.5%-100% accuracy), Pre-Chill WOG (78% variance, 87.5%-100% accuracy), Post-Chill WOG (91% variance, 69.7%-100% accuracy), Deboned Fillets (85% variance, 66.7%-100% accuracy). It remains sensitive despite 26% misclassification rates. Conversely, the SVM model shows lower sensitivity and specificity (54.8%-69.7% accuracy). BPNN surpasses SVM in predicting myopathies across processing stages. Keywords: Support Vector Machines, Backpropagation Neural Networking, Woody breast, Meat myopathies, Spaghetti meat, Bio-electrical impedance analysis, Machine learning, Artificial intelligence