Deep learning with keratin/poly lactic acid/thymol/mulberry anthocyanin nanofilm in pork preservation and freshness assessment

Abstract In this study, an intelligent nanofilm was developed for pork freshness monitoring and preservation. A 10% pig nail keratin and 5% polylactic acid (PLA) matrix was fabricated via electrospinning. The nanofilm, designated keratin/PLA/Thy5/MA3, was incorporated with 5% thymol (Thy) and 3% mulberry anthocyanins (MA). Structural characterizations of the nanofibers were confirmed, and the antioxidant activity of the Thy5/MA3-loaded nanofibers was 87.21% ± 0.1%. The keratin/PLA/Thy5/MA3 film exhibited antimicrobial activity against Staphylococcus aureus (22.19 ± 1.29 mm), Escherichia coli (13.48 ± 0.76 mm), Aspergillus niger CBS 513.88 (23.75 ± 0.56 mm), and yeast (36.47 ± 0.64 mm). The nanofilm extended the shelf life of pork by delaying deterioration for up to 12 d and was coupled with classical machine learning algorithms with random forest (RF) for non-destructive, quantitative determination of total volatile basic nitrogen (TVB-N), thiobarbituric acid reactive substances (TBARS), total viable count (TVC), and pH. The coefficient of determination (R2) values of the training set ranged from 0.770 to 0.847, indicating that the model provided a good fit for the data. A clear color change was observed during pork spoilage, transitioning from light pink to light brown, and then to brownish-green, which was highly correlated with pork deterioration stages. Notably, the deep features of the obtained keratin/PLA/Thy5/MA3 images were extracted through model (ResNet50), achieving 100% classification accuracy in distinguishing the pork freshness levels (fresh, less fresh, and spoiled). In conclusion, the multifunctional keratin/PLA/Thy5/MA3 composite nanofiber film, combined with advanced learning models, shows great potential for pork preservation and real-time freshness monitoring.