Filme inteligente à base de colágeno incorporado com antocianinas da flor de catharanthus roseus e sua aplicabilidade no monitoramento do frescor de oreochromis niloticus por meio de inteligência artificial

Abstract

This study aimed to evaluate the impact of different concentrations of anthocyanin extract (CAE), extracted from Catharanthus roseus flowers, on the development of smart films applied to monitoring the freshness of tilapia fillets (Oreochromis niloticus), using artificial intelligence for real-time detection of changes in fish quality. Hydroalcoholic extraction of anthocyanins was performed from the flowers, while collagen was obtained from tilapia skin by hot water extraction; both were characterized in terms of their physicochemical properties. For the elaboration of the films, collagen and pectin were used as polymeric matrices, incorporating different concentrations of anthocyanin extract in the FC (control), F10, F20, and F30 formulations. The physical, mechanical, thermal, functional, and morphological properties were evaluated, including the response to pH variations. The film with the best performance was selected for application to tilapia fillets stored under refrigeration for eight days. The film's color changes, associated with total volatile basic nitrogen (TVB-N) and pH values, were used to create a database for training artificial intelligence models. Complementary analyses were performed to validate the correlation between the colorimetric response and the degree of fish deterioration, including TBARS, microorganism count, texture, volatile compounds, and sensory evaluation. The results indicated a collagen extraction yield of 6.31%, with a hydroxyproline content of 4.83 g/100 g, and preservation of the triple helix structure confirmed by FTIR. The anthocyanin extract showed a concentration of 4.78 mg CGE/g and high antioxidant activity, with a phenolic profile composed of phenolic acids, flavonoids, and anthocyanins, in addition to color variation as a function of pH, highlighting its potential as a freshness indicator. The films exhibited alterations in physical and morphological properties, such as reduced thickness, transparency, luminosity, and elasticity, due to interactions with the polymeric matrix. The incorporation of the extract did not compromise non-toxicity, water vapor permeability, or mechanical integrity, maintaining adequate solubility and thermal stability. The F30 formulation stood out for its higher antioxidant activity and sensitivity to ammonia. During storage, an increase in pH and TVB-N, a reduction in hardness and chewiness, and the formation of deterioration compounds were observed, accompanied by a loss of sensory acceptability. The colorimetric response of the film correlated with these parameters and was applied in a machine learning system integrated into the SmartFish application, which classified the freshness of tilapia into three levels with 93% accuracy. Thus, the system demonstrated high efficiency and potential for practical application in monitoring fish freshness. Smart films with anthocyanins from Catharanthus roseus have proven effective, especially the F30 formulation, and integration with artificial intelligence has shown a promising approach to increase food safety and reduce waste.