Colorimetric quality monitoring of carob syrup using smartphone‑embedded machine learning

Carob (Ceratonia siliqua L.) syrup is a functional and nutritionally rich product widely used in food applications. This study compared two preparation methods, a rapid 30-min decoction (syrup 1) and a conventional 24-h maceration (syrup 2), and evaluated their conversion into molasses, using smartphone-based color assessment and artificial neural network (ANN) modeling for real-time monitoring. Physicochemical analysis showed that both syrups had slightly acidic pH and similar total soluble solids, but syrup 1 exhibited higher density, conductivity, and soluble sugar content, whereas syrup 2 retained higher mineral levels due to its prolonged extraction. Moreover, syrup 2 contained higher polyphenols (2311 vs. 581 mg GAE/100 g), flavonoids (199 vs. 128 mg QE/100 g), and tannins (93.8 vs. 595.3 mg TAE/100 g). These differences translated into superior functional properties, including stronger antioxidant activity (89.5% DPPH inhibition vs. 71.4%) and greater emulsifying capacity (61.1% vs. 35.5%). No antibacterial effect was observed against S. aureus or E. coli. Color analysis revealed progressive darkening during concentration, with a steady decline in RGB and L a b parameters. The ANN regression model accurately predicted syrup quality from smartphone-captured images, showing strong correlations between color indices and chemical/functional attributes. Overall, prolonged maceration improved the nutritional and functional quality of carob syrups, while smartphone-based ANN monitoring proved effective for rapid, non-destructive quality control in food processing.