Intelligent optimization of cinnamon essential oil extraction: A hybrid RSM and machine learning approach for maximized yield and bioactivity.

To optimize cinnamon essential oil extraction and assess its potential, this study utilized response surface methodology and machine learning models. Based on single-factor experiments, these approaches were employed to predict and compare the optimal extraction process. The RSM-predicted optimal parameters included a material-to-liquid ratio of 1:10.4, distillation time of 105 min, an ultrasonic temperature of 63.0 °C, and an ultrasonic time of 48.8 min, resulting in a predicted yield of 4.61 %(actual:4.56 %). Meanwhile,the machine learning predicted parameters were a material-to liquid ratio of 1:9.3, distillation time of 90 min, an ultrasonic temperature of 55 °C, and an ultrasonic time of 40 min, yielding a predicted 4.65 % (actual: 4.60 %). machine learning demonstrated significant potential for process optimization. GC-MS analysis identified 40 components in the oil, primarily consisting of aldehydes, alkenes, and esters. The most abundant components were trans-cinnamaldehyde (19.76 %), α-copaene (16.98 %), and cinnamyl dimethyl acetal (12.92 %). Antioxidant tests showed a favorable dose-effect relationship, with over 90 % scavenging rates for ABTS•+ and DPPH•free radicals at low concentrations. Antibacterial tests revealed significant inhibition zones against Staphylococcus aureus and Escherichia coli, with minimum inhibitory concentrations of 1 mg/mL. Hemolysis tests confirmed good biocompatibility. Overall, this study elucidated the optimal extraction process, main components, and biological activities of cinnamon essential oil, indicating its potential in food preservation and medical hygiene.