Hydrothermal-Assisted Extraction of Cinnamomum camphora Essential Oil Optimized by Machine Learning

This study presents a combined methodology employing hydrothermal pretreatment alongside steam distillation to achieve efficient extraction of essential oils from the leaves of Cinnamomum camphora. The optimization of process parameters was conducted utilizing Response Surface Methodology (RSM) and a Genetic Algorithm-optimized Artificial Neural Network (GA-ANN) model. To prevent data leakage, all replicates of the same experimental condition were averaged, yielding 63 unique data points, which were then split into training, validation, and test sets under a grouped constraint. The optimized GA-ANN model demonstrated good predictive performance, achieving a coefficient of determination (R²) of 0.954 on an independent test set. The GA-ANN identified the following optimum conditions: hydrothermal liquid-to-solid ratio of 3.8 mL/g, temperature of 160 °C, and duration of 117 min, followed by steam distillation with a liquid-to-solid ratio of 14.3 mL/g for 62 min. Under these parameters, an essential oil yield of 4.59 ± 0.02 % was achieved, which marked a 52 % enhancement over conventional steam distillation and also surpassed the yield obtained under the RSM-optimized conditions (4.43 %). The GA-ANN model exhibited superior generalization ability, with an external validation R² of 0.954 compared to 0.821 for RSM. Mechanistic insight was gained through systematic monitoring of structural changes via scanning electron microscopy (SEM) and Fourier transform infrared (FTIR) spectroscopy. It was demonstrated that hydrothermal pretreatment induced structural alterations in the lignocellulosic matrix of the cell walls. SEM and FTIR analyses revealed morphological changes and spectral shifts indicative of structural alterations, which collectively are likely to have facilitated the release and diffusion of essential oils during distillation. Gas chromatography-mass spectrometry (GC‑MS) profiling tentatively identified 22 plant‑derived constituents in the hydrothermal‑assisted oil and 12 plant‑derived constituents in the conventional steam‑distilled oil, accounting for 99.77 % and 99.88 % of the total volatile content, respectively.