Crocus sativus L. (Iridaceae), commonly known as saffron, is a highly valued medicinal plant often termed 'red gold' for its economic importance and vibrant flower colour. Among its bioactive constituents, safranal is notable for its antioxidant, anti-inflammatory, and neuroprotective properties. However, standardising its extraction remains challenging due to variable processing conditions. This study optimised safranal extraction using response surface methodology (RSM) and adaptive neuro-fuzzy inference system (ANFIS). Initial screening showed that microwave-assisted extraction (MAE) provided the highest yield (0.950% in 30 min), while ultrasound-assisted extraction (UAE) achieved 0.808% in only 10 min. Extraction parameters including temperature, irradiation time, solvent concentration, particle size, solvent-to-solute ratio, pH, and extraction steps were systematically evaluated. Box-Behnken design (BBD) optimised conditions (60 °C, 10 min, 50% ethanol, 7.5 g/mL solid-to-solvent ratio) yielded a maximum of 0.605% safranal. Box-Behnken design (BBD) showing an excellent fit with the quadratic model (R2 = 99.38%). The adaptive neuro-fuzzy inference system (ANFIS) model further refined these predictions. High-performance thin-layer chromatography (HPTLC) quantified the yield of safranal (0.605%) under optimised conditions. The ANFIS model with 81 fuzzy rules successfully modelled non-linear interactions. The HPTLC method was validated with a Limit of Detection (LOD) of 120 µg/spot and a Limit of Quantitation (LOQ) of 360 µg/spot. The results demonstrate a rapid, eco-friendly, and energy-efficient extraction process, which is a sustainable approach for safranal production to support its growing pharmaceutical and fragrance applications.
Keywords: Crocus sativa L.; HPTLC; optimisation; response surface methodology; safranal.