Optimization of accelerated solvent extraction of fatty acids from Coix seeds using chemometrics methods

This study investigated the optimization of accelerated solvent extraction (ASE) of fatty acids (FAs) from three Coix seeds (SCS small Coix seed; BCS big Coix seed; TCS translucent Coix seed) by chemometrics methods. Partial least-squares regression (PLSR) and backpropagation neural network (BPNN) were applied to build models that reflect the relationship between content of FAs and extraction conditions (temperature, time, and extraction solvent). Genetic algorithms (GAs) and particle swarm optimization (PSO) were utilized to optimize the combination of extraction conditions. The composition of FAs was analysed by gas chromatography-mass spectrometry (GC-MS). The PLSR models could reflect the relationship of FA content in both BCS and SCS and extraction conditions well, while the BPNN model was more suitable for TCS. The optimal extraction conditions for BCS and SCS were obtained by GAs, whereas those of TCS were obtained by PSO. The FA compositions of the three Coix seeds exhibited differences. The results show that ASE combined with chemometrics methods can rapidly and effectively obtain the optimal conditions for the extraction of FAs from Coix seed and there are differences in the extraction conditions and compositions of FAs among different varieties of Coix seed, but all the extraction time is shorter than other extractions methods.