RESUMO
Celery seeds are medicinal herbs used for the prevention and treatment of gout as these have the ability to inhibit the activity of xanthine oxidase and reduce the concentration of serum uric acid. In this study, the relationship between xanthine oxidase inhibitory effects and high-performance thin-layer chromatography data of celery seed extracts was established using multilayer neural network (MNN) in combination with principal component analysis (PCA). The constructed MNN-PCA model was stable and had accurate prediction ability with coefficient of determination = 0.9998, leave-one-out coefficient = 0.7371, root mean square error = 0.0025, and mean absolute deviation = 0.0019 for the training set and coefficient of determination = 0.8124, root mean square error = 0.0784, and mean absolute deviation = 0.0645 for the test set. This model can be used to identify the main compounds related to the xanthine oxidase inhibitory effect of celery seed extract. These results can be applied not only to celery extract but also to other herbal medicines.