Pharmacognostic standardization and machine learning-based investigations on Akebia quinata and Akebia trifoliata

Abstract

Currently, Akebiae Caulis is being used in clinical practice, but there are few reseaches on its different varieties. To ensure the accuracy and effectiveness of clinical practice, this study distinguished the Akebia quinata (Thunb.) Decne. and Akebia trifoliata (Thunb.) Koidz, using organoleptic evaluation, microscopic observation, fluorescence reaction, physicochemical properties, thin-layer chromatography, IR spectroscopy, HPLC, four machine learning models, and in vitro antioxidant methods. Analysis of the powders of these two varieties using optical microscopy revealed the presence of starch granules, cork cells, crystal fibers, scalariform vessels, and wood fibers. Scanning electron microscopy revealed the presence of scalariform vessels, pitted vessels, wood fibers, and calcium oxalate crystals. Several tissues, including the cork layer, fiber population, cortex, phloem, pith, xylem, and ray, were found in the transverse section. In addition, thin-layer chromatography was used to identify two components: oleanolic acid and calceolarioside B; 11 common peaks were identified in 15 batches of SAQ and 5 batches of SAT by using HPLC. Support vector machine, BP neural networks, and GA-bp neural networks were able to predict 100% accurately of the different origins of stem of Akebia quinate (Thunb.) Decne (SAQ) and Akebia trifoliata (Thunb.) Koidz (SAT). Extreme learning machine achieved a correct rate of 87.5%. Meanwhile, Fourier-transform infrared spectroscopy fingerprint identified nine characteristic absorption peaks of the secondary metabolites of SAQ and SAT. 2,2-Diphenyl-1-1-picrylhydrazyl experiment revealed that the IC₅₀ values of SAQ and SAT extracts were 155.49 and 128.75 μg/ml, respectively. For the 2,2'-azino-bis-(3-ethylbenzothiazoline-6-sulfonic acid) assay, the IC₅₀ value of SAT extract was found to be 269.24 μg/ml, which was lower than that of SAQ extract (IC₅₀ = 358.99 μg/ml). This study successfully used different methods to differentiate between A. quinata (Thunb.) Decne. and A. trifoliata (Thunb.) Koidz., to help decide on which type to use for clinical application.

Keywords: antioxidant activity; machine learning; microscopy; stem of Akebia quinata (Thunb.) Decne; stem of Akebia trifoliata (Thunb.) Koidz.