The purpose is to use the Analytic Hierarchy Process (AHP)-CRITIC method for determining the weight of indicators related to the mixed weighting method, grey correlation analysis, and backward propagation artificial neural network (BP-ANN) to optimize the water extraction process of Qi-Zhi-Fang, providing experimental basis for the optimization of the preparation process and quality standards of Qi-Zhi-Fang. The method uses L9 (34) orthogonal experiment, combined with AHP-CRITIC mixed weighting method to determine the weight coefficients of the components astragaloside, saikosaponin, isoflavone glycoside, and the mass fraction and dry extract yield of cynanchum with off-site alkyl glycoside III, 36'-diglycosyl, and evaluates the comprehensive scores of various factor levels in the orthogonal experiment as evaluation indicators to select the optimal process parameters, and investigates the effects of the number of extractions, extraction time, and solvent usage on the water extraction process of Qi-Zhi-Fang through intuitive analysis, variance analysis, and grey correlation analysis; meanwhile, a BP-ANN analysis model is established to backward predict the optimal extraction process factor levels of this compound and verify the optimized process parameters. The AHP-CRITIC mixed weighting method determines the weight coefficients of 5 indicators, astragaloside, saikosaponin, isoflavone glycoside, and the mass fraction of astragaloside, saikosaponin, isoflavone glycoside, and the dry extract yield are 25.7%, 20.82%, 16.41%, 12.45%, 15.96%, and 8.67%, respectively. The optimized extraction process parameters are water extraction 3 times, each time for 1 hour. The network prediction results of BP-ANN samples are consistent with the results of the orthogonal experiment, and the mean square error (MSE) of network prediction value and actual measurement value is < 1%. The stable and feasible water extraction process of Qi-Zhi-Fang obtained through related mathematical model analysis and prediction significantly increases the extraction rate of effective ingredients of astragaloside and saikosaponin, the average comprehensive score of the verification experiment is 90.85 points, and the relative standard deviation (RSD) is 1.55%. Conclusion This study establishes the water extraction process of compound Qi-Zhi particles, and the optimal extraction process can effectively improve the extraction efficiency of astragaloside and saikosaponin, providing beneficial reference for the optimization of the preparation process and the establishment of quality standards for other clinical empirical formulas.

Qi-Zhi-Fang; Analytic Hierarchy Process (AHP); Indicator Relevance Indicator Weighting Method (CRITIC); Orthogonal Experiment; Grey Correlation Analysis; Backward Propagation Artificial Neural Network (BP-ANN); Extraction Process