目的 立足于药食同源产品高质量发展的要求，以小儿消食颗粒（Xiaoer Xiaoshi Granules，XXG）为示范，基于质量源于设计（quality by design，QbD）理念建立药食同源颗粒剂的智能驱动式成型工艺优化及质量一致性评价方法。方法 首先，使用层次分析法（analytic hierarchy process，AHP）对XXG的成型关键质量属性（critical quality attributes，CQAs）进行权重分配并通过风险评估法确定关键影响因素；其次，在成型关键影响因素的单因素考察后，通过比较Box-Behnken设计-响应面法（Box-Behnken design-response surface methodology，BBD-RSM）与遗传算法-反向传播神经网络法（genetic algorithm-backpropagation neural network，GA-BPNN）优选最佳成型参数组合；最后，根据XXG药食同源的属性特点，对其总黄酮含量的化学质量属性和基于电子舌的口感质量属性进行一致性评价研究。结果 确定赋形剂性质、润湿剂乙醇体积分数、润湿剂乙醇用量和辅药比为XXG成型关键影响因素，赋形剂种类及配比为麦芽糊精-甘露醇2∶1、乙醇体积分数84%、乙醇用量0.18 mL/g、辅药比1.5∶1为XXG的最佳成型参数组合，综合评分为103.88。在有效成分含量方面，3批颗粒的总黄酮质量分数依次为8.46、8.82、8.82 mg/g，RSD仅为1.95%。在电子舌智能感官评价方面，3批颗粒在电子舌7根传感器（SCS、ANS、CTS、NMS、AHS、PKS和CPS）上响应值的RSD均在3%以内。结论 优选出的XXG最佳成型参数组合稳定可行，该成型工艺的优化方法能为药食同源产品的开发、中药智能制造的发展提供参考价值。

Objective To meet the requirements for high-quality development of food-medicine homology products, taking Xiaoer Xiaoshi Granules (XXG, 小儿消食颗粒) as an example, this study established an intelligent-driven granulation process optimization and quality consistency evaluation method for food-medicine homology granules based on the quality by design (QbD) concept. Methods Firstly, the analytic hierarchy process (AHP) was used to assigned weights to critical quality attributes (CQAs) of XXG, with subsequent risk assessment identifying granulation critical influencing factors. Secondly, following single-factor screening of critical influencing factors, the optimal combination of granulation parameters was selected by comparing the Box-Behnken design-response surface methodology (BBD-RSM) and the genetic algorithm-backpropagation neural network (GA-BPNN). Finally, leveraging the medicine-food homology properties of XXG, a quality consistency assessment was performed targeting both chemical attributes (total flavonoid content) and taste quality attribute based on electronic tongue. Results The excipient properties, ethanol volume fraction of wetting agent, ethanol amount, and excipient-to-drug ratio were determined as critical influencing factors, the optimal granulation parameters for XXG were determined as follows: an excipient system of maltodextrin to mannitol at a 2∶1 ratio, an ethanol concentration of 84%, an ethanol amount of 0.18 mL/g, and an excipient-to-drug ratio of 1.5∶1, yielding a comprehensive score of 103.88. Regarding bioactive constituents, the RSD of total flavonoid content across three batches was as low as 1.95%. For instrumental sensory evaluation, the e-tongue response values demonstrated batch-to-batch consistency with RSDs below 3% across all seven sensors. Conclusion The optimized granulation parameter set demonstrated stability and feasibility, and this process optimization approach can offer valuable insights for developing medicine-food homology products and advancing intelligent manufacturing in traditional Chinese medicine.

R283.6

京津冀科技创新协同项目(Z251100005225010);国家重点研发计划项目(2022YFC3501902);国家重点研发计划项目(2023YFC3504505);国家自然科学基金项目(82274110);北京科技新星交叉项目(20230484458)