近红外光谱结合化学计量学快速测定蓝芩口服液原药材水分含量

目的 通过近红外光谱法建立蓝芩口服液板蓝根、栀子、黄芩、黄柏、胖大海原药材中水分的快速定量方法，同时研究了水分通用模型的可行性。方法 采用了多种预处理方法进行模型优化，选用竞争自适应重加权采样法（competitive adaptive reweighted sampling，CARS）进行关键变量筛选，建立了5种药材的专属、通用偏最小二乘回归（partial least square regression，PLSR）模型。结果 5个原药材的专属、通用CARS-PLS模型Rc值>0.96，RSEP值<5%。与专属模型相比，通用模型的预测准确度稍有下降，但仍满足应用要求。此外，通过配对t检验验证模型预测能力，6个PLSR模型预测值与HPLC测得的参考值皆不存在显著性差异。结论 近红外光谱与化学计量学相结合建立通用模型是一种可靠的方法，可用于蓝芩口服液5种原药材的水分含量快速检测。对于水分含量检测，选用药材的通用模型比专属模型更为简便，更符合高效化的生产需求。

Rapid Determination of the Moisture Content of the Original Medicinal Materials of Lanqin Oral Solution by Near-infrared Spectroscopy Coupled with Chemometric Algorithms

OBJECTIVE To establish a rapid quantification method of moisture contents in the raw materials of Lanqin oral solution, Isatidis Radix, Gardeniae Fructus, Phellodendri Chinensis Cortex, Scutellariae Radix and Sterculiae Lychnophorae Semen by near-infrared spectroscopy(NIRs), and the feasibility of an universal moisture content model was investigated. METHODS Preprocessing methods were applied to optimize the model, and competitive adaptive reweighted sampling (CARS) was chosen for key variable screening. Exclusive and universal partial least square regression(PLSR) models for five medicinal materials were constructed. RESULTS The Rc values of CARS-PLS models were higher than 0.96, and the RSEP values were lower than 5%. Compared with the exclusive model, the prediction accuracy of the universal model was slightly lower, but it still met the application requirements. In addition, the predictive ability of the model was verified by paired t-test, and there was no significant difference between the predicted values and the reference values. CONCLUSION The overall results showed that the combination of NIRs and chemometrics to establish an universal model was a reliable method, which can be applied for rapid detection of the moisture content of the five medicinal materials of Lanqin oral solution. For moisture content detection, the universal model of medicinal materials is simpler than the exclusive model, and it is more in line with high-efficiency production requirements.