中药近红外定量模型跨尺度传递方法研究

As a process analytical technology，NIR spectroscopy is widely used in the pharmaceutical field. However，NIR quantitative models need to be constantly updated and transferred in the preparation process of Chinese patent drugs because of the complex background variation in TCM system. The aim of this project is establishment of cross-scale transfer methods from laboratory NIR quantitative models to semi-works models using the orthogonal projection technique and the model cross-scale updating technique after systematic research on the influence factors and transfer errors of NIR quantitative model transfer，which will realize the pattern of the research results transfer from lab to semi-works or from semi-works to production and guide the rapid applications of the laboratory research findings in TCM fields. Taking the water extraction process of Flos Lonicerae Japonicae and the drying process of Centalla Total Glucosides as the research carriers, the method for NIR quantitative model cross-scale transfer is researched. After the effects of temperature and raw material variation on the model performance are clarified，the laws for process variation and model transfer will be studied，then the systematic error of spectral background between the laboratory samples and semi-works samples will be corrected through the orthogonal projection technique，and the accidental error among the semi-works samples spectra will be diminished through model-updating algorithms. The method system of model transfer about “variant sample recognition—error correction—model updating” will be constructed，through which the cross-scale transfer of NIR quantitative model will be realized，and which will strengthen the ability of semi-works model to cope with the complex background of herbal medicine and promote the application of NIR quantitative model in the preparation process of TCM.

近红外光谱作为过程分析技术，被广泛应用于制药领域。但是在中药制剂过程中，由于中药体系复杂背景的变动，近红外定量模型需要不断地更新和传递。本项目拟在实验室近红外定量模型的基础上，对影响模型传递的因素及产生的误差开展系统研究。利用正交投影和模型跨尺度更新技术，建立近红外定量模型由实验室向中试跨尺度传递的方法，实现中药的研究结果由实验室向中试乃至由中试向生产转移的模式，以指导中药实验室研究成果的快速应用。因此，本项目以金银花水提和积雪草总苷干燥过程为载体，在明确温度、原料变异等因素对模型性能的影响后，进一步研究过程变异和模型传递规律，采用正交投影技术“校正”中试样本与实验室样本间的系统误差，应用模型更新算法“包容”中试样本间的偶然误差，构建“变异样本识别—误差校正—模型更新”的模型传递体系，建立近红外定量模型的跨尺度传递方法，增强模型应对中药背景变动的能力，促进近红外定量模型在中药生产过程的应用

中药作为天然产物，其制剂原料内在成分受地域、气候等多方面因素的影响而产生变动，同时中药制剂过程中不可控的制剂条件的波动，不同批次间原料药制剂过程的NIR光谱背景复杂，一定范围内样本所建立的NIR定量模型具有各自的适用范围。在中药制剂过程中，由于中药体系复杂背景的变动，近红外定量模型需要不断地更新和传递。本课题组以金银花水提过程和积雪草总苷制粒过程为载体，分别在小试规模下和中试规模下设计了来源不同的金银花和药用辅料糊精来模拟不同批次原料制剂过程中的NIR光谱背景差异，在制剂过程中采集NIR光谱的同时收集样本，分别以金银花水提过程绿原酸浓度和积雪草总苷制粒干燥过程水分含量为参考值，建立两个制剂过程的NIR定量模型，讨论了不同批次和不同制剂规模的NIR定量模型的适用性。在金银花水提的液体体系和积雪草总苷制粒的固体体系两个制剂过程，当NIR定量模型对新来源批次样本和不同制剂规模的样本预测失效时，以新批次和同制剂规模的制剂过程的近红外光谱定量预测作为NIR定量模型传递的目标，建立了“DOSC-SBC”不同批次样本间同尺度模型传递方法和“光谱背景校正与以中试变异样本为指导的模型更新相结合” （DOSC-SBC-SIC-MU）小试样本向中试样本跨尺度模型传递方法，为中药近红外定量模型的传递提供参考方法，促进近红外定量模型在中药制剂过程中的应用。