串联型中药制药过程质量传递模型和全局优化方法研究

Modeling of the pharmaceutical process for quality control is an important way to enhance the precision and stability of the manufacturing process of Chinese medicine preparations. The current practice of process modeling and optimization of Chinese medicine products is limited to the production unit, while ignoring the cumulative effect and time varying effect of the production system. In order to overcome such limitations, the overall process modeling and optimization methods of Chinese medicine products are studied in this work, with the serial manufacturing process of Qingkailing Injection and Ginkgo Leaf Tablet taken as the research objects. The multi-source dispersive quality attributes and process parameters from the production system are integrated and preprocessed first. Then, the multi-block partial least squares (MB-PLS) algorithm and the expanding modeling technique are used to establish the quality transferring model of the serial manufacturing process of Chinese medicine products. The multidimensional combination and connection of production units and process parameters will be evaluated. The cumulative and time varying effects of the production system will be investigated. Then contribution of both the production units and process parameters can be quantified. Based on the established expanding MB-PLS models, the suitable process control strategies will be set. The reliability-based overall process optimization methods will be researched, considering the constraints within both the primary variable space and the latent variable space of the process model. Optimization of the serial manufacturing process will be carried out unit by unit, guaranteeing the stability and traceability of the process quality trajectory. The methodology established in this research project is ultimately aiming at lifting the quality control level of the manufacturing process of Chinese medicine products.

过程质量控制模型化是提高中药制剂工艺稳定性和精密度的重要方法和途径，现有的中药制药过程建模和优化仅限于单元工艺，未体现系统质量累积效应和时变效应。针对此局限性，本项目立足于中药生产全过程，以清开灵注射液和银杏叶片串联制备工艺为研究对象，将过程系统中多源分散的质量参数和过程参数数据融合利用，采用多块偏最小二乘（MB-PLS）算法和递进建模策略建立串联型中药制药过程质量传递模型，评价工艺单元及过程变量之间的多维交互作用，考察中药制药过程系统的质量累积和时变效应，量化各工艺单元和变量对质量传递的影响，制定合理的工艺控制策略。在递进MB-PLS模型的基础上探索以可靠性为导向的全局优化方法，同时考虑高维工艺参数在原始变量空间和潜变量空间内的约束，结合最优化算法实现串联型中药制药过程逐步优化，保证中药生产过程质量轨迹稳定且可追溯，提高中药生产过程质量控制水平。

国际制药质量源于设计理念将药品质量控制模式前移，即从传统的以检验为主的制药过程到以科学和风险为主并经过详细研究和理解的生产过程。中药制药过程具有原料复杂、流程长、控制对象多等特点，借鉴QbD的方法和原则深入理解中药制药过程特性，开发符合中药特点的生产过程质量控制和优化方法，对于确保产品质量在各工艺环节间的准确传递，并从根本上实现中药产品“安全有效、稳定均一”具有重要的科学意义和应用价值。本项目立足于中药生产全过程，突破单元界限，从工艺系统角度研究了中药制药过程质量传递模型。以液体制剂提取、浓缩、醇沉、收醇、精制串联型工艺和固体制剂混合、制粒、干燥、压片等共性关键制剂环节为研究对象，研究了过程近红外光谱的适宜性评价和应用，建立了中药原料粉末、中间体颗粒物理指纹谱评价方法，融合各单元异质、分散数据，建立了中药串联型制药过程多块偏最小二乘（MB-PLS）质量传递模型。基于模型全面评价了中药制药过程“关键原料属性-关键工艺参数-产品关键质量属性”之间的多维组合和相互作用，从关键工艺参数辨识、关键工艺单元辨识、以及关键工艺参数对产品质量属性的系统影响等方面对中药制药过程的质量传递规律进行了解析。建立了达标概率、Hotelling T2和预测平方误差（SPE）集成的优化目标，同时考虑待优化过程参数在原始变量空间和潜变量空间内的约束，通过模拟过程扰动和实施多阶段序贯前馈优化评价了全局优化方法在稳定中药产品质量中的作用，实现了以可靠性为导向的串联型中药制药过程系统全局优化，为提高产品质量一致性注入可靠性元素。本项目第一标注发表论文25篇，其中SCI收录6篇、EI收录1篇、中文核心期刊论文16篇、一般期刊论文1篇、会议论文1篇；培养硕士研究生1名，协助培养博士研究生2名；相关成果得到《中国中药杂志》（2017年第6期）专题报道。