The Whole Production Process Optimization Methodology Of Tcm Research

The quality of traditional Chinese medicine (TCM) products is formed during their pharmaceutical production process, each unit of which has more or less impact on the final product. Along with the development and wide application of modern information technology, analytical technology, sensor and control technology, a number of advanced pharmaceutical innovations has been applied in the manufacturing process of TCM products. Some enterprises built and implemented the automatic and integrated control systems. All these works have greatly increased the level and efficiency of TCM production process, as well as the inner quality of the products.However, with the good development trend of TCM industry, there still exist some quality related and key problems that have been not solved properly. Firstly, variations from the raw herbal materials and the manufacturing process could propagate during the production, leading to the variability of the intermediate and the product. The quality control of Chinese medicine often emphasizes on the analytical aspect. Conventional analytical techniques are able to capture the phenomenon of process variation, but are insufficient to maintain the quality consistence across herbal products. Secondly, the optimization technologies, including the unit and system optimization, don’t match up with the automatic and integrated production systems.In order to solve the above mentioned problems, the new concept and methodology named overall process optimization (OPO) of the TCM production is proposed in this thesis. The OPO methodology inherits the guidance from the TCM theory and system science theory. The most important character of OPO methodology is that it considers the transfer of products’quality from the overall production process point of view. The advanced information technology, analytical technology, pharmaceutical and engineering technology will be adapted and integrated in the implementation of OPO methodology. The main contents of OPO are to research the overall process modeling and simulation methods, the overall process optimization and control methods, and the risk evaluation and control methods. The goals of OPO are to realize the safe, smooth and optimized manufacture of the TCM product, to assure the predefined target of the TCM production, to stabilize the quality of the final products, and to accomplish the unity of "process control, process optimization, quality control, safety and efficacy". It is hoped that the OPO methodology will lay the foundation for the quality control of the overall process of TCM production. Around the content and aim of OPO. four parts are organized in this thesis. 1. Preliminarily theoretical study of overall process optimization of TCM productionThe study of overall process optimization of TCM production belongs to the multidisciplinary research area. Therefore, a clear interpretation of the OPO theory will help to guide the production practice. In this part of the thesis, the relationship between the OPO theory and the TCM theory is analyzed, and the guiding role of TCM theory is established. The relationship between the OPO theory and the modern system science theory is also discussed, and how to apply the system science theory to the overall process optimization is investigated. The core theory of TCM pharmaceutics is investigated, and the position and function of OPO theory in the theory of TCM pharmaceutics is clarified. Based on the theoretical studies, the possible ways to realize the OPO methodology in actual production process as well as the research plan of this thesis are brought forward.2. Characteristic study of the unit process of TCM productionIn this part, the popular TCM production unit, i.e. alcohol precipitation, is taken as the research object. And with the help the near infrared (NIR) spectroscopy, a process analytical technology (PAT) tool, the character of TCM production unit is analyzed from both the quantitative and qualitative aspects. The process NIR spectra could reflect the whole feature and fingerprint of the pharmaceutical process. In the quantitative analysis, the PLS model is utilized to extract the response variable related information from the process NIR spectra. In qualitative analysis, the PCA algorithm and multivariate statistical process control (MSPC) model is used to extract the main pattern from the process NIR spectra. Results revealed that there was batch-to-batch variation in the TCM production, and such variation could be due to the process uncertainty. The quantitative and qualitative models need to be updated to cope with the process variations. The PAT methods are helpful to observe the phenomena of batch-to-batch variations, but deserve the limited competence of control within the production unit.3. Study of the quality transferring law of TCM productionBy applying the concept of ICH’s Quality by Design (QbD), the establishment of design space for the TCM production process is proposed. Design space stands for a higher level and deep understanding of the TCM production process, and its development is based on process models. In this part, four kinds of data-driven approximation models, i.e. polynomial model, least square support vector machine (LS-SVM), radical basis function neural network (RBFNN) and Kriging model, are used to build surrogates for the water extraction and alcohol precipitation process of gardenia fruit. The hyperparameters of LS-SVM, RBFNN and Kriging models are optimized by a combination of genetic algorithm (GA) and adaptive modeling technique. Based on the process models, the relationships between the critical process parameters and critical quality attributes of the process unit are investigated by the response surface methodology, uncertainty analysis and sensitivity analysis.With the established process unit models and the help from the system science theory, a "partition-integration" modeling method is used to develop the multi-unit combinational model for the "extraction-concentration-alcohol precipitation" processes of gardenia fruit. Based on the multi-unit combinational model, the Monte Carlo simulation (MCS) is applied to approximate the multi-unit combinational design space, which is visualized by the parallel coordinate plot. Through the multi-unit combinational design space, the quality transfer laws of Geniposide and Genipin-1-β-D-gentiobioside are studied.4. Study of the strategy for the overall process optimization of TCM productionThis part presents a new strategy, target-oriented overall process optimization (TOPO), which can be used to assure the consistent quality in TCM products. The methodology of TOPO includes four parts, target definition, data pretreatment, process modeling and overall process optimization. The Bayesian approach is integrated into the optimization step. The mechanism of TOPO involves optimizing multiple units of the production system step by step, giving each unit optimal operating conditions consistent with the quality target. The probability trajectory is adjusted to monitor and optimize the production process. The proposed TOPO strategy was successfully applied to seven-unit manufacturing process used to produce Lonicerae Japonicae extract and six-unit production process of gardenia extract. Results demonstrated that TOPO could keep the production process in line with the predefined target and reduce the variability of the final products. In general, TOPO explores the full potential of legacy batch production data with respect to understanding and optimizing the herbal production process, and provides systematic ways to the overall process optimization of the TCM production.All in all, this thesis integrates the process unit research and multi-unit research, the laboratory scale research and production scale research, the theoretical study and practical application together. The proposed OPO methodologies and technologies for TCM production will help to enhance the level of quality control of TCM products as well as core competiveness of TCM industry.