| Traditional Chinese medicine granule is one of the main forms of modern traditional Chinese medicine preparation,which is a dry granule preparation with a certain particle size prepared by mixing raw materials and appropriate excipients.Compared with ordinary traditional Chinese medicine granulation methods,dry granulation does not require processes such as soft material making,extrusion granulation,drying and granulation.It has the advantages of simple production process,high production efficiency,and less auxiliary materials.It is widely used in Chinese medicine solid preparations.In the process of dry granulation,the powder properties of the Chinese medicine extract and the setting of process parameters are closely related to the quality of the granules.Under the existing production model of"experience and trial and error",it will not only cause a waste of time and raw materials,but also difficult to ensure the consistency of the quality of granules.Therefore,it is of great significance to study the correlation between the powder properties of traditional Chinese medicine extract,process parameters and the quality of granules building a prediction model for the quality control of granules.Quality by Design(Qb D)aims to transition the drug quality supervision system from relying solely on final product inspection to quality control in the production process,product design and research phases,which can comprehensively improve the stability of product quality and the adjustability of production processes.Traditional Chinese medicine extracts are composed of different types of components,which have the characteristics of diverse chemical components and complex composition system.Among them,flavonoids,saponins and polysaccharides are important effective components of traditional Chinese medicine.The natural properties of different components of traditional Chinese medicine raw materials determine the differences in their properties,which will be transmitted to the quality attributes of the final product through the process,leading to the change of particle quality.Based on the concept of QBD,this project takes the extracts of Scutellaria baicalensis Georgi,Anemarrhena asphodeloides and Astragalus membranaceus as the model drugs of flavonoids,saponins and polysaccharides,constructs the design space,selects the control space,and establishes the prediction model,so as to move the product quality control center forward to the raw material control and pharmaceutical process control and save time and excipients,ensure the quality of traditional Chinese medicine.The quality control center moved forward to raw material control and pharmaceutical process control in order to save time and auxiliary materials,while provids reference for ensuring the uniformity and stability of the quality of Chinese medicine granules.In summary,this article mainly includes the following five parts:The first part was the Bayesian network fault diagnosis of traditional Chinese medicine dry granulation based on FMEA.FMEA was used to identify the main risk factors affecting the quality of dry granulation of traditional Chinese medicine,and the CPPs of dry granulation of traditional Chinese medicine,namely feeding speed,roller speed,and roller pressure.Based on FMEA,Bayesian fault diagnosis model of dry granulation of traditional Chinese medicine was constructed.The occurrence of failure modes and BN’s powerful mathematical logic inference calculation function solved the probability of occurrence of failure modes.The bad state value of"unqualified particle quality"was set to 100%,and then the probability parameters of the whole network were updated.Under the condition of unqualified particle quality,the bad failure probability(bad)of"low particle forming rate","poor particle solubility"and"poor particle fragility"were all more than 50%.Therefore,the particle forming rate,brittleness and dissolution time were taken as the CQAs of the particles to study the influence of process and prescription factors on the dry granulation process of traditional Chinese medicine in the follow-up experiments.The second part was based on the Qb D concept of flavonoid component Chinese medicine(Scutellaria baicalensis Georgi extract)powder properties-process parameters-particle quality correlation research and prediction model construction.Using Box-Behnken experimental design was used to establish model of key process units,and constructed the design space,from which the control space was selected,and its accuracy was verified by Monte Carlo method.The powder samples of flavonoids extract were prepared by mixture design,and the dry granules were prepared by selecting the specific process parameters in the space.Taking the above sample data as the model data,the model data with strong correlation were extracted by principal component analysis(PCA),and the reobtained variables and other input variables(including process parameters and sheet thickness)were used as the model data Radial basis function neural network(RBFNN),generalized regression neural network(GRNN)and back propagation neural network(BPNN)were used to construct the input layer of the prediction model.The correlation of powder properties,process parameters and dry particle quality was studied,and the prediction accuracy of different models was verified.The results showed that the feeding speed was between 5Hz and 11Hz,and roller speed was between15Hz and 27Hz,and roller pressure was between 0kg/cm~2and37kg/cm~2,and the accuracy of particle forming rate and melting time reach 100%and 99.4%respectively.The MRE of PCA-RBFNN prediction model constructed of molding rate,melting time and friability were all in the range of 6%,which was reduced to different degrees compared with PCA-BPNN and PCA-GRNN.It showed that the PCA-RBFNN model constructed in this chapter had relatively good prediction accuracy for the quality of flavonoid extract particles,and could predict the CQAs of flavonoid components of traditional Chinese medicine dry particles more accurately.The third part studied on the correlation between the powder properties,process parameters and particle quality of saponin components of traditional Chinese medicine(Anemarrhena asphodeloides extract)and the construction of prediction model based on the concept of Qb D.The Box-Behnken experimental design was used to prepare dry granulation.The control space of dry granulation of saponin component extract was determined on the basis of previous experiments,which was verified by Monte Carlo method.The saponin component extract powder sample was prepared by the mixture design,and the specific process parameter values were selected in this space to prepare the dry method Granules,obtaining corresponding data samples of dry granules under the premise of prescription changes.Based on the above sample data as the model data,PCA method was used to process the input indexes with strong correlation in the model,and the RBFNN model of saponin component extract powder properties process parameters particle quality prediction was constructed and verified to study the correlation among the three.The results showed that the feeding speed was between 10Hz and 26Hz,and roller speed was between10Hz and 18Hz,and roller pressure was between 10kg/cm~2and 50kg/cm~2,and the accuracy of particle forming rate and melting time reaches 100%.The MRE of the PCA-RBFNN prediction models for the molding rate,dissolution time and brittleness were all within the range of 6%,indicating that the PCA-RBFNN model constructed in this chapter has relatively good prediction accuracy for the quality of saponin component extract granules,and could accurately predict the CQAs of saponin component traditional Chinese medicine dry process granules.The fourth part was based on the Qb D concept of polysaccharide component Chinese medicine(Astragalus extract)powder properties-process parameters-particle quality correlation research and prediction model construction.The Box-Behnken experimental design was used to prepare dry-process particles,and the correlation between process parameters and the quality of dry-process particles was quantitatively investigated through this experimental design which was verified by Monte Carlo method.The mixture design was used to prepare the polysaccharide component extract powder sample,and the specific process parameter values were selected in this space to prepare the dry method granules,and obtained corresponding data samples of dry granules under the premise of prescription changes.Based on the above sample data,PCA method was used to deal with the input indexes with strong correlation in the model,and the RBFNN prediction model of polysaccharide extract powder properties process parameters particle quality was constructed and verified to study the correlation among the three.The results showed that the feeding speed was between 10Hz and37Hz,and roller speed was between 7Hz and 19Hz,and roller pressure was between 0kg/cm~2and 50kg/cm~2.The accuracy of particle forming rate and melting time was 95%and 94%respectively.The MRE of the PCA-RBFNN prediction model for molding rate,melting time and brittleness were all in the range of 6%,indicating that the PCA-RBFNN model constructed in this chapter had relatively good network prediction accuracy for the quality of polysaccharide extract granules,and could accurately predict the CQAs of polysaccharide extract granules.The fifth part was the construction of the prediction model of powder properties-process parameters-particle quality of compound Chinese medicine extracts.Mixture design was used to simulate the preparation of compound Chinese medicines with different physical properties.Three components of flavonoids,saponins and polysaccharides were selected to control the overlapping part of the dry granulation process parameters of traditional Chinese medicines,and specific process parameters were selected to prepare dry granules.PCA method was used to deal with the input indexes with strong correlation in the model,and the RBFNN model was established to predict the powder properties,process parameters and particle quality of traditional Chinese medicine extracts.The results showed that the MRE of the PCA-RBFNN model was within 7%,which could accurately predict the quality of compound components of traditional Chinese medicine granules.In summary,this article is based on the study of the correlation between the properties of flavonoids,saponins and polysaccharides of three different physical properties of traditional Chinese medicine powder properties-process parameters-dry particle quality,and the prediction results of the constructed model can be used to determine the formability and solubility of particles.The actual results of mechanical strength provide a reference,and the application of Qb D concept has a certain reference significance for the formulation and adjustment of process parameters to determine the best granule quality,and provides a new idea for the establishment of dry granule quality control methods for component traditional Chinese medicine. |
