In-line Monitoring Research Of API Content In Fluidized Bed Mixing Process By Near Infrared Spectroscopy

With the promulgation of industrial guidelines,near infrared spectroscopy(NIRS)monitoring the uniformity of solid oral preparations has become one of the most reported studies.In the fluidized bed mixing process,the content uniformity of Active Pharmaceutical Ingredients(API)is a critical quality attribute(CQA).However,fluidized bed mixing has always been carried out under strict process parameters.The entire process seems like a black box,making it difficult to understand the true state and physical and chemical properties of the materials in the chamber.In this study,a portable near-infrared sensor was introduced to perform "visual" in-line monitoring of the fluidized mixing process.As a fast,non-destructive,green and environmentally friendly process analytical technology(PAT),NIRS could achieve the purpose of in-line monitoring of API content during the mixing process.Due to the dynamic characteristics of materials and the interference of other influencing factors,the in-line model has poor predictive ability.The collection of in-line spectra requires more materials than off-line spectra,and regular maintenance of the model requires more economic investment.In order to improve the accuracy of in-line models,save costs,reduce the economic costs of calibration set spectrum collection and model maintenance,in this chapter,chemometrics is used to study the transfer of models between offline and online spectra.The model is established after transfer.The new model is used for in-line monitoring of the API content in the pilot production process,realizing the theoretical and practical research from laboratory to pilot application.The following is the main research content:(1)Feasibility study of NIRS in-line monitoring API content in fluidized bed mixing processA portable miniature near-infrared sensor is used to collect off-line spectra during the small-scale experimental fluidized bed mixing process.Limited materials are used to establish a calibration model.Through the experimental design,the influence of formula factors on the model results during the mixing process was investigated.In addition,by studying different preprocessing and band selection methods to optimize the model,the R2,Rev2,RMSEC,RMSECV,Rp2 and RMSEP of the model were 0.977,0.973,1.6776,1.8588,0.981 and 1.5296 respectively.The methodological investigation of the near-infrared monitoring method shows that the near-infrared measurement has good accuracy(RMSEP=1.5296%),precision(SD=2.2%)and linearity(R=0.921),indicating that the feasibility of near-infrared measurement monitoring the API content.(2)In-line monitoring of API content in fluidized bed mixing process by NIRSIn this study,an experimental fluidized bed was used and 64 batches of experiments were conducted under different environmental conditions and process parameters.The corresponding near infrared spectrum is collected inline.Then,the spectra selection method of the angle cosine distance is proposed and the spectrum selection is performed on the in-line spectrum.Finally,the quantitative model of API content in the mixing process is established.After spectral preprocessing and variable selection,the Rc2,Rcv2,RMSEC and RMSECV of the model are 0.957,0.954,2.3103 and 2.3143,respectively.The methodological investigation of the NIRS method showed that the in-line near-infrared measurement had good accuracy(RMSEP=2.3629%),precision(SD=4.8%)and linearity(R=0.903),which verified the feasibility of in-line monitoring of API content in the fluidized bed mixing process.This research can be used as a reference for the in-line monitoring by miniature near-infrared sensors in the pilot and industrial fluidized bed mixing process,to achieve in-line monitoring of the fluidized bed mixing process,and to connect CQA with key process parameters to achieve the intelligent production of solid preparation.(3)Research on NIRS model transfer and pilot application in fluidized bed mixing processThe piecewise reverse standardization(PRS)algorithm is used to study the model transfer between off-line and in-line spectra.After model transfer,the modeling results of the in-line model are improved.Rc2,Rcv29 RMSEC and RMSECV are 0.960,0.950,2.2368 and 2.5009,respectively.The model established after the model transfer is used for in-line monitoring,and the near-infrared real-time monitoring value of API content and the real value are well fitted overall.The results show that this portable near-infrared sensor can be used for on-line API content monitoring in the pilot fluidized bed.Moreover,through model transfer,the modeling cost is reduced and the accuracy of the model is improved.The innovations of this study include:(1)In the fluidized bed mixing system,the experimental design was used to investigate the influence of the raw and auxiliary materials variation factors,process factors and environmental factors on the modeling results.The equivalence of the NIRS model and off-line measurement was verified from the perspective of spectrum monitoring,which provided method support for real-time monitoring of API content in the fluidized bed mixing process.At the same time,in the case of the need to detect disturbances and changes in special reasons from the downstream of the pharmaceutical process,the application of spectrum-based PAT has shown great application prospects.(2)The cosine distance spectra selection method was proposed and used.The spectral selection of the in-line spectrum collected during the mixing process improved the predictive ability of the model,and the method was universal.(3)The PRS model transfer method was used to study the model transfer between off-line and in-line spectra.Through model transfer,the modeling cost was reduced and the accuracy of the model was improved.