Setting the process parameters for the coating process in order to assure tablet appearance based on multivariate analysis of prior data.

Designing efficient, robust process parameters in drug product manufacturing is important to assure a drug's critical quality attributes. In this research, an efficient, novel procedure for a coating process parameter setting was developed, which establishes a prediction model for setting suitable input process parameters by utilizing prior manufacturing knowledge for partial least squares regression (PLSR). In the proposed procedure, target values or ranges of the output parameters are first determined, including tablet moisture content, spray mist condition, and mechanical stress on tablets. Following the preparation of predictive models relating input process parameters to corresponding output parameters, optimal input process parameters are determined using these models so that the output parameters hold within the target ranges. In predicting the exhaust air temperature output parameter, which reflects the tablets' moisture content, PLSR was employed based on prior measured data (such as batch records of other products rather than design of experiments), leading to minimal new experiments. The PLSR model was revealed to be more accurate at predicting the exhaust air temperature than a conventional semi-empirical thermodynamic model. A commercial scale verification demonstrated that the proposed process parameter setting procedure enabled assurance of the quality of tablet appearance without any trial-and-error experiments.