Multivariate Analysis In The Pharmaceutical Industry PDF Download

Manufacturing of pharmaceutical products has a prominent role in the healthcare industry. Generally, the ultimate aim of pharmaceutical development is to release to the market products with acceptable quality. As advanced pharmaceutical manufacturing technologies such as continuous tablet manufacturing, are developed and embraced, it is essential to adopt a scientific, risk-based, and proactive approach for pharmaceutical development. The work presented in this dissertation focuses on using multivariate analysis tools to establish a predictive capability for pharmaceutical process and product development, especially when the amount of materials available is limited. Importantly, the methodologies developed in this dissertation can be applied easily to powder handling and processing in a wider range of industries, such as cosmetic, catalyst, chemical, petrochemical, and food. In this work, methods for analyzing flow properties of raw materials and predict process performance were developed. A method to analyze shear cell data of powders measured under different initial consolidation stresses was introduced. The method was shown to reduce significantly the complexity of shear cell data, and to enabled comparison of materials measured under different initial consolidation stresses. In addition, a predictive correlation between material flow properties and feeder performance was developed. By using multivariate models, the feeding performance of a material with given flow properties can be predicted and quantified. Using a quality-by-design approach, the cohesion of a powder mixture can be predicted based on the concentration of each ingredient. The prediction model was further supplemented by a study investigating two mixing systems. Using statistical analysis, the effect of lubrication on blend flow properties was discussed. By quantifying the correlations between different flow property measurements, mixing systems that have different mixing mechanism were compared. Disadvantages of widely used dissolution comparison methods were addressed. Statistically reliable methodologies to analyze, compare, and predict drug in vitro release profiles were proposed. The proposed methods were shown to be able to consider the self-correlated intrinsic nature of dissolution profiles, and to use within-group variability to estimate the reliability of observations. Additionally, the work presented a case study to improve real-time release testing for advanced tablet manufacturing processes by achieving predictive capability for nondestructive dissolution testing. Using hierarchical multivariate analysis, the validated prediction models were able to predict dissolution profile of an individual tablet based on its NIR spectrum.