Multivariate Analysis for Process Understanding, Continuous Process Verification, and Condition Monitoring of Lyophilization Processes

This chapter explores the use of multivariate data analysis (MVDA) tools as applied in the context of lyophilization. Since MVDA offers the advantage of being capable to handle large datasets, allows rapid visualization of process trends, enables automatic fault/anomaly detection of a batch, and takes into account the correlation in between variables, these mathematical tools are well suited to applications such as the continuous process verification of lyophilization batches or the condition monitoring of the equipment. To achieve this objective, a basic theoretical overview of MVDA tools will be presented as well as the knowledge required to understand the mathematical tools and language required in building principal component analysis (PCA) and projection to latent structure (PLS) models. Then examples will be given when applied to lyophilization processes.