Multivariate process analytical technology tools for fluidized bed granulation and drying analysis: A review

The United States Food and Drug Administration (FDA) has encouraged the adoption of Process Analytical Technology (PAT) in the pharmaceutical industry for improving manufacturing processes and product quality. Multivariate PAT tools are particularly useful for analysing vast amounts of data collected through process analysers. Univariate analysis methods are inadequate for studying these datasets due to the collinearity between variables and the large number of variables in combination with a small number of measurements. This review focuses on the use of multivariate methods for analysing data generated by PAT analysers that monitor fluidized bed granulation and drying processes. A brief theoretical and critical overview of multivariate methods is presented, with a focus on the most commonly used techniques. The first part of the review focuses on the use of principal component analysis (PCA), partial least squares (PLS) regression and multiple linear regression (MLR). The methods for analysis of multiway data, such as multiway PCA (MPCA) and multiway PLS (MPLS), are presented in the second part. The final part of the review explores less commonly used techniques that show promise to become valuable multivariate PAT tools in the future (cluster analysis, discriminant analysis, artificial neural networks (ANN), support vector machine (SVM) and random forests).