Summary: | The effect of uncertainty in biopharmaceutical manufacturing can be a barrier to robust, scalable process design. The ideal is for a process in development to complete technology transfer to full scale manufacturing with no redevelopment costs or surprises. Essential to achieving this is a systematic method for analysing large complex datasets and extracting critical combinations of fluctuations that lead to product loss and scheduling delays. This thesis describes a dynamic database-driven decision-support tool to facilitate such efforts and identify robust optimal purification strategies to match the high productivity cell cultures whilst coping with uncertainties. The benefits of a databasedriven approach using MySQL (MySQL AB, Uppsala, Sweden) are harnessed to capture the process, business and risk features of multiple biopharmaceutical purification sequences in a multi-product facility and better manage the large datasets required for multiple processes, uncertainty analysis and optimisation. Principal component analysis combined with clustering algorithms are used to analyse the complex datasets from complete batch processes for biopharmaceuticals. The challenge of visualising the multidimensional nature of the dataset was addressed using hierarchical and k-means clustering as well as parallel co-ordinate plots to help identify process fingerprints and characteristics of clusters leading to facility fit issues. Industrially-relevant case studies are presented that focus on tech transfer challenges for therapeutic antibodies moving from early phase to late phase clinical trials.
|