Advancement of cell culture process understanding and control through real-time multivariate process monitoring, use of statistical process modes and deployment of process analytical technologies

Commercial bioprocess manufacturing facilities generate considerable amounts of process data which are routinely used for process monitoring. Through the analysis of batch over batch trends it is possible to ensure that the process remains in a state of continuous statistical control. Real time process data can also be used to drive rapid root cause analysis and to enable timely interventions to prevent performance parameter excursions. Finally, these process data also provide significant opportunities to further understand the interaction between operating parameters, process equipment and raw materials to enable improved process control and optimization. To use the available process data effectively, systems are required to store, aggregate, and visualize the data. These data can then be used for advanced analytics, such as development of process models or analysis of variance across similar product modalities. These advanced analytics allow for rapid identification of process excursions and identification of true special cause variation (e.g. raw material variability). These tools also enable continued process verification (CPV) as part of the lifecycle approach to process validation. This presentation will describe the deployment of advanced process monitoring approaches, the development of predictive process models using multivariate process and product data and how to leverage these tools to drive process improvements. Finally, the application of newer technologies to understand and control process performance will be discussed including a case study describing the development of closed loop feeding strategies using Raman spectroscopy. These advances will be described in the context of continued process verification and as part of a future vision for portfolio-wide approaches to process monitoring, understanding and optimization