Process Characterization & Statistical Modeling — Transforming Data Into Knowledge Throughout The Product Lifecycle

It’s no surprise that data plays a pivotal role in the biopharmaceutical industry. Parsing through the large amount of valuable data acquired during a product lifecycle and turning it into actionable knowledge is a critical task for drug development pre- and post-regulatory approval. At Cytovance Biologics, process characterization not only characterizes activities in the process validation landscape, but also provides a platform leveraging enablers that turn data into product and process knowledge. This ensures desired product quality and process consistency start to finish.

In this webinar, the featured speaker describes their process characterization services based on Quality by Design (QbD) principles, emphasizing risk and knowledge management and transforming data into the currency of product and process knowledge.

The speaker covers the applications of:

• PC enablers, including a powerful data management platform — OSIsoft PI
• An effective statistical tool DoE (design of experiments)
• A machine learning technique known as SVEM (self-validating ensemble modeling) and
• A JMP Pro multi-purpose platform functional data explorer (FDE)

Through the integrated applications of DoE with SVEM and OSIsoft PI with FDE, statistical modeling transforms DoE and functional data, respectively, into actionable knowledge. Although there are many types of data, the webinar will focus primarily on DoE and functional data analysis.

The speaker will demonstrate the benefits of employing a definitive screening design, a super-saturated design that features fewer experimental runs in the JMP DoE platform and SVEM, which builds high-performance predictive models through auto-validation techniques. These predictive models can be constructed to support the efficient development of analytical procedures and drug processes. Moreover, much of the process data is generated as a time function. Therefore, different tools and techniques are required to handle the complexity of functional data.

This webinar highlights how turning data into knowledge can enhance process understanding. In addition, it explores how predictive models can be constructed using DoE data and validated in parallel via SVEM algorithms. This approach reduces the number of experimental runs required for a validated model. By integrating advanced enablers in the workflow like sophisticated statistical modeling techniques and real-time data collection, many of the challenges encountered throughout a product lifecycle can be overcome.