Multivariate data analysis on historical IPV production data for better process understanding and future...

Area of Science:

• Biopharmaceutical Manufacturing
• Data Science in Biotechnology

Background:

• Historical manufacturing data offers valuable insights for optimizing biopharmaceutical processes.
• Multivariate Data Analysis (MVDA) is a key technique for extracting information from complex datasets.

Purpose of the Study:

• To apply MVDA to historical inactivated polio vaccine (IPV) production data.
• To identify sources of operational variation and assess process consistency.
• To leverage data analysis for process improvement and troubleshooting.

Main Methods:

• Utilized multivariate data analysis (MVDA) techniques, including Principal Component Analysis (PCA).
• Analyzed historical manufacturing data from over 50 batches across two production scales (700-L and 1,500-L).
• Performed explorative analysis on single unit operations and input parameters.

Main Results:

• Confirmed consistent manufacturing across analyzed IPV batches.
• Successfully identified known outliers, such as rejected batches, using PCA.
• Pinpointed variation in input materials, like media, as the primary source of operational variability.
• Found that other process parameters were in control and could not be correlated with product quality attributes.

Conclusions:

• MVDA of historical IPV data provides valuable insights into process consistency and variation.
• Digitalizing and analyzing manufacturing data aids in troubleshooting and identifying areas for improvement.
• Understanding operational variability is crucial for enhancing the robustness of vaccine production processes.