Implementation of machine learning tool for continued process verification of process chromatography unit operation.

Area of Science:

• Biopharmaceutical Manufacturing
• Process Analytical Technology (PAT)
• Industry 4.0

Background:

• The biopharmaceutical industry is integrating Industry 4.0 technologies, driven by AI and ML.
• US FDA guidelines mandate a three-stage process validation: design, qualification, and continuous process verification (CPV).
• Implementing CPV is crucial for ensuring consistent biopharmaceutical production quality.

Purpose of the Study:

• To propose and demonstrate a strategy for achieving CPV in cation exchange chromatography (CEX).
• To leverage real-time data analytics for process monitoring and control.
• To facilitate the biopharmaceutical industry's transition to Industry 4.0.

Main Methods:

• Utilized statistical process control (SPC) charts with real-time data from in-built sensors (pH, conductivity, UV, pressure).
• Integrated Raman spectroscopy and Near-Infrared (NIR) spectroscopy for critical quality attributes (CQAs) like charge variant composition and concentration.
• Developed a Python-based program for real-time data analysis and deviation response, including Deep Neural Network (DNN) models for PAT tools.

Main Results:

• Developed PAT models for NIR and Raman spectroscopy achieved R² values > 0.90, indicating statistical significance.
• Demonstrated a control strategy using Raman spectroscopy for CEX eluate charge variant composition.
• Intentional perturbations in CEX load resulted in consistent charge variant composition (acidic ~20%, main ~62%, basic ~18%).

Conclusions:

• Advanced data analytics, soft sensors, and appropriate analyzers enable effective CPV.
• The proposed strategy supports the biopharmaceutical industry in implementing Industry 4.0.
• Real-time monitoring and control of CEX operations ensure consistent product quality and process robustness.