Simultaneous prediction of 16 quality attributes during protein A chromatography using machine learning based Raman...

Area of Science:

• Biopharmaceutical Manufacturing
• Process Analytical Technology
• Spectroscopy

Background:

• Process Analytical Technology (PAT) is crucial for advancing biopharmaceutical manufacturing.
• Raman spectroscopy is a promising PAT tool but faces limitations in predicting multiple quality attributes.
• Current applications are limited due to challenges in spectral data interpretation.

Purpose of the Study:

• To overcome limitations in Raman spectroscopy for biopharmaceutical process monitoring.
• To develop a robust method for predicting multiple product quality attributes in-line.
• To enhance process understanding through continuous quality data generation.

Main Methods:

• Applied Butterworth filters for novel Raman spectra preprocessing.
• Integrated machine learning algorithms for calibration model development.
• Utilized laboratory automation for automated data collection and model training.

Main Results:

• Successfully predicted 16 different product quality attributes using Raman spectroscopy.
• Demonstrated enhanced spectral feature extraction through advanced filtering.
• Achieved in-line, continuous monitoring of product quality attributes.

Conclusions:

• The developed Raman-based multiattribute in-line technology is a breakthrough for bioprocess monitoring.
• Implementation will significantly increase process understanding and enable real-time control.
• This approach facilitates new workflows in process development, characterization, validation, and control.