Towards Rapid Calibration of Bioprocess Quantification Models Using Single Compound Raman Spectra: A Comparison of...

Area of Science:

• Biotechnology and Bioprocess Engineering
• Analytical Chemistry
• Spectroscopy

Background:

• In-line Raman spectroscopy provides real-time bioprocess monitoring of key parameters.
• Traditional model calibration requires extensive data, leading to process-specific and sensitive models.
• Simplifying data generation or using computational methods for synthetic spectra can overcome these limitations.

Purpose of the Study:

• To investigate methods for simplifying the development and implementation of Raman spectroscopy quantification models.
• To assess the use of computational approaches, including synthetic and augmented spectral data, for model calibration.
• To minimize experimental effort while maintaining model robustness for bioprocess monitoring.

Main Methods:

• Utilized a small experimental dataset (16 single compound spectra) for initial model calibration using Partial Least Squares (PLS) and Indirect Hard Modeling (IHM).
• Employed isolated spectral features from IHM to generate synthetic Raman spectra for PLS model calibration.
• Augmented spectra from a single batch process with isolated spectral features to improve calibration.

Main Results:

• PLS and IHM models showed comparable performance for glucose, ethanol, and biomass quantification using limited experimental data.
• Calibration with fully synthetic spectral datasets yielded low relative root mean square error of prediction (rRMSEP) for glucose (3.2%) and ethanol (14.5%).
• Augmenting spectra reduced rRMSEP by 18.6% for glucose and 4.3% for ethanol compared to models calibrated solely on process data.

Conclusions:

• Computational methods, including synthetic and augmented spectral data, significantly reduce experimental effort for Raman spectroscopy model development.
• Achieved robust model calibration, even with complete independence from process-specific data.
• Demonstrated a pathway for rapid implementation of Raman spectroscopy in bioprocess monitoring.