Related Experiment Video
Updated: May 16, 2025
12:02
Automated Modular High Throughput Exopolysaccharide Screening Platform Coupled with Highly Sensitive Carbohydrate Fingerprint Analysis
Published on: April 11, 2016
11.4K
Continuous glucose feedback control using Raman spectroscopy and deep learning models for biopharmaceutical processes
Mohammad Rashedi1, Matthew Demers2, Hamid Khodabandehlou1
1Operations Transformation and Digital Strategy, Amgen Inc., Thousand Oaks, California, USA.
Biotechnology Progress
|April 2, 2025
Summary
Continuous glucose control using deep learning and Raman spectroscopy improves bioprocess efficiency and product quality. This advanced method enhances glucose measurement accuracy, reduces byproducts, and optimizes yields in complex cell cultures.
Area of Science:
- Biotechnology
- Bioprocess Engineering
- Computational Biology
Background:
- High-consumption cell culture processes present challenges in maintaining stable glucose levels, impacting product quality and yield.
- Traditional bolus feeding strategies often lead to glucose fluctuations and suboptimal process performance.
- Accurate, real-time monitoring and control are crucial for optimizing complex biomanufacturing operations.
Purpose of the Study:
- To implement and evaluate continuous glucose control (CGC) strategies in high-consumption cell culture.
- To leverage advanced deep learning models and Raman spectroscopy for precise glucose monitoring and control.
- To assess the impact of CGC on critical quality attributes, process yield, and byproduct formation.
Main Methods:
- Utilized Raman spectroscopy coupled with deep learning models (CNNs, VAE JIT learning) for in-situ glucose monitoring.
- Developed and implemented a continuous glucose calculator (CGC) as a scalable alternative to Raman spectroscopy.
- Compared continuous glucose control strategies against traditional bolus feeding in bioreactor systems across multiple cell lines.
Main Results:
- Deep learning-derived process monitoring significantly enhanced glucose measurement accuracy and stability.
- Continuous glucose control strategies maintained set-point stability, reduced high mannose (HM) levels, and improved overall titer productivity.
- Both Raman-based and CGC-driven strategies minimized glucose fluctuations, reduced undesirable byproducts, and optimized process yields across different cell lines.
Conclusions:
- Continuous glucose control, powered by deep learning and advanced monitoring techniques, offers a robust solution for dynamic, high-consumption bioprocesses.
- The developed CGC provides a scalable and effective alternative for manufacturing environments, improving bioprocess efficiency and product quality.
- This approach enables systematic evaluation of critical quality attributes and addresses the challenges of glucose variability in cell culture.
