Sign in Subscribe
Machine learning

Deep Learning Method Optimizes Monoclonal Antibody Production

Strategic Insights

— 1 min read

Researchers at the University of California have developed a deep learning method for optimizing monoclonal antibody production processes, resulting in a 28.1% increase in antibody titer and a 27.9% improvement in volumetric productivity. The method uses a hybrid Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) architecture and was validated using industry data from 50 products over 18 months.

Key Takeaways:

  • The deep learning method was developed and validated using industry data from 50 products over 18 months.
  • The proposed design outperforms statistical models, machine learning algorithms, and other deep learning models, achieving a root mean squared error of 0.412 g/L and R^ 2 value of 0.947 for mAb titer prediction.
  • Feature importance analysis identified temperature, dissolved oxygen, and pH as the most critical parameters affecting mAb production.
  • In silico optimization, experiments demonstrated a 28.1% increase in mAb titer and a 27.9% improvement in volumetric productivity.
  • The model's robustness and generalizability were validated across cell lines and bioreactor scales (50L to 2000L).
  • A novel Dynamic Trajectory Similarity (DTS) score was introduced to quantify the model's ability to capture process dynamics, yielding a score of 0.923.
  • The research also discussed limitations, including interpretability challenges and the need for uncertainty quantification in future work.
  • The study was published in the Journal of Advanced Computing Systems in 2024.

Statistics:

  • 50 products: The number of industry data used to validate the proposed design.
  • 18 months: The duration of the validation period.
  • 0.412 g/L: The root mean squared error of the model for mAb titer prediction.
  • 0.947: The R^ 2 value of the model for mAb titer prediction.
  • 28.1%: The increase in mAb titer achieved through in silico optimization.
  • 27.9%: The improvement in volumetric productivity achieved through in silico optimization.
  • 0.923: The Dynamic Trajectory Similarity (DTS) score of the model.

Sources:

  • A Deep Learning Approach for Optimizing Monoclonal Antibody Production Process Parameters. Journal of Advanced Computing Systems, 2024,4(12). Published by Scientific Publication Center.

https://doi-org.sdpl.idm.oclc.org/10.69987/JACS.2024.41203

  • Wenxuan Zheng, Applied Math, University of California, Los Angeles, CA, United States.
  • Mingxuan Yang, Decheng Huang, Meizhizi Jin.

Read more