Model-Enabled Knowledge Transfer Across Cell Lines, Culture Scales and Conditions.

Area of Science:

• Biotechnology
• Bioprocess Engineering
• Computational Biology

Background:

• Mechanistic models are crucial for understanding and optimizing Chinese hamster ovary (CHO) cell culture processes.
• Current models are often limited by parameter sets specific to particular cell lines, scales, or operating conditions, hindering broad applicability.

Purpose of the Study:

• To introduce and evaluate an ensemble-based framework using the ensemble Kalman filter (EnKF) for dual state and parameter estimation in bioprocessing.
• To enable mechanistic model adaptation across distinct CHO cell culture systems without reparameterization.

Main Methods:

• Application of the ensemble Kalman filter (EnKF) for recursively assimilating process measurements.
• Updating uncertain kinetic parameters and predicting system states using an ensemble-based approach.
• Evaluating the framework across six diverse CHO cell datasets (different scales, cell lines, temperatures, feeding strategies).

Main Results:

• The EnKF framework successfully adapted a single calibrated model to new systems using only one experimental dataset.
• Accurate reconstruction of system dynamics was achieved across all evaluated datasets.
• Progressive improvement in long-term predictions was observed as more data was assimilated.

Conclusions:

• The EnKF provides a practical method for knowledge transfer across different bioprocess systems, maintaining mechanistic transparency.
• This approach strengthens the role of mechanistic modeling in data-informed bioprocess understanding and control.
• The evolving parameter ensembles offer valuable time-resolved sensitivity analysis for identifying influential parameters under new conditions.